Processing and Properties of Advanced Ceramics and Composite Ceramic Transactions

Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites

Ceramic Transactions, Volume 203 A Collection of Papers Presented at the 2008 Materials Science and Technology Conference (MS&T08) October 5-9, 2008 Pittsburgh, Pennsylvania

Edited by

Narottam P. Ban sal J. P. Singh

®WILEY A John Wiley & Sons, Inc., Publication

Copyright © 2009 by The American Ceramic Society. All rights reserved. Published by John Wiley & Sons, Inc., Hoboken, New Jersey. Published simultaneously in Canada. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic format. For information about Wiley products, visit our web site at www.wiley.com.

Library of Congress Cataloging-in-Publication Data is available. ISBN 978-0-470-40845-2 Printed in the United States of America. 10 9 8 7 6 5 4 3 2 1

Contents

Preface

ix

MICROWAVE PROCESSING Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria Powder

3

M. A. Imam, A. W. Fliflet, K. L. Siebach, A. David, R. W. Bruce, S. B. Qadri, C. R. Feng and S. H. Gold

Microwave Irradiation-Assisted Method for the Rapid Synthesis of Fine Particles of α-ΑΙ 2 0 3 and a-(AI1.xCrx)203 and Their Coatings onSi(100)

15

Anshita Gairola, A. M. Umarji, and S. A. Shivashankar

CHEMICAL VAPOR DEPOSITION Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor System Via CVD

25

J. C. Flores-Garcia, A. L. Leal-Cruz, and M. I. Pech-Canul

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

35

A. L. Leal-Cruz, M. I. Pech-Canul, E. Lara-Curzio, R. M. Trejo, and R. Peascoe

COMBUSTION SYNTHESIS MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

4

Podbolotov Kirill Borisovich and Diatlova Evgenija Mihajlovna

v

Finite Element Analysis of Self-Propagating High-Temperature Synthesis of Strontium-Doped Lanthanum Manganate

53

Sidney Lin and Jiri Selig

REACTION FORMING AND POLYMER PROCESSING Comparison of Bulk and Nanoscale Properties of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

65

Arif Rahman, Suraj C. Zunjarra, and R. P. Singh

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

77

D. Agaogullari and I. Duman

SINTERING AND HOT PRESSING Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

93

S. F. Li, H. Izui, M. Okano, W. H. Zhang, and T. Watanabe

Comparison of Slip Cast to Hot Pressed Boron Carbide

10

T. Sano, E.S.C. Chin, B. Paliwal, and M. W. Chen

AMORPHOUS CERAMICS Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis of Ultra-High Temperature Ceramics

119

H. Kimura

Preparation and Characterization of Fused Silica Based Ceramic Cores Used in Superalloy Casting

131

M. Arin, S. Sevik, and A. B. Kayihan

COATINGS AND FILMS Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

143

S. Ramanathan, M. Tsuchiya, C. L. Chang, and C. Ko

Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings for Use in Gas Turbine Engines

153

Joseph Kell and Heather McCrabb

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology J. Wang and J. Villafuerte

vi

· Processing and Properties of Advanced Ceramics and Composites

161

COMPOSITES Foreign Object Damage Versus Static Indentation Damage in an Oxide/Oxide Ceramic Matrix Composite

171

Sung R. Choi, Donald J. Alexander, and David C. Faucett

Distinguished Functions Making the Best Use of the Unique Composite Structures

181

Toshihiro Ishikawa

Effects of Environment on Creep Behavior of NEXTEL™720/ Alumina-Mullite Ceramic Composite at 1200 °C

193

C. L Genelin and M. B. Ruggles-Wrenn

Performance of Composite Materials in Corrosive Conditions: Evaluation of Adhesion Loss in Polymers Via Cathodic Disbondment and a Newly Developed NDE Technique

205

Davion Hill, Colin Scott, Ayca Ertekin, and Narasi Sridhar

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

215

C. August, M. Jitianu, and R. Haber

MODELING Modeling of the Pressure in 1-D Green Ceramic Bodies during Depressurization from Conditions of Supercritical Extraction of Binder

229

Kumar Krishnamurthy and Stephen J. Lombardo

Models of the Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

239

Stephen J. Lombardo and Rajiv Sachanandani

Finite Element Modeling of Steel Wire Drawing through Dies Based on Encapsulated Hard Particles

249

Daniel J. Cunningham, Erik M. Byrne, Ivi Smid, John M. Keane

Author Index

255

Processing and Properties of Advanced Ceramics and Composites

· vii

Preface

Two international symposia "Innovative Processing and Synthesis of Ceramics, Glasses and Composites" and "Ceramic Matrix Composites" were held during Materials Science & Technology 2008 Conference & Exhibition (MS&TO8), Pittsburgh, PA, October 5-9, 2008. These symposia provided an international forum for scientists, engineers, and technologists to discuss and exchange state-of-the-art ideas, information, and technology on advanced methods and approaches for processing, synthesis and characterization of ceramics, glasses, and composites. A total of 105 papers, including 12 invited talks, were presented in the form of oral and poster presentations. Authors from 15 countries (Belarus, Canada, China, France, Germany, India, Iran, Japan, Mexico, Norway, Russia, South Korea, Taiwan, Turkey, and the United States) participated. The speakers represented universities, industries, and government research laboratories. These proceedings contain contributions on various aspects of synthesis, processing and properties of ceramics, glasses, and composites that were discussed at the symposium. Twenty two papers describing the latest developments in the areas of combustion synthesis, microwave processing, reaction forming, polymer processing, chemical vapor deposition, electrophoresis, spark plasma sintering, mechanical amorphization, thin films, composites, etc. are included in this volume. Each manuscript was peer-reviewed using The American Ceramic Society review process. The editors wish to extend their gratitude and appreciation to all the authors for their cooperation and contributions, to all the participants and session chairs for their time and efforts, and to all the reviewers for their useful comments and suggestions. Financial support from The American Ceramic Society is gratefully acknowledged. Thanks are due to the staff of the meetings and publications departments of The American Ceramic Society for their invaluable assistance. It is our earnest hope that this volume will serve as a valuable reference for the researchers as well as the technologists interested in innovative approaches for synthesis and processing of ceramics and composites as well as their properties. NAROTTAM P. BANSAL

NASA Glenn Research Center J. P. SINGH

U.S. Army International Technology, Center-Pacific (ITC-PAC) ix

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Microwave Processing

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

CONTINUOUS MICROWAVE-DRIVEN POLYOL PROCESS FOR SYNTHESIZING YTTERBIUM-DOPED YTTRIA POWDER M.A. Imam, A.W. Fliflet, K.L. Siebach*, A. David*, R.W. Bruce**, S. B. Qadri, C. R. Feng and S.H. Gold Materials Science and Component Technology Directorate, Naval Research Laboratory, Washington, DC, USA ABSTRACT The continuous microwave polyol process is a promising novel approach to the synthesis of metallic and ceramic nanopowders. Current efforts are directed toward synthesizing ytterbia-doped yttria (Yb203:Y,03) for use as a polycrystalline laser host material. The process involves pumping a mixture of yttrium nitrate and ytterbium nitrate dissolved in hydrated diethylene glycol through a pressurized quartz tube contained in an S-Band waveguide driven by a 2.45 GHz microwave source at powers up to 6 kW. As the solution moves along the waveguide, it absorbs the co-propagating microwave energy and is heated rapidly to a temperature above 200°C causing a reaction to occur. Condensation reactions then form particles with ytterbium-doped yttria crystal structure. The rapid heating and cooling serve to limit the growth of the crystals so that they are on submicron and fairly uniform in size. The production of doped yttria was confirmed by x-ray diffraction. INTRODUCTION In the polyol process, an organic solvent such as glycol or alcohol is used to reduce a dissolved metal salt to the metal [1,2J. This is commonly done in a boiling, reflux system where the glycol solution of the metal salt is heated to boiling, and the evaporating solvent is condensed and fed back into the solution. At the elevated boiling temperature, the glycol solvent acts as a reducing agent, converting the dissolved metal salt first to a metal oxide and then to the metal. The process results first in the formation of metal atoms suspended in the glycol solvent. These then aggregate, first into clusters and then into larger metallic particles. The process is capable of producing metallic particles in the nanometer size range (1-100 nm), and the particles produced are protected from oxidation or nitridation by the organic solvent and can also be further protected by organic coatings generated during the process from additives. This process has been used for about a decade in production of nanophase powders of metals and mixtures of such metals and films or coatings of these [3], and a wide range of metals can be produced in this manner. The process can also be used to produce metal oxides, sulfides and selenides [4,5]. The limiting factor is the chemical energy available from the solvent vs. the enthalpy of formation of the metal oxides. This makes it very difficult to obtain nanophase metals such as lithium, aluminum, yttrium, magnesium, zirconium, e.g., Groups I-IV, without resorting to much higher processing temperatures, although nanophase oxides of many of these metals can be produced [6]. However, most of the balance of the metals in the periodic table can be produced by this process, e.g., Fe, Co, Ni, Cu, Ru, Rh, Pt, Au, etc., as well as intimate mixtures of these-Fe/Co, Co/Pt, Fe/Pt, Ni/Ag, Cu/Ni, Co/Ni, Co/Ni/Cu, etc. The conventional polyol process, where the processing is done with a high boiling point solvent such as ethylene glycol, heated in a reflux system by a heating mantle, is adequate for production of small quantities of experimental powders, e.g., 1-10 g of product from a 1 liter batch that may take 0.5-2 hours to process. However, the process is intrinsically limited in scalability. In the heating mantle/flask system, scaling to larger volumes results in much greater product variability from varying thermal histories in the larger volume with different convection

3

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

cell structure, and also typically results in larger particle sizes from longer processing times. The typical approach for increasing production rates is to set up a large number of reactors running simultaneously. This approach raises issues of batch-to-batch and reactor-to-reactor product variability, and may never be capable of producing larger quantities for production uses at reasonable costs. One process variation that has been adopted to overcome the limitation on process temperature resulting from boiling of the solvent is to drive the process in sealed containers heated in a microwave cavity [5-10]. This has been done using adaptations of systems intended for microwave digestion. This permits raising the process temperature to as high as 240°C vs. a boiling point for ethylene glycol of ca. 185°C, and avoids the complications associated with solvent boiling. While this process has some advantages over the conventional flask/heating mantle systems, it also has extreme limitations in terms of process size. The typical reactors, Teflon-lined PEI, that, in theory, are limited to temperatures of 170°C, are only 100 ml in volume. This permits production of only about 0.01 g of product per reaction vessel, typically processed in about 15 min. Again, this may be useful for preliminary research purposes, but is clearly not scalable. Going to larger reactors in a more powerful microwave system would result in unacceptably high stresses in the vessels (the stresses scale with reactor diameter), that are already being operated above their rated continuous operating temperature. We have been investigating the use of microwave and millimeter-wave systems in various types of materials processing for about a decade [1,11-17]. As part of this program we have been exploring the use of microwave and millimeter-wave heating of polyol solutions for the production of nanophase metals and oxides. EXPERIMENTAL Initially, this process was demonstrated with millimeter-wave beam heating (83 GHz) of batch polyol systems [14]. This demonstrated that greatly reduced process times could be achieved with the millimeter-wave beam heating, which is capable of rapidly heating polyol solutions in bulk, but still left the problems with the economics of the batch process. Subsequently, continuous polyol processing with millimeter-wave beam heating was demonstrated with very interesting results. In this case, shown in Figure 1 an ethylene glycol/copper acetate solution was heated by the millimeter-wave beam to approx. 200°C in a few seconds as it passed through a silica reaction tube, approx. 10 mm in ID, producing nanophase copper metal particles from the copper acetate. This is an extreme case of microwave polyol processing, as the beam configuration permits very high deposited power densities in the solution, perhaps as high as 100-500 W/cm3 in the liquid, but it's very interesting here that this process can be driven to completion in a few seconds in this continuous system, rather than the hours required in the conventional process. Use of the millimeter-wave system for large-scale production of nanophase metals was considered, but several practical considerations argued otherwise. The millimeter-wave system is very costly, high operating costs, and the continuous system involved extensive plumbing and plumbing connections within the millimeter-wave processing chamber, making operations and maintenance rather difficult and system reliability questionable. A continuous processing system is only economical when it can be operated continuously for long periods at low cost. These considerations led to the decision to use a dedicated, lower frequency, S-band source (2.45 GHz)

4

· Processing and Properties of Advanced Ceramics and Composites

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

Figure 1. Schematic of millimeter-wave driven continuous polyol process using 83 GHz gyrotron source operating at 2-3 kW and producing nanophase copper with less than 10 s reaction time. for a continuous microwave polyol system much more likely to be successful and economically feasible in large scale production of nanophase metals. One pleasing but fortuitous result with this change was much better coupling to the polyol solution. It happens that the peak in absorption in ethylene glycol is very close to 2.45 GHz; thus we have very efficient coupling at this frequency into small volumes of solution. The small volume consideration has significant implications for operation under pressure. We initially experimented with an S-band analogue of the millimeter-wave system, using a resonant cavity made from a shorted waveguide section, tuned to place the maximum in the field at the location of a silica reaction tube passing transversely through the waveguide section. This was found to be impractical because of the field distribution within the cavity that resulted in a heated region only 1-2 cm in length, with inadequate time for the process to occur. We subsequently went to a traveling wave applicator that has been very successful for this process and provides a much better geometry for a well-controlled continuous polyol process [18]. This configuration is shown schematically in Figure 2. In this system, the direction of propagation of the microwaves down the waveguide and the direction of flow of the polyol solution through the inner quartz tube can be in the same or opposite sense. These produce different temperature distributions in the polyol solution. Temperature distributions, based on modeling results, can also be controlled by variations in input microwave power vs. polyol flow rate. Presently, we are operating with the propagation and flow directions the same and with process parameters that result in a fairly uniform temperature distribution. The other choice, with the directions opposite, produces a more gradual temperature rise in the solution with a shorter length of fairly uniform temperature. For the case we are using, the liquid traveling down the tube is heated by absorption of the microwaves traveling down the waveguide, with the power deposited proportional to the electric

Processing and Properties of Advanced Ceramics and Composites

· 5

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

Figure 2. Schematic of microwave polyol process with solution pumped through silica tube placed along centerline of S-Band waveguide-the microwaves that propagate down the waveguide heat the solution flowing along the tube. field squared and the dielectric loss in the liquid. The absorption in the liquid causes attenuation of the microwaves; in our present system and with our current operating parameters, all of the microwave power is absorbed in the polyol solution in a distance of 20-40 cm, as the solution is heated from its initial temperature to a maximum of 180-240°C (depending on system pressure) as shown in Figure 3. Because of the relatively simple geometry and well-defined boundary In this system, the direction of propagation of the microwaves down the waveguide and the direction of flow of the polyol solution through the inner quartz tube can be in the same or opposite sense. These produce different temperature distributions in the polyol solution. Temperature distributions, based on modeling results, can also be controlled by variations in input microwave power vs. polyol flow rate. Presently, we are operating with the propagation and flow directions the same and with process parameters that result in a fairly uniform temperature distribution. The other choice, with the directions opposite, produces a more gradual temperature rise in the solution with a shorter length of fairly uniform temperature. For the case we are using, the liquid traveling down the tube is heated by absorption of the microwaves traveling down the waveguide, with the power deposited proportional to the electric field squared and the dielectric loss in the liquid. The absorption in the liquid causes attenuation of the microwaves; in our present system and with our current operating parameters, all of the microwave power is absorbed in the polyol solution in a distance of 20-40 cm, as the solution is heated from its initial temperature to a maximum of 180-240°C (depending on system pressure) as shown in Figure 3. Because of the relatively simple geometry and well-defined boundary conditions, it is possible to calculate accurately the electric fields in this system; this calculation can include the temperature dependence of the dielectric properties of the polyol solution. From the dielectric loss of the ethylene glycol/metal salt solutions, the energy deposited in the ethylene glycol and heating can be determined, as shown in Figure 4. This heating causes a rapid increase in the temperature of the glycol solution that is offset somewhat by the attenuation of the electric field in the propagating microwaves. Calculation of temperature increases in the ethylene glycol currently include corrections for thermal losses from the silica tube to the waveguide (free

6

· Processing and Properties of Advanced Ceramics and Composites

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

Figure 3. Schematic of basis for model for waveguide microwave heating of flowing polyol solution

Figure 4. Schematic of electric field distribution in waveguide containing silica tube with flowing polyol solution; field in polyol solution is reduced because of high permittivity (dielectric constant), but energy absorption is highly localized in polyol solution because of very high dielectric loss. convection in a closed system) and from the outside of the waveguide (free convection), though these are not significant, corrections with present conditions. Other thermal losses (radiation, conduction) are negligible. With present parameters, the outside of the waveguide, cooled by

Processing and Properties of Advanced Ceramics and Composites

· 7

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

free convection, remains below 60°C, with the polyol solution at 240°C. The result of these calculations, as confirmed by experiments with thermocouple probes at various locations in the polyol solution is a rapid increase, over about 10 cm, in temperature of the polyol solution to a plateau value as shown in Figure 5. This plateau value is maintained, for appropriate choices of microwave power

Figure 5. Schematic of temperature distribution in continuous microwave polyol system with polyol flow and microwave propagation directions coincident with experimental results shown for comparison. and solution flow rate, over a substantial distance-30-60 cm. With this present arrangement, the solution being processed can be maintained at a desired process temperature over this distance which corresponds to a residence time, in the reaction tube of 30-60 s. With this system at ambient pressure, temperatures slightly in excess of boiling can be reached, approx. 180-200°C, for ethylene glycol solutions. We currently are operating the system at an overpressure of 0.20.3 MPa, using pressure regulators on the outlet side and positive displacement pumps to move the solution, and thus can operate at about 240°C, without boiling occurring. The use of this overpressure is greatly facilitated by the relatively small diameter of our silica processing tube that should be capable of operating at pressures as high as 70 MPa. An overall picture of the actual system is shown in Figure 6. This is a somewhat earlier version of the system but shows most of the current features. The S-Band source is to the left is a self-contained unit, -1.5 m in height and 1 m in width, and requires 440V, 30A three-phase electrical power as well as cooling water. The power is taken out of the unit on the top through conventional aluminum and brass waveguide. The working portion of the waveguide, containing the silica reaction tube for the polyol, is placed vertically here. Any power not absorbed in the polyol process is collected in a water load at the upper right of the waveguide. Input power to the system is monitored as well as power reflected back into the source. If needed, power into the water load can be monitored as well either by RF power meters or by measurement of cooling water temperature load. During normal operations, temperature of the polyol solution is monitored at the outlet (at the top). A well either by RF power meters or by measurement of cooling water temperature load. During

8

· Processing and Properties of Advanced Ceramics and Composites

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

Figure 6. Overall view of continuous microwave polyol system based on 6 kW maximum output S-Band, 2.45 GHz source. normal operations, temperature of the polyol solution is monitored at the outlet (at the top). A fine wire thermocouple probe can be used, if necessary to check the temperature at various points along the length of the reaction tube, but is normally not needed with stable continuous operation. Viewports are also available which can be used to observe the process visually at roughly the midpoint of the reaction tube. The system is currently being operated under overpressure; this is achieved via a positive displacement pump driving the polyol solution, currently capable of achieving about 0.7 MPa pressure. The pressure is controlled via a pressure regulator on the outlet. The product solution is rapidly cooled via a stainless steel heat exchanger in an ice water bath and collected for use or analysis. Nearly all of the parts of the present system through which the reactants and products pass are either stainless steel or silica, with thick wall silicone rubber tubing used in some areas. These components will tolerate the reactants used, the temperatures involved (currently up to 240°C), and the overpressures used to suppress boiling. The entire system will tolerate cleaning operations such as flushing with nitric acid to remove metallic residues, followed by distilled water, alcohol and ethylene glycol flushes. This capability is critical if the system is to be used for more than one type of material and purity of products is critical. In the present work, the chemicals used were diethylene glycol (Alfa Aesar), yttrium nitrate (99.999%, Metall), ytterbium nitrate (99.99%, Metall), urea (Fisher Chemical), and

Processing and Properties of Advanced Ceramics and Composites

· 9

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

distilled water. Diethylene glycol was used instead of ethylene glycol to minimize the onset of boiling during processing as it has a higher boiling point. Three thermocouples were used to monitor the process at various points—one at the intake flask, another after the waveguide to measure internal temperature, and another at the collection flask to determine how much the solution has cooled after going through the cooling coil. An S-Band 2.45-GHz Cober high power microwave generator provided the power. The pump system was made up of a peristaltic pump, quartz tubing through an s-band waveguide, a stainless-steel pressure-regulation system, and rubber tubing that led to a stainless-steel cooling coil in a water bath as shown in Figure 6. A precursor solution was created containing diethylene glycol, yttrium nitrate, and ytterbium nitrate. Dissolving agents and catalysts, urea and water, were also added to the solution. Each run contained 1200-mL of diethylene glycol, 30-mL of water, and 24-g of urea. The amount of yttrium nitrate and ytterbium nitrate was determined by weight according to the dopant concentrations of 0%, 5%, and 10%. When mixing the precursor solution, a magnetic stirrer and a hot plate were used to warm the mixture to 70°C, until homogeneous. Before the precursor solution is pumped through the system, the reaction pathway is primed with pure diethylene glycol to eliminate air from the system. Next microwave power is increased slowly until a temperature of ~ 212°C is achieved at the waveguide output at a pressure of 20-psi. This temperature and pressure is maintained as the solution is pumped at 0.37 mL/s through the system and collected in a flask that was maintained at room temperature. The solution is then cleaned in an alcohol rinsing procedure, which requires a centrifuge. The solution is first put in centrifuge tubes and centrifuged for 30-min at 10,000-rpm. The diethylene glycol is then decanted. The tubes are then filled with the alcohol reagent and spun for 15-min at 10,000-rpm two more times. The result was a highly compacted white paste. The paste was then placed on a Petri-dish and warmed to 70°C in open air to evaporate off the alcohol reagent. The powder was calcined in a furnace for 150-min at about 700°C. The resulting powder is analyzed using x-ray diffraction and scanning electron microscopy. X-ray diffraction scans on Y,03 and Yb203 powder samples were obtained using Cu Κα radiation from a rotating anode x-ray source and a high resolution powder diffractometer. Figures 7a shows the scan taken for Y,03 samples after microwave processing. The vertical lines correspond to the expected diffraction pattern for Y203 from JCPDS card (Pdf # 00-043-1036) [19]. It is clear from these figures that a pure Y : 0 3 phase was obtained although in Fig. 7a we also see some extra peaks that are not identifiable. Figures 8a and 9a correspond to a mixture of Y,03 with 5 and 10 % Yb203, respectively. Superimposed on these scans are the red vertical lines corresponding to pure Y : 0 3 phase and blue vertical lines (Pdf # 00-043 1037) [19] corresponding to Yb,03. As can be seen there is a shift between in the position of the peaks with their centroids located in between these two vertical lines. This indicates a change in lattice parameters with increasing Yb,03 composition and also the ytterbium is occupying the yttrium site substitutionally. This is expected if a solid solution is formed between Y203 and Yb,03 since both are isostructural having a space group Ia3(206) [20]. SEM micrographs, Figures 7b, 8b and 9b, show particle size ranging from 400 to 700 nm. We did not attempt to control the particle size by regulating flow rate or temperature. DISCUSSION The steps of reaction for the formation of yttrium oxide can be assessed through the color changes during processing. The solution is initially clear, precipitation occurs while heating and becomes translucent, and then slowly changes to milky-tan. These reaction steps indicate that

10

· Processing and Properties of Advanced Ceramics and Composites

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

Figure 7a. X-ray diffraction scans on Y203 sample obtained by polyol processing

Figure 7b. SEM micrograph of Y20,

Figure 8a. X-ray diffraction scans on Y20, + 5 wt%Yb20, sample obtained by polyol processing

Figure 8b. SEM micrograph of Y,03 +5 wt%Yb203

the precursor first reacts to form an insoluble intermediate, which then slowly becomes yttria by means of a continuous reaction. A likely mechanism for this reaction is given below: W Y(N0 3 ) 3 + ΟΗ->Υ(Ν0 3 ) χ (ΟΗ) 3 _ χ -> Y(OH)3 (2)

Y(OH)3 -> Y(OH)x03_x -> Y 2 0 3

A similar reaction should apply for the formation of ytterbium oxide from ytterbium nitrate. Doping essentially involves introducing an active ion into the crystal lattice. In order to create a good material, there needs to be a homogeneous mixture of the ytterbium oxide within the

Processing and Properties of Advanced Ceramics and Composites

· 11

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

Figure 9a. X-ray diffraction scans on Y,03 + 10 wt%Yb20, sample obtained by polyol processing

Figure 9b. SEM micrograph of Υ,Ο, +10 wt%Yb,03

yttrium oxide crystal structure. By using diethylene glycol in the process, the ytterbium nitrate and yttrium nitrate can be suspended in solution to create a homogeneous solution. The suspension results in a dispersion that helps to create a well dispersed ytterbium-yttrium oxide structure. The resulting powder should have the proper proportions of ytterbium and be evenly distributed since the solution itself was evenly distributed. X-ray diffraction results give the indication that ytterbium was introduced into the yttrium crystal lattice as substitutional element. To date, we have demonstrated that the continuous microwave polyol system can be used in production of sizeable quantities of nanophase ytterbium doped yttrium oxide. The process is energy efficient and can produce material with the purity level determined by the precursor materials. The particle size can be controlled which may help in further processing by optimizing temperature, pump speed, and pressure to fully take advantage of the microwave-assisted polyol process. Note that one significant advantage of operation under overpressure is the ability to use lower boiling point solvents or glycols that typically have higher solubility for the metal salts used as precursors. This could be useful in increasing production rates for materials whose precursors have very limited solubility in high temperature diols such as ethylene glycol. Another approach that may be possible to increase production rates, but has not been explored by the authors, is to employ suspensions of fine particles of precursors in a diol. This would eliminate the restrictions on concentrations imposed by solubilities, but might lead to larger metal particle size in the product because of the very different reaction path. The economics of this process should be far superior to any of the batch polyol processes with much larger production quantities. We would expect that the process here could be readily scaled to higher production quantities. One obvious approach is to move to a higher power SBand microwave source. The higher power, coupled with either a larger reaction tube and/or higher flow rates, would permit production rates several times higher than possible with our prototype system. A limitation with this approach is the high loss in solvents such as ethylene glycol, which limit the penetration depth of microwave energy. For sufficiently large reaction tubes, heating would no longer be relatively uniform across the diameter of the tube and greater process variability would occur. Another, also relatively practical scaling approach, is to move to a lower frequency system, such as L-band (915 MHz) for which low cost, commercial sources and hardware are also available. In this case, with a frequency about 1/3 that of S-band, the

12

· Processing and Properties of Advanced Ceramics and Composites

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

waveguide are approx. 3 times larger, the absorption is somewhat less than at S-Band, and reaction tubes 3-4 times larger than our 10 mm ID could easily be accommodated. An L-Band system, with a 40 mm ID reaction tube, and about 100 kW output, could process hundreds of liters of solution per hour and produce tens of kilograms of nanophase product per shift. There are a number of cautions that need to be cited here. While the economics of the continuous microwave polyol process are quite good, with low capital cost, low operating and labor costs, and low raw materials cost, there are some other critical issues. One is the matter of recovering the nanophase metal or other powder product from the diol solvent and other reaction products. This is done on small scales now, but economical processes (settling, centrifuging, ultrafiltration, controlled agglomeration) need to be available for the separation process on much larger scales. There are also generic problems associated with handling nanophase powders, which tend to agglomerate readily and are highly reactive with atmosphere, but these are not peculiar to polyol-derived nanophase powders. Processing hundreds or thousands of liters of glycol solution per shift also require economical techniques for recycling large quantities of usable solvents and reactants and disposing of waste products, that may or may not be hazardous. Assuming that these problems noted here could be overcome, the continuous microwave polyol process could be a viable technique for economical large-scale production of a wide range of nanophase materials-metals, metal alloys and mixtures, metal oxides and other materials such as selenides and sulfides. CONCLUSIONS 1. The continuous microwave polyol system can be used in production of sizeable quantities of nanophase ytterbium doped yttrium oxide. 2. The economics of this process should be far superior to any of the batch polyol processes. 3. X-ray diffraction results give the indication that ytterbium was introduced into the yttrium crystal lattice as substitutional element. 4. A likely mechanism for the formation of ytterbia/yttria from ytterbia/yttria nitrate is proposed. ACKNOWLEDGEMENTS This work was supported by the U.S. Office of Naval Research. FOOTNOTES * Summer student ** Summer Faculty REFERENCES 'D. Lewis III, M. A. Imam, R. W. Bruce, L. Kurihara, A. W. Fliflet and S. H. Gold, Production of Nanophase Metals via the Continuous Microwave Polyol Process, TMS Proceedings on Powder Materials, pp. 157-168 (2003) 2 D. Larcher and R. Patrice, Preparation of Metallic Powders and Alloys in Polyol Media: A Thermodynamic Approach, J. Solid State Chem. 154,405-411 (2000). 3 G. M. Chow, J. Zhang, Y. Y. Li, J. Ding and W. C. Goh, Electroless polyol synthesis and properties of nanostructured NixCo100x films, Mater. Sci. and Eng. A304-A306, 194-199 (2001). 4 D. Chen, G.Z. Shen, K. B. Tang, X. Jiang, L. Y. Huang, Y. Jin and Y. T. Qian, Polyol mediated synthesis of Nanocrystalline M3SbS3 (M=Ag, Cu), Mater. Res. Bull. 38 (3) 509-513 (2003)

Processing and Properties of Advanced Ceramics and Composites

· 13

Continuous Microwave-Driven Polyol Process for Synthesizing Ytterbium-Doped Yttria

Ή. Grisaru, O. Palchik, A. Gedanken, V. Palchik, M. A. Slifkin and A. M. Weiss, Preparation of the Cd(l-x)Zn(x)Se alloys in the nanophase form using microwave irradiation, J. Mater. Chem. 12 (2) 339-344 (2002) 6 T. Yamamoto, Y. Wada, H. B. Yin, T. Sakata, H. Mori, S. Yanagida, Microwave-driven polyol method for preparation of Ti02 nanocrystallites, Chemistry Letters 10, 964-965, (2002). 7 M. Tsuji, M. Hashimoto and T. Tsuji, Fast Preparation of Nano-sized Nickel Particle Under Microwave Irrad. without Using Catalyst for Nucleation, Chemistry Letters, 1232-1233 (2002). 8 W. Tu and H. Liu, Rapid synthesis of nanoscale colloidal metal clusters by microwave irradiation, J. Mater. Chem. 10, 2207-2211 (2000). 9 S. Komarneni, L. Dongsheng, B. Newalkar, H. Katsuki and A. S. Bhalla, Microwave-Polyol Process for Pt and Ag Nanoparticles, Langmuir 2002,18 5959-5962 (2002). I0 D. Li and S. Komarneni, Microwave-Assisted Polyol Process for Synthesis of Ni Nanoparticles, Journal of the American Ceramic Society. 89, 1510 - 1517(2006). "R. W. Bruce, A. W. Fliflet, D. Lewis III, R. J. Rayne et al., Microwave Sintering of Pure and Doped Nanocrystalline Alumina Compacts, Mat. Res. Soc. Symp. Proc, 430, 139-144 (1996). ,2 R. W. Bruce, A. W. Fliflet, R. P. Fischer, D. Lewis III et al., Millimeter-Wave Processing of Alumina Compacts, Ceramic Transactions Vol. 80 Microwaves: Theory and Application in Materials Processing IV, 287-294 (1997). n A.W. Fliflet, R. W. Bruce, R. P. Fischer, D. Lewis III, L. K. Kurihara, B. A. Bender, G.-M. Chow and R. J. Rayne, A Study of Millimeter-Wave Sintering of Fine-Grained Alumina Compacts, IEEE Trans. On Plasma Sci., Vol. 28 No. 3, 924-935 (2000). U L. K. Kurihara, D. Lewis, A. M. Jung, A. W. Fliflet and R. W. Bruce, Millimeter-Wave Driven Polyol Processing of Nanocrystalline Metals, Proc. MRS, Vol. 634 (2001) 15 S. H. Gold, D. Lewis, A. W. Fliflet, B. Hafizi and J. R. Penano, Interference and Guiding Effects in the Heating of Ceramic Slabs and Joints with Millimeter-Wave Radiation, J. Mater. Synth, and Proc, 9, No. 5 287-297 (2002). 16 D. Lewis, M. A. Imam, L. K. Kurihara, A. W. Fliflet, S. Gold, R. W. Bruce, Material Processing with a High Frequency Millimeter-Wave Source, Mater, and Manuf. Proc, 18, No. 2 151-167(2003). I7 M. A. Imam, D. Lewis III, R. W. Bruce, A. W. Fliflet and L. K. Kurihara, Processing of Advanced Materials with a High Frequency, Millimeter-Wave Beam Source and Other Microwave Systems, Materials Science Forum, 426-432, 4111-4116 (2003). 18 S.H. Gold, R.W. Bruce, A.W. Fliflet, D. Lewis III, L.K. Kurihara, and M.A. Imam, System for Continuous Production of Namophase Material using a Microwave-Driven Polyol Process, Rev. Sci. 78, 02309, pp. 1-6(2007). 19 MacKenzie, K.J.D., Gainsford, G.J., Ryan, M.J, J. Eur. Ceram. Soc, vl6 p553 (1996) 20 Zachariasen, W.H. Skr. Nor. Vidensk.-Akad., Kl. 1: Mat.-Naturvidensk. Kl., v 1928 pi (1928)

14

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

MICROWAVE IRRADIATION-ASSISTED METHOD FOR THE RAPID SYNTHESIS OF FINE PARTICLES OF α-Α1203 AND a-(Al,. x Cr x ) 2 03 AND THEIR COATINGS ON Si(100) Anshita Gairola, A.M. Umarji, and S.A. Shivashankar* Materials Research Centre, Indian Institute of Science, Bangalore, Karnataka -560012, India ABSTRACT Chromium-substituted β-diketonate complexes of aluminium have been synthesized and employed as precursors for a novel "soft chemistry" process, wherein microwave irradiation of a solution of the complex yields, within minutes, well-crystallized needles of a-(Ali_xCrx)203 measuring 20-30 nm in diameter and 50 nm long. By varying the microwave irradiation parameters and using a surfactant such as polyvinyl pyrrolidone, the crystallite size and shape can be controlled and their agglomeration prevented. These microstructural parameters, as well as the polymorph of the Crsubstituted AI2O3 formed, may also be controlled by employing a different complex. Samples of a(Ali.xCrx)203 have been characterized by XRD, FTIR, and TEM. The technique results in material of homogeneous metal composition, as shown by EDAX, and can be adjusted as desired. The technique has been extended to obtain coatings of a-(Ali.xCrx)203 on Si(100). INTRODUCTION Solid state synthesis has been a well-studied area of research for preparing solid solutions of ceramic oxides. The use of two metal salts to obtain a solid solution of oxide is a strategy that has generally been employed [1-7]. The use of a single precursor, which is a metalorganic complex containing two metal ions, would thus be a new approach to the synthesis of substituted (or bimetallic) oxides. Such an approach has been explored in the work presented here, using microwave assisted synthesis, to obtain a-(Ali.xCrx)203. Α12Οι and Cr203 share the same crystal structure, i.e., corundum, where the metal ions occupy two-thirds of the available octahedral sites in the hexagonal close-packed oxide ion arrangement. Aluminium has an ionic radius of 0.53 A, that of chromium being 0.615 A [8], and the two differing by 13.8%. Hence, at normal temperatures, the two phases are expected to be miscible, without segregating, at low concentrations of the substituting ion. These mixed metal oxides are important refractory materials, and find applications as hard coatings and as refractory coatings. Preparation of homogeneous solid solutions by solid-state synthesis generally requires high temperatures and long processing periods. Such a requirement has prompted the development of nonconventional methods, such as sol-gel synthesis, that provide a high degree of compositional homogeneity under moderate processing conditions [4]. The microwave irradiation technique, apart from providing short reaction times and expanded reaction ranges, also provides a "non-conventional" route to deposit thin films on dielectric substrates. These non-conventional techniques can sometimes result in the production of non-thermodynamic or metastable reaction products [9], as opposed to solid state methods that generally give the thermodynamic product. The microwave irradiation technique also provides a "non-conventional" route to deposit thin films on dielectric substrates. This novel approach of preparing nanoparticles and coatings of bimetallic oxides using a "single source" metalorganic precursor is presented here. This rapid and effective method of preparing coatings has not been described previously in the literature. In this work, a mixed-metal acetylacetonate (denoted as 'acac') complex containing aluminium and chromium, Alo.9oCro.io(acac)3, is used to prepare nanoparticles, and thin films, of the mixed aluminium-chromium metal oxide, (AlxCr].x)203. The

15

Microwave Irradiation-Assisted Method for the Rapid Synthesis of Fine Particles on Si(100)

complex Al(acac)3 was used to make nanoparticles of α-Α^Ο^ by microwave irradiation. The use of such acetylacetonate complexes of aluminium and chromium as precursors for the chemical vapour deposition (CVD) of oxides is well known [10]. These complexes facilitate the formation of oxides because of the direct metal-oxygen bonds in the molecular structure. EXPERIMENTAL The chemical precursor, viz., the substituted metal complex Alo.9oCro.io(acac)3, was synthesized using the co-synthesis techniques, wherein the metal salts Al2(S04)3-16H20 and CrCh.oFbO, were dissolved in stoichiometnc amounts of water. The ligand acetylacetone was added to it, and the solution was neutralized using 1:1 NF^FLO, to obtain a precipitate of Alo.9oCro.iu(acac)3. The complex Al(acac)3 was synthesized in a similar manner. The crude was recrystallized using acetone. The formation of complex containing both metal ions was confirmed by FTIR spectroscopy (The spectra were recorded in KBr pellets within a wavelength range of 400-4000 cm'1.) Melting point determination (using the Buchi-B540 melting point apparatus) indicated that the compound has a single, sharp, and congruent melting point. This was confirmed by simultaneous thermogravimetry and differential thermal analysis (TG/DTA). The metal ion composition of the complex was determined using refinement of single crystal X-ray diffraction data obtained on the complex. The microwave irradiation-assisted synthesis of nanoparticles of (AlxCri-x)203 was carried out in a domestic-type microwave operating at 2.45 GHz, and rated at 800 W. The metalorganic complexes Al(acac)3 Al90Cr i0(acac)3 were used as precursors. For this purpose, 0.5 g of the precursor was taken and dissolved in 20 ml of chloroform. To this, 0.05 g of a capping reagent, i.e., polyvinyl pyrrolidine (PVP), and 5 ml of glycerol were added, the latter to aid dissolution. A double-necked round bottom flask fitted with a condenser (and a substrate holder fitted on the other neck) was used for the synthesis. To obtain nanoparticles, the reaction mixture in the round bottom flask was placed at the centre of the microwave oven, and subjected to irradiation at full power (800 W) for 15 minutes. To obtain coatings, a Si(100) substrate measuring ~1 cm' was placed in the substrate holder. Microwave irradiation of the solution results in a cloudy suspension, which is subjected to centrifugation at 5000 rpm for 10 minutes, leading to a precipitate, which was washed several times with acetone. It was then calcined at 500°C for 4 hours to remove the capping agent PVP. In a similar manner, the coated substrate was also calcined to remove the PVP. The crystalline phase of the calcined powder was identified by X-ray diffraction (XRD) using a Siemens model D500 diffractometer with Cu Ku source in the Bragg-Brentano geometry. The average particle size of the particles was estimated using the Scherrer formula. FTIR spectra of the powder samples were obtained using a Digilab FTS60A. These samples were made by pressing KBr pellets at a particle-to-KBr mass ratio of 1:25. The samples were examined by transmission electron microscopy (TEM) in a JEOL CX200 microscope (operated at 120 kV). The TEM specimens were prepared by dispersing the powder in cyclohexane with the aid of ultrasonic agitation. A few drops of this were poured onto a porous carbon film supported on a copper grid, then dried in air. RESULTS AND DISCUSSION The substituted complex was characterized by FT-IR spectroscopy to confirm the formation of the complex. The FT-IR spectrum is shown in figure 1. The spectrum shows the presence of Cr—O and Al—O indicating the presence of both chromium and aluminium in the complex. The peaks at 400

16

· Processing and Properties of Advanced Ceramics and Composites

Microwave Irradiation-Assisted Method for the Rapid Synthesis of Fine Particles on Si(100)

- 500 cm"1 correspond to the metal-oxygen bond, confirming that there is bonding of the metal ion with oxygen of the ligand. The peaks at 460 cm"1, 594 cm"1, 659.00 cm'1 confirm the Cr-0 bond and the peaks at 491 cm"1 and 577 cm"1 confirm the presence of Al-0 bond. In case of the pure Al(acac)3 complex, only the Al-0 bonds were observed.

Fig. 1 FTIR spectrum for the precursor Alo.9oCro.io(acac)3 The various other bands observed were matched with the standard stretching and bending modes for various bonds. The precursor showed a single congruent melting point at 195°C. This indicates that chromium is substituted into the Al(acac)3 lattice and there is no phase segregation. This is important as it ensures the uniformity, homogeneity and constant composition of the precursor. The complex Al(acac)3 showed the lower melting point of 192°C, consistent with the variation of melting points in a solid solution series.. The metal ion ratio (Al:Cr) in the substituted complex was estimated to be 90:10, on the basis of the refinement of the lattice parameters of the crystalline complex, using single crystal X-ray diffraction. Microwave irradiation of the substituted complex yielded a uniformly green precipitate, green being characteristic of a-Cr 2 03. XRD indicated that this powder was crystalline, and that the pattern (Fig. 2) could be indexed to the corundum phase. Microwave irradiation of the aluminium precursor gave a white powder whose XRD pattern could be matched to α-Α1203 ( JCPDS No. 85-1337 )

Processing and Properties of Advanced Ceramics and Composites

· 17

Microwave Irradiation-Assisted Method for the Rapid Synthesis of Fine Particles on Si(100)

Fig.2. XRD pattern of (Α1 χ Ον χ ) 2 0 3 The lattice constants deduced from the data were close to those of a-C^Oi (JCPDS card no.850869). As the metal ion ratio in the precursor complex was Al:Cr = 90:10, and as the powder was green in color, it was surmised that the microwave irradiation of the substituted complex had led to the separation of AI2O3 and Cr 2 03. Thus, the powder XRD pattern was analyzed carefully for the presence of peaks due to a-A^C^. But, no separate peaks were found assignable to (X-AI2O3. These data indicate that, though the metal ion ratio in the precursor complex is Al:Cr=90:10, the metal ion ratio in the crystalline powder that results from microwave irradiation of the complex is "skewed" in favor of Cr. That is, Al substitutes partially for Cr in a-Cr2U3 (as confirmed by the elemental analysis related below). The width of the peaks in the x-ray pattern indicated that the particles were nano-sized. Using the Scherrer equation and the FWHM of the broad peaks, the average particle size was calculated to be 30-35 nm. FT-IR of the nanoparticles recorded in KBr showed the presence of Cr—O and Al—O. The peak at 566 cm"1 corresponds to the Al-0 bond and the one at 626 cm"1 corresponds to the Cr-0 bond. FT-IR spectra of the AI2O3 nanoparticles showed only the Al-0 bonds. TEM micrographs of the powder sample are shown in figure 3. The nanoparticles were observed to be needle like in shape, with a length of 100 nm and a diameter of 20-30 nm.

Fig.3 TEM micrographs of the Al doped Cr2C>3 nanoparticles.

18

· Processing and Properties of Advanced Ceramics and Composites

Microwave Irradiation-Assisted Method for the Rapid Synthesis of Fine Particles on Si(100)

EDS compositional analysis showed a uniform composition of 80% Cr and 20% Al in the powder. Spot EDS taken on various nanoparticle aggregates showed that the metal ion distribution was uniform and homogeneous throughout. These EDS results are consistent with the XRD results, where the pattern matches that of CX-O2O3, but with a slight shift in the 2Θ values of all the peaks. These results are counterintuitive, in that the precursor is aluminium-rich. This implies that, during the microwave processing under the conditions used here, the interaction of the complex with the microwave field "segregates" the "aluminium part" of the precursor from its "chromium part". Since the Al—O bond is smaller than the Cr—O bond, and the Al(acac)3 molecule is smaller than the Cr(acac)3 molecule, Al(acac)3 is expected to be more stable. Thus, during the microwave processing, Cr(acac)3 decomposes to form Cr2U3 while much of the Al(acac)3 remains undecomposed in the solution. This would explain the formation of chromium-rich nanoparticles. A complete understanding of the process leading to the formation of oxide nanoparticles from the metalorganic complex must take into account the interaction of the solvent (chloroform) and the capping agent (PVP) with the microwave field. The presence of water in the solution can also influence the reaction process. A detailed investigation is under way. The coating formed on Si(100) was characterized by XRD. The pattern matches that of aΟ2Ο3, with all peaks shifted by 0.5°. There are no peaks due to a second phase, viz., a-alumina (corundum), just as ion the case of the powder formed under microwave irradiation. Thus, the coating on Si( 100) was deduced to be that of Al-substituted Cr^C^. The XRD pattern is shown in fig.4.

Fig.4 XRD pattern of coating of Al doped Cr 2 0 3 on Si(100) EDS analysis showed that the composition of the coating was homogenous and uniform throughout, and that the metal ion ratio in the coating corresponds to Al:Cr::20:80. This result is consistent with the composition from EDS obtained on the nanoparticles. This is expected as the thin film and the nanoparticles were produced using similar reaction conditions. A SEM micrograph of the coating is shown in Fig.5, displaying its morphology.

Processing and Properties of Advanced Ceramics and Composites

· 19

Microwave Irradiation-Assisted Method for the Rapid Synthesis of Fine Particles on Si(100)

Fig.5 SEM of the film on the silicon substrate It can be seen that coating is continuous, but is not of uniform thickness, and that it is spotted by aggregations. CONCLUSIONS Nanoparticles of bimetallic oxide (Ali.xCrx)203 have been synthesized by using a novel microwave irradiation technique using a single precursor of the type (AlxCri-x) (acac)3. Similarly, nanoparticles of AI2O3 have been prepared using the microwave irradiation technique using the metalorganic precursor Al(acac)3The particles of the substituted oxide have been analyzed by XRD, EDAX, FTIR and TEM. X-ray diffraction confirms the formation of α-(Οο.8Α1ο.2)2θ3, while EDS confirms the metal ion ratio and compositional homogeneity in the crystallites. TEM analysis shows that particle diameter is in the range 20-30 nm and that they are 50 nm long. The microwave irradiation technique has also been extended to obtain continuous, adherent, and compositionally uniform coatings of a-(Cro.8Alo.2)203 on Si(l00). The microwave-assisted technique offers a means for the rapid synthesis of nanoparticles of refractory oxides, and a one-step method for the synthesis of substituted metal oxides. Preliminary results indicate that the technique can be extended to obtain oxide coatings on dielectric substrates. REFERENCES 1.

A. Neuhaus, Physics and Chemistry of High Pressure, ed. Edited by the Society of Chemical Industry. 1963, New York: Gordon and Breach Publishers.

2.

D.M. Roy and R.E. Barks, Nature Phys.Sci, 235 (1972). 118.

3.

W. Sitte, Mater. Sci. Monogr., 28A (1985). 451.

4.

B. Durand, Ceramics Powders. 1983, Amsterdam,The Netherlands: Elsevier Science Publishers. 413-420.

5.

R.M. Spring and S.L. Bender, Journal of the American Cerammic Society, 45(10), (1962). 506506.

20

· Processing and Properties of Advanced Ceramics and Composites

Microwave Irradiation-Assisted Method for the Rapid Synthesis of Fine Particles on Si(100)

6.

L.R. Rossi and W.G. Lawrence, Journal of American Ceramic Society, 53(l 1), (1970). 604608.

7.

W.H. Gitzen, Alumina as a Ceramic Material. 1970, Westerville,OH: The American Chemical Society. Chapter 4.

8.

R.D. Shannon and C.T. Prewitt, Acta Cryst., B25 (1969). 925.

9.

P. Lidstorm, J. Tierney, B. Wathey, and W. Jacob, Tetrahedron, 57 (2001). 9225-9283.

10.

D.F. Bradley and A.R. Barron, Chemical Vapour Deposition, 7(2), (2001). 62-66.

Processing and Properties of Advanced Ceramics and Composites

· 21

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Chemical Vapor Deposition

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

SYNTHESIS AND CHARACTERIZATION OF Si/Si2N20/Si3N4 COMPOSITES FROM SOLID-GAS PRECURSOR SYSTEM VIA CVD J. C. Flores-Garcia, A. L. Leal-Cruz and M. I. Pech-Canul* Centro de Investigacion y de Estudios Avanzados del IPN Unidad Saltillo Carr. Saltillo-Monterrey km 13, Coahuila, Mexico, 25900. ABSTRACT Si/Si2N20/Si3N4 composites were prepared via chemical vapor deposition (CVD) using a solid-gas precursor system. In this method the Si-F gas species generated from the thermal decomposition of a solid precursor react with a nitrogen precursor gas to produce in situ silicon nitride and oxynitride. Cylindrical preforms (3 cm in diameter x 1.25 cm long) with 50 % porosity - prepared by the uniaxial compaction of Si powders with average particle size of 12.4 μιη - were infiltrated in a multiple step mode with high purity nitrogen (HPN) and the Si-F gas species in a chemical vapor deposition (CVD) reactor. The specimens were heated at a rate of 15 °C/min up to a processing temperature of 1300 °C, and maintained isothermally for 70 minutes; then they were cooled down to room temperature. Results from the characterization by XRD and SEM show the deposition of Si2N20 and S13N4 on the Si particles, occupying the interstices of the porous preforms. S13N4 is typically deposited with a sponge-like structure and compact deposits, while Si2N20 is formed with a pin-like morphology and as whiskers. Depending on the number of infiltration steps, Si/Si2N20/Si3N4 composites with tailorable porosity can be prepared. INTRODUCTION The broad range of studies devoted to the development of composite materials has embraced two of the most important advanced ceramics, silicon nitride (S13N4) and oxynitride (Si2N20). This fact is not surprising because both compounds possess excellent properties ideal equally for structural and functional applications. Interestingly, application of composites' paradigm to these materials includes not only dense, but also porous materials. Individually, S13N4 has received much attention because of its high mechanical properties, excellent chemical resistance and high thermal properties. On the other hand, Si2N20 has been recognized by its excellent oxidation resistance in a variety of environments. Accordingly, Si3N4/Si2N20 composites are promising candidates for applications demanding high mechanical strength and oxidation resistance. Moreover, this trend includes incorporation of a third constituent, like in the case of Si3N4-Si2N20-TiN composites. Porous composites of these materials offer the potential to be used as hot gas and molten metal filters, support for catalysts, thermal insulation materials, and preforms for the manufacture of metal/ceramic (using further liquid metal infiltration) or dense ceramic/ceramic composites. Specific suggested applications for porous Si3N4/Si2N20 composites include: filters for the purification systems of air and water and Diesel Paniculate Filters (DPF), targeting the reduction of diesel PM (Paniculate Matter) emissions in the automotive area. In addition to compatibility with the catalyst, DPFs must have excellent thermal shock resistance, thermal and chemical stability, and high surface area. Several processing routes have been reported in the recent literature for these composites [1-11]. For instance porous Si3N4-Si2N20 bodies have been fabricated by the multi-pass extrusion process. This route consists of a nitridation process performed at 1400 °C in flowing N2 gas for 20 hours [1, 2]. From another perspective, silicon nitride and oxynitride composites are

25

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

sought out for superplasticity related applications [3,4]. As it is now well known, silicon nitride exhibits superplastic-like behavior, offering the potential for the fabrication of complex near-netshape components, and reducing the high costs associated to machining. Preparation of superplastic nitrides typically starts from ultrafine P-Si3N4 and involves a lengthy series of operations, including ball-milling in hexane, and hot-pressing. Obviously these operations impact the processing costs; for example, hot pressing is conducted at a pressure of 20 MPa, and at 1750 °C under nitrogen at 1 atmosphere [3,4]. The hybrid precursor chemical vapor deposition method (HYSYCVD) offers the potential to produce silicon nitride and oxynitride composites with advantageous processing characteristics as compared to many previous processing routes. HYSYCVD is a method recently developed at Cinvestav-Saltillo (Cinvestav is the Spanish acronym for Center for Research and Advanced Studies; Saltillo is the location city), for the production of advanced ceramics (for example, alpha- and beta-Si3N4) in which solid-gas reaction systems or hybrid systems are used [5]. This method is based on the ability of some solids like sodium hexafluorosilicate (Na2SiF6) to produce highly reactive gas species (S1F3, S1F2, SiF and Si) which, by the reaction with nitrogen precursors allow the formation of condensed phases as stable solids [6]. As in the conventional CVD route for the processing of S13N4, a number of parameters such as nitrogen precursor, nitrogen precursor flow-rate, substrate condition, and processing time and temperature, significantly influence the occurrence and amount of phases (S12N2O, a- and P-S13N4). HYSYCVD allows reducing processing times and temperatures (70 min per infiltration step, at a processing temperature of 1300 °C), and more importantly, does not require to start from any kind of nitride seed (S13N4 or S12N2O) as is the case in other processing routes which use ultrafine a- or p-Si3N4. Of prime importance -in terms of the paradigm of composites- is the fact that the method allows producing reinforcing phases in an in situ mode, resulting in selfreinforced materials. Depending on the processing parameters, fine powders, whiskers and fibers, which can act as reinforcements, are usually obtained. Another quintessential feature of the method is the feasibility to produce either porous or dense composites by programming successive infiltrations. In this particular investigation, authors report on the microstructure characteristics of composites containing silicon nitride and oxynitride using the HYSYCVD method, obtained by the multiple infiltration approach. EXPERIMENTAL PROCEDURES Processing of composites Silicon (Si powders of 12.4 μιτι average particle size) cylindrical preforms (3 cm in diameter x 1.25 cm high) with 50 % porosity and sodium hexafluorosilicate (Na2SiF6) compacts were prepared by the uniaxial compaction of the corresponding powders in a steel-die. Both, the Si and Na2SiF6 powders were supplied by Sigma Aldrich Inc. Processing of composites was carried out in a hybrid CVD reactor, which consists of a horizontal tube furnace (high-alumina tube, internal diameter =3.175 cm and length = 76.2 cm) with end-cap fittings. The reactor is equipped with gas inlets and outlets to supply the nitrogen precursor (high purity nitrogen HPN, 99.997%, 02(g) (< 5 ppm) or Η 2 0 (?) (< 5 ppm)) as well as with devices to control flow rate, pressure and process atmosphere. Figure 1 is a schematic of the experimental set-up used in the investigation. Sip porous performs were consistently positioned within the tube at the center of the reaction chamber (in the high temperature zone) and the silicon hexafluorosilicate compacts

26

· Processing and Properties of Advanced Ceramics and Composites

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

were placed in a low temperature zone nearby the gas entrance. The thermal gradients along the tube longitudinal axis are used to induce the Na2SiF6 decomposition, for the production of silicon-fluorine gaseous reacting species necessary for the formation of silicon nitride (Si3N4) and oxynitride (Si2N20), and their subsequent deposition. The Sip porous preforms (the subscript p stands for paniculate) were heated in high purity nitrogen (HPN) at a rate of 15 °C/min up to 1300 °C and then maintained at this temperature for 70 min at a constant gage pressure of 12 mbars. Simultaneously, Na2SiF6 was heated (in the temperature range 250-550 °C) and decomposed to generate the Si-F gas species. After completing the test time, the specimens were cooled down to room temperature at a rate of 20 °C/min. In order to maximize deposition, the Sip preforms were processed in four steps by alternating the preform deposition sides from one to another stage. Table I summarizes the conditions under which the trials were carried out. After the infiltration tests, the specimens were removed from the reactor for characterization by XRD, SEM, and EDS.

Fig. 1. Schematic representation of the experimental set-up. rable I. Processing parameters used in the lrihltration tests. 1 Processing parameters Temperature 1300°C - . - . - . Heating rate 15°C/min I Time 70 min Gage pressure 12 mbar Nitrogen precursor High purity nitrogen (HPN) N2 99.997 % Flow rate of HPN 15cnrVmin Porosity of porous preform (Si) 50 volume % of ceramic 1 Quantity of Na2SiF6 (mass) 20 g

Processing and Properties of Advanced Ceramics and Composites

] . | | 1 | 1 |

· 27

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

It should be noted that the processing conditions used in the current work were optimized in previous investigations by the same authors, using only one infiltration step [71. Characterization of raw materials and specimens Initially sodium hexafluorosilicate and silicon powders were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), and thermal analysis (TG/DTA). In order to identify the phases formed in situ, representative samples from each of the specimens were analyzed by XRD. A Philips model 3040 x-ray diffractometer was used under the following conditions: excitation voltage of the anode of 40 kV and current of 30 mA; monochromatic Cu Ka radiation (λ = 1.541838 A); scanning range of 10-80 2Θ degrees, at a scanning speed of 0.02 degrees/sec. In order to determine phase type and morphology, distribution and composition, the specimens were analyzed by SEM and EDS using a Philips XL 30 ESEM scanning electron microscope provided with an EDAX energy dispersive x-ray spectroscopy (EDS) microanalysis device. Both, secondary and backscattering electron modes were used in the analysis at an acceleration voltage between 20 and 30 kV. Prior to the analysis, the specimens were coated with gold. Thermogravimetry (TGVdifferential thermal analysis (DTA) was used only for Na2SiF6. Thermal analysis was performed in a Perkin Elmer model TGA7 thermogravimetric analyzer using a heating rate of 10 °C/min, scanning temperature range from 50 to 1300 °C, and nitrogen atmosphere (HPN) at flow rate of 15 cnrVmin. RESULTS AND DISCUSSION Sodium hexafluorosilicate (Na2SiF6) Figure 2 shows an x-ray diffraction pattern of sodium hexafluorosilicate powders. All reflections correspond to Na2SiF6 according to the Joint Committee on Powder Diffraction Standards (JCPDS) - International Centre for Diffraction Data (ICDD) file No. 72-1115.

Fig. 2. X-ray diffraction pattern of Na2SiF6.

28

· Processing and Properties of Advanced Ceramics and Composites

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

Figure 3a) is a SEM photomicrograph showing sodium hexafluorosilicate with rosettelike geometry, constituted by prismatic particles with sizes in the range of 25 to 40 μπι. Analysis by EDX allows confirming the presence of Si, F and Na element peaks corresponding to Na2SiF6.

Fig. 3. a) SEM photomicrograph and b) EDS spectrum of Na2SiF6 powders. Results from the thermal analysis of Na2SiF6 are shown in Figure 4.

Fig. 4. Thermal analysis of Na2SiF6by TG/DTA. The thermogram in Fig. 4 (curve A) shows a maximum weight loss of approximately 52 % in the temperature range 543-565 °C, which is attributed to the decomposition of Na2SiF6 into silicon-fluorine gaseous species (SiFx where x=0-4) and sodium fluoride (NaF). A second weight loss of approximately 29 % which occurs in the temperature range 965-1200 °C is attributed to the superficial evaporation of molten NaF. In curve B, DTA results show three endothermic events; the first occurs at 559 °C and corresponds to the decomposition of the sodium hexafluorosilicate; the second appears at 972 °C and can be attributed to the transition from NaF(s) to NaF(D (m. p. 993 °C), while the third, which appears at 1024 °C may be associated to the evaporation of molten NaF.

Processing and Properties of Advanced Ceramics and Composites

· 29

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

Specimen phase analysis and microstructure Figure 5 corresponds to a representative X-ray diffraction pattern of specimens before the infiltrations. Before the treatment, reflections pertaining only to the silicon particles are revealed. 4000 η

Si*

3500-

ro

o o I . I .

ro

ai

§

i►

♦

I . I .

g o

Intensity ι counts)

3000-

500-

01

10

1

20

1

1 1

30

♦

.

1

40

1

1

1

1

50

1

1

60

1

1

1

70

1

80

2Θ (degrees)

Fig. 5. X-ray diffraction patterns of specimen before thermal treatment. Si ♦

1/ Γ1000+■· c3 800o o >» 600< / c> *■*

0)

c

400-

▼

10

20

30


40

50

60

—τ— 70

2Θ (degrees)

Fig. 6 a). X-ray diffraction patterns of a representative specimen after 2 infiltrations. JCPDSICCD file numbers for a-Si3N4 and Si2N20 are 76-1409 and 83-2142, respectively. After the processing, the presence of silicon nitride and oxynitride reflections -besides Siis also observed. According to the analysis by XRD of processed specimens, S13N4 and S12N2O

30

· Processing and Properties of Advanced Ceramics and Composites

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

phases were formed on the Sip porous preforms. A representative X-ray diffraction pattern of a specimen infiltrated two times is shown in Figure 6 a). On the other hand, Figure 6 b) shows the XRD pattern of a specimen after four infiltrations. Although no results from phase quantification are shown in this particular work, qualitatively, both the diffractograms in Figures 6 a) and b), and the microstructure analysis by SEM suggest an augment in the deposition of the nitrides. Moreover, according to recent publications [7,8], oxygen content in the nitrogen precursor is sufficient for inducing the formation of silicon oxynitride in gas phase. It should also be pointed out that nitridation of the thin film of silica normally present on silicon plays an important role in the formation of both nitrides, Si2N20 and S13N4. 1400-

Si ♦

1200

'1000-

O

800

=

600-

Έ

400Η

c ω

200 LX^JLLK^^JUWA^J — 1 — 20

30

40

50

— i — 60

—\— 70

2Θ (degrees) Fig. 6 b). X-ray diffraction patterns of a representative specimen after 4 infiltrations. Figure 7 shows SEM photomicrographs at two magnifications (500X and 5000X) of a specimen before processing. It can be observed that the particles exhibit irregular morphology. In Figure 8 SEM photomicrographs of specimens after two infiltrations are shown.

Fig. 7. SEM photomicrographs of Sip porous preform before infiltration.

Processing and Properties of Advanced Ceramics and Composites

· 31

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

Fig. 8. SEM photomicrographs of specimens after 2 infiltration steps. Results from the analysis by SEM show that initially Si3N4 is deposited with a sponge-like morphology and that after subsequent infiltrations, it changes into compact deposits. As for S12N2O, in the first steps it is deposited with a pin-like morphology and with time, it changes into whisker shape. Figure 9 shows typical photomicrographs of specimens after four infiltrations.

Fig. 9. SEM photomicrographs of specimens after 4 infiltration steps. Results from the characterization by XRD are supported by SEM and EDS analysis, as indicated in the spectra shown in Figure 10. It can be noted that after processing, the EDS spectra reveal the presence of silicon (Si), nitrogen (N) and oxygen (O) peaks (Fig. 10 b). By contrast, in the spectra before processing, nitrogen and oxygen peaks are absent (Fig. 10 a). The gold (Au) peak in the spectrum is due to the coatings for analysis by SEM and EDS.

32

· Processing and Properties of Advanced Ceramics and Composites

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

Fig. 10. EDS spectra of specimens a) before and b) after infiltration. A visual analysis at macroscopic level of the specimens shows a change in the color of the porous bodies due to processing. Formation of S13N4 and S12N2O phases caused a color change from dark-gray (Sip porous preform) to light-gray/white (after processing), as shown in Figure 11.

Fig. 11. Color change in the porous bodies from a) dark-gray before infiltration trials to b) light-gray/white after processing. SUMMARY AND CONCLUSIONS In this work the authors report on the preparation of Si/Si3N4/Si2N20 porous composites using a multiple step infiltration approach. To the naked eye it is evident that the nitrides are deposited all over the Sip cylindrical preforms. Even with a four-step processing, the process times involved are still below those used in conventional processing routes. The characterization by XRD and EDS reveal formation of both nitrides under the experimental conditions used in the current work. In addition, according to the analysis by SEM, the S13N4 and S12N2O phases are deposited in a variety of morphologies, ranging from sponge-like, whiskers, and fibers to compact deposits. The reason behind the use of multiple infiltration steps lies not only in the possibility of varying porosity in the composites but also in promoting different reinforcement morphology and size, depending on the desired final application.

Processing and Properties of Advanced Ceramics and Composites

· 33

Synthesis and Characterization of Si/Si2N20/Si3N4 Composites from Solid-Gas Precursor

ACKNOWLEDGEMENTS Authors gratefully acknowledge CONACyT for financial support under contract No. SEP-CONACYT-2005-1/24322. Also, Mr. J. C. Flores-Garcia expresses his gratitude to CONACyT for providing a scholarship. Finally, the authors also thank Mrs. M. Rivas-Aguilar and S. Rodriguez-Arias for technical assistance during the analysis by SEM and XRD, respectively. REFERENCES 1. R. K. Paul, C. W. Lee, H. D. Kim and B.-T. Lee, Microstructure characterization of in situ synthesized porous Si3N4-Si2N20 composites using feldspar additive, J. Mater. Sci., 42,4701-06(2007). 2. B. T. Lee, R. K. Paul, C. W. Lee and H.-D. Kim, Fabrication and microstructure characterization of continuously porous Si3N4-Si2N20 ceramics, Mater. Lett., 61, 2182-86 (2007). 3. R. J. Xie, M. Mitomo, and G. D. Zhan, Superplastic deformation in silicon nitride-silicon oxynitride in situ composites, J. Am. Ceram. Soc, 83 (10), 2529-35 (2000). 4. R. J. Xie, M. Mitomo, F.-F. Xu, G. D. Zhan, Y. Bando and Y. Akimune, Microstructure and mechanical properties of superplastically deformed silicon nitride-silicon oxynitride in situ composites, J. Eur. Ceram. Soc, 22, 963-71 (2002). 5. A. L. Leal-Cruz and M. I. Pech-Canul, In situ synthesis of S13N4 from Na2SiF6 as a silicon solid precursor, J. Mater. Chem Phys., 98, 27-33 (2006). 6. A.L. Leal-Cruz and M.I. Pech-Canul, In situ synthesis of S13N4 in the Na2SiF6 - N2 system via CVD: kinetics and mechanism of solid-precursor decomposition, Solid State Ionics, 111, 3529-36 (2007). 7. A. L. Leal-Cruz, Thermodynamics, kinetics and microstructural study of Na2SiF6 decomposition- silicon nitrides formation in systems Na2SiF6-nitrogen precursor-diluent, Ph. D. Thesis, Cinvestav Saltillo, Saltillo Coah. Mexico, 2007. 8. M. I. Pech-Canul, J. L. de la Pena, and A. L. Leal-Cruz, Effect of processing parameters on the deposition rate of Si3N4/Si2N20 by chemical vapor infiltration and the in situ thermal decomposition of Na2SiF6, Appl. Phys. A., 89, 729-35 (2007). 9. R. G. Duan, G. Roebben, J. Vleugels and O. Van der Biest, In situ formation of S12N2O and TiN in Si3N4-based ceramic composites, Acta Mater., 53, 2547-54 (2005). 10. R. G. Duan, G. Roebben, J. Vleugels and O. Van der Biest, Thermal stability of in situ formed Si3N4-Si2N20-TiN composites, J. Eur. Ceram. Soc, 22, 2527-35 (2002). 11. J. L. de la Pena and M. I. Pech-Canul, Microstructure and kinetics of formation of S12N2O and Si3N4 into Si porous preforms by chemical vapor infiltration, Ceram. Int., 33, 134956 (2007).

34

· Processing and Properties of Advanced Ceramics and Composites

Processing Processing and and Properties Properties of of Advanced Advanced Ceramics Creamics and and Composites Composites Edited P. Bansal Bansal and and J.J. P. P. Singh Singh Edited by by Narottam Narottam P. Copyright Copyright O O 2009 2009 The The American American Ceramic Creamic Society. Society.

EFFECT OF FLOW RATE, NITROGEN PRECURSOR AND DILUENT ON Si2N20 DEPOSITION BY HYSYCVD A. L. Leal-Cruz1, M. I. Pech-Canul1*, E. Lara-Curzio2, R. M. Trejo2, and R. Peascoe2. 'Cinvestav-Saltillo. Carr. Saltillo-Monterrey km. 13, Saltillo, Coah., Mexico, 25900. 2 Oak Ridge National Laboratory. One bethel Valley Road, Bldng. 4515. MS-6069. Oak Ridge TN, USA, 37831. ABSTRACT The effects of flow rate, nitrogen precursor and diluent on the synthesis of S12N2O by the hybrid precursor chemical vapor deposition method (HYSYCVD) were investigated. Based on an L9 Taguchi experimental design, the processing parameters were studied at three levels: flow rate (10, 15 and 20 cm3/min), nitrogen precursor (UHP-N2, 50% N2-balance ammonia, and 5 % N2-balance ammonia), and diluent (Ar, He, and with no use of diluent). Anova shows that flow rate of nitrogen precursor has the highest relative contribution (46 %) to the variability in the formation of Si2N20, followed by the type of nitrogen precursor (44 %) and, the type of diluent (8 %). Deposition of S12N2O is maximized by using 10 cnrVmin of UHP-N2 and with no use of diluent. Results from the characterization of additional trials by XRD and SEM show that in the temperature range 1153.15-1603.15 K and processing times between 0-70 min, S12N2O is formed as spheres, rough and fine fibers. INTRODUCTION Silicon base ceramics such as silicon carbide (SiC), silicon nitride (S13N4) and silicon oxynitride (S12N2O) have been considered promising structural materials for high temperature applications because of their low density, high strength, good chemical inertness and excellent oxidation resistance. S12N2O was discovered in 1926 as a new mineral in samples from the Jajh deh Kot Lalu meteorite fallen in the village with the same name in the Sind province of Pakistan, and it was recognized later on by electron microprobe analysis [1]. Schumb and Lefevre determined its composition with reasonable accuracy in 1954 [2]. Nonetheless, it was not until 1967 when other systematic studies started to appear in the scientific literature [3]. During the last forty years, it has been recognized as a promising material for high temperature applications due to its excellent oxidation and thermal shock resistance in a variety of environments [2-6]. It can keep good oxidation resistance in air up to 1873.15 K and high flexural strength up to 1673.15 K without degradation. In addition, it possesses very low theoretical density (2.81 g/cm3), high hardness (Hv: 17-22 GPa), low thermal expansion coefficient (3.5x10"6 K'1), and high thermodynamic stability temperature (~ 2073.15 K) [5]. Accordingly, S12N2O has been regarded mostly for structural applications. Although from the perspective of ceramics engineering S12N2O has received less attention than S13N4, recent studies suggest an apparent opposite trend. This is explained by the growing interest in this ceramic material for use in microelectronic devices [7-13]. More specifically, a variety of functional applications including interlevel dielectrics, passivation, diffusion barriers and antireflection coatings exemplify some of the potential applications for silicon oxynitride in the microelectronics industry [7]. In this respect, and given the increase in the spectrum of applications for S12N2O, it would be desirable to offer new or alternative processing routes to potential users. Advanced ceramics, such as aand p-Si3N4, as well as S12N2O can be produced under suitable processing conditions by the hybrid precursor chemical vapor deposition method (HYSYCVD) [14-17]. HYSYCVD is a

35

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

method recently developed at Cinvestav-Saltillo (Cinvestav is the Spanish acronym for Center for Research and Advanced Studies; Saltillo is the location city), for the production of advanced ceramics (for example, alpha- and beta-Si3N4) in which solid-gas reaction systems or hybrid systems are used [14]. This method is based on the ability of some solids like sodium hexafluorosilicate (Na2SiF6) to produce highly reactive gas species (S1F3, S1F2, SiF and Si) which, by the reaction with other proper precursors (for instance, nitrogen precursors), allow the formation of condensed phases as stable solids [15, 16]. As in the conventional CVD route for the processing of S13N4, a number of parameters such as nitrogen precursor, nitrogen precursor flow-rate, substrate condition, and processing time and temperature, significantly influence the occurrence and amount of phases (S12N2O, a- and P-Si3N4). Systematic investigations on the quantitative effect of the processing parameters in this field are rather scarce [14, 15, 17]. Studies on the effect of gas diluents upon the amount, phase type and morphology are also very limited [17, 18]. In the case of S12N2O, nothing is documented so far on the quantitative effect of various processing parameters, including the use of diluents. Based on a Taguchi experiment design, this work is aimed at studying the effect of nitrogen precursor, nitrogen precursor flow-rate, substrate porosity and diluent, on the formation of S12N2O in Na2SiF6-nitrogen precursor-diluent systems by HYSYCVD. EXPERIMENTAL PROCEDURES An L9 Taguchi experiment design and analysis of variance (Anova) were used to study the effect of the following processing parameters in three levels, on the deposition of S12N2O in Na2SiF6-nitrogen precursor-diluent systems: nitrogen precursor (N2, 50 vol. % N2-balance NH3, and 5 vol. % N2-balance NH3), diluent (argon, helium, and no diluent), nitrogen precursor flowrate (10, 15, and 20 cnr/min ) and SiCp/Sip substrate porosity (40, 50, and 60 vol. %). Table I shows the L9 Taguchi experiment design considered in the study.

An L9 Taguchi experiment design allows studying the effect of up to 4 parameters in 3 levels each, with a minimum number of trials, and optimizing the processing parameters. Anova provides insight into the optimum process parameters and a means of estimating the percent contribution of each of the parameters tested to the variability in the measured response variable (for this particular case, the amount of Si2N20 formed) [19]. The silicon-solid-precursor, sodium hexafluorosilicate (Na2SiF6) was prepared in our laboratory by means of the dissolution of Si0 2 (obtained from rice hulls) in hydrofluoric acid (HF) and the subsequent precipitation with a

36

· Processing and Properties of Advanced Ceramics and Composites

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

solution of sodium hydroxide (NaOH). The nitrogen precursors (N2, 50 % N2-balance NH3 and, 5% N2-balance NH3), and the diluents He and Ar - all of ultra high purity (99.9999 %) - were supplied by INFRA Air products. According to the suppliers, the gases contain oxygen either as 02(g) (< 5 ppm) or as H20(g) (< 5 ppm). The substrates consisted of SiC/Si porous preforms (40, 50, and 60 vol. %) with 5 wt. % Si each, prepared by the compaction of the corresponding amount of the powders (both with particle size: -325 mesh). The silicon powders were supplied by Sigma Aldrich Inc., and the silicon carbide powders were provided by Microabrasivos de Mexico. Synthesis tests were carried out in a HYSYCVD reactor at constant pressure, slightly above to that of the atmospheric pressure (gage pressure = 17 mbars). The reactor consists of a horizontal furnace with an alumina tube (3.8 cm in diameter x 76.2 cm long) provided with devices to control the processing temperature, pressure, and gas flow rate; the system also includes a powder collector and a neutralizer of the gas by-products. Na2SiF6 compacts (3.175 cm in diameter x 2.00 cm long) and SiCp/Sip cylindrical porous preforms (3.175 cm in diameter x 3.00 cm long) were prepared by the uniaxial compaction of the corresponding powders in a steel die. The 25 g compacts of Na2SiF6 were placed within the tube between the gas entrance and the SiCp/Si porous substrate. The SiCp/Sip preforms were positioned strategically within the tube and heated at a rate of 15 °C/min up to 1573.15 K either in N2 or N9-NH3 in accordance with the experimental plan; then the specimens were maintained isothermally at this temperature for 2 hours under the conditions established in the L9 Taguchi experimental design. Systematically, leak tests were performed in the reaction chamber before starting each deposition trial. Additional test of synthesis of Si2N20 at different times (0-70 minutes) and temperatures (1153.15-1603.15 K) were also carried out. After the synthesis tests, the specimens were removed from the reactor for chemical and microstructural characterization by X-ray diffraction (XRD), scanning electron microscopy (SEM), and energy dispersive X-ray spectroscopy (EDS). The XRD tests were performed using a Philips XRD equipment (Cu Ka radiation, anode excitation of 40 kV, and current of 30 mA). For quantification operations by XRD, specimens were previously prepared by finely grinding (500 mesh) the CVD samples in an agate mortar; slurries were then prepared with these fine powders and acetone, and placed on silicon singlecrystal substrates for analysis at a step size of 0.02 degrees and a rate/count time of 1 degree/min, from 15 to 80 2Θ degrees. Refinement of the XRD patterns by Rietveld's method, identification and quantification of deposited phases were performed by means of the Jade™ computer software. The microstructural characterization was carried out in low vacuum, at an acceleration voltage of 30 kVe, in a Philips SEM equipment provided with a Falcon™ EDS device. Previous to the SEM analysis, the samples were coated with graphite. RESULTS AND DISCUSSION Identification and quantification of the deposited phases by XRD technique Characterization by XRD reveals that different phases such as silicon oxynitride (Si2N20), and a- and p-Si3N4 were deposited under the various conditions tested in the L9 experimental design. Figure 1 shows the XRD qualitative results of deposited phases under the conditions of the L9 Taguchi experimental design.

Processing and Properties of Advanced Ceramics and Composites

· 37

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

Figure 1. XRD patterns corresponding to all tests in the experiment design. (·) SiC, (♦) Si, (o) Si2N20, (a) a-Si3N4, and (β) p-Si3N4. Results from the quantitative analyses are shown in Table II. These results reveal that Si2N20 can be obtained in the composition range of 3.75-100 wt.%, a-Si3N4 of 0-96.25 wt.% and p-Si3N4 of 0-20 wt.% under the conditions of L9 Taguchi experimental design at 1300°C for 2 h. In addition, these results indicate that it is possible to deposit Si2N20 as single phase under the conditions of trial LI (40 v % substrate porosity, nitrogen precursor, 10 cm7min flow rate, and Ar as diluent).

38

· Processing and Properties of Advanced Ceramics and Composites

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

Table II. Quantitative analyses of phases deposited by HYSYCVD under the conditions of the L9 Taguchi experimental design. COMPOSITION wt.% TRIAL Si2N20 P-Si3N4 a-Si3N4 LI 100.00 0.00 0.00 X L4 73.00 11.00 16.00 L7 83.00 1.70 15.30 21.89 0.00 78.11 Z g « L2 L5 19.60 1.80 78.60 o «^ L8 93.40 3.30 3.30 L3 28.00 52.00 20.00 L6 65.80 0.00 34.20 0.00 3.75 96.25 L9 Analysis of variance for the deposition of S12N2O Results from the analysis of variance (Anova) are shown in Table III. The last column in Table III indicates the percent contribution of the tested parameters - flow rate, nitrogen precursor and diluent - to the variability in the amount of deposited S12N2O. Table III. ANOVA table for the amount of Si2N2Q formed by HYSYCVD. DF F Factors S V P(%) 2 SP Pooled 54.00 44 2 NP 941.36 1525.00 1525.00 2 990.74 46 1605.00 1605.00 Q 8 2 179.01 D 290.00 290.00 2 2 Error 27.00 18.00 54.00 4 1158.00 100 Total 3476.00 SP is substrate porosity, NP is nitrogen precursor, Q is flow rate of nitrogen precursor, D is type of diluent, DF is degrees of freedom, S is sum of squares, V is variance, F is variance ratio and P is percentage contribution. According to Anova results, flow rate of nitrogen precursor has the highest relative contribution (46 %) to the variability in the deposition of Si2N20, followed by the type of nitrogen precursor (44 %) and, the type of diluent (8 %). On the other hand, analysis of the main effects shows that the deposition of Si2N20 is optimized by using 10 cnrVmin of UHP-N2 and, with no use of diluent. Characterization results of the deposition of Si2N20 in the temperature range of 1153-1603 K from 0 to 70 minutes XRD quantitative results of the amount of Si2N20 deposited in samples of additional synthesis trials, as a function of time for each temperature are shown in Figure 2.

Processing and Properties of Advanced Ceramics and Composites

· 39

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

Figure 2. Amount of deposited S12N2O as a function of time for various temperatures. The curves of S12N2O amount versus time showed that in the temperature range 1153 to 1603 K and from 0-70 minutes, S12N2O deposition has a parabolic behavior, typical of many oxidation reactions. According to the results, the maximum amount of S12N2O is achieved at 1603 K for 70 min. On the other hand, analysis by SEM reveals that S12N2O is present in different morphologies such as spheres, snow-like deposits, fine fibers and thin fibers. Analysis by EDS reveals the presence of silicon (Si), nitrogen (N) and oxygen (O) corresponding to chemical elements of the S12N2O phase. Carbon in the EDS spectrum can be attributed to SiC from the substrate and to graphite used to coat the samples before the analysis. Figure 3(a) shows an SEM photomicrograph representing the typical microstructure.

7000 6000 ^5000

z>

. 4000

«Tax»

c O 2000 O 1000 0

0

o

N

100

200

300

400

Energy (eV) Figure 3. (a) SEM photomicrograph of S12N2O on the SiC porous substrate and its corresponding EDS spectrum.

40

Figure 3(b) is a typical EDS spectrum.

· Processing and Properties of Advanced Ceramics and Composites

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

CONCLUSIONS Silicon oxynitride (Si2N20) can be deposited via (HYSYCVD) in a composition range of 3.75-100 wt.%, a-Si3N4 in the range 0-96.25 wt.% and p-Si3N4 between 0 and 20 wt.% under the conditions of the L9 Taguchi experimental design at 1300°C for 2 h. Despite the broad composition range for S12N2O deposition, it can be produced selectively, i.e., as a single phase under the conditions of Trial LI (40 v % substrate porosity, nitrogen as precursor, 10 cm /min flow rate, and Ar as diluent). Flow rate of nitrogen precursor has the highest relative contribution (46 %) to the variability in the formation of Si2N20, followed by the type of nitrogen precursor (44 %) and, the type of diluent (8 %). The deposition of S12N2O is optimized by using 10 cm3/min of UHP-N2 and with no use of diluent. After optimization, and in order to evaluate only the effect of time and temperature, additional tests were carried out in the temperature range of 1153-1603 K from 0 to 70 minutes. The results show that Si2N20 deposition has a parabolic behavior and that the maximum amount of S12N2O is achieved at 1603 K for 70 min. The effects of the other parameters (flow rate, type of diluent, etc.) were evaluated previously in the L9 Taguchi experimental design.

ACKNOWLEDGEMENTS Authors gratefully acknowledge CONACyT for financial support under contract No. SEP-CON AC YT-2005-1Z24322. Also, Dr. Leal-Cruz expresses his gratitude to CONACyT for providing a postdoctoral scholarship under the same research project. REFERENCES PI [2] [3] [41 [5] f61 [7] [81

C. A. Andersen, K. Keil, and B. Mason, Silicon Oxynitride: A meteoritic Mineral, Science, 146,256-57(1964). R.W. Cahn, P. Haasen, and E. J. Kramer, Encyclopedia of materials science and technology, Wiley-VCH, Germany, 2004. M. Ohashi, S. Kanzaki, and H. Tabata, Processing, Mechanical Properties, and Oxidation Behavior of Silicon Oxynitride Ceramics, J. Am. Ceram. Soc, 74, 109-13 (1991). Q. Tong, J. Wang, Z. Li, and Y. Zhou, Low-Temperature Synthesis/Densification and Properties of Si2N20 Prepared with Li 2 0 Additive, J. Eur. Ceram. Soc, 27, 4767-72 (2007). M. Heim, J. Chen, and R. Pompe, Dry and Wet Oxidation of an S12N2O - Ζ Γ 0 2 Composite Material, J. Mater. Set., 32, 4025-30 (1997). R. Larke, Reaction Sintering and Properties of Silicon Oxinitride Densified by Hot Isostatic Pressing, J. Am. Ceram. Soc, 75, 62-66 (1992). E. P. Gusev, H. C. Lu, E. L. Garfunkel, T. Gustafsson, and M. L. Green, Growth and Characterization of Ultrathin Nitride Silicon Oxide Films, IBM J. Res. Develop., 43, 265-86(1999). H. G.Tompkins, S. Smith, D. Convey, R. B. Gregory, M. L. Kottke, and D. Collins, Determinin the Amount of Si-Si Bonding in CVD Oxynitrides, Surf. Interface. Anal., 35, 136-40(2003).

Processing and Properties of Advanced Ceramics and Composites

·

41

Effect of Flow Rate, Nitrogen Precursor and Diluent on Si 2 N 2 0 Deposition by HYSYCVD

[9] [10] [11] [12] [13] [14] [15] [16] [17]

[18] [19]

42

O. Sanchez, J. M. Martinez-Duart, R. J. Gomez-Sanroman, M. A. Aguilar, C. Falcony, F. Fernandez-Gutierrez, M. Hernandez-Velez, SiOxNy Films Deposited with SiCU by Remote Plasma Enhanced CVD, J. Mater. Sci., 34, 3007-12 (1999). W. Bohne, W. Fuhs, J. Rohrich, B. Selle, I. Sieber, A. del Prado, E. San Andres, I. Martil and G. Gonzalez-Diaz, Surf. Interface Anal. 34, 749-53 (2002). O. Jintsugawa, M. Sakuraba, T. Matsura, J. Murota, Surf. Interface Anal. 34, 456-59 (2002). D. W. Hess, IBM J. Res. Develop. 43, 127-43 (1999). K. A. Ellis, R. A. Buhrman, I. B. M. J. Res. Develop. 43, 287-99 (1999). A. L. Leal-Cruz, "Synthesis and characterization of silicon nitride reinforcements by the thermal decomposition of Na2SiF6 in nitrogen containing atmospheres", M.Sc. Thesis, Department of Ceramics Engineering, CINVESTAV-Saltillo, Mexico, (2004). A. L. Leal-Cruz and M.I. Pech-Canul, In Situ Synthesis of S13N4 from Na2SiF6 as a Silicon Solid Precursor, Mat. Chem. and Phys., 98, 27-33 (2006). A. L. Leal-Cruz and M.I. Pech-Canul, In Situ Synthesis of S13N4 in the Na2SiF6-N2 system via CVD: Kinetics and Mechanism of Solid-Precursor Decomposition, Solid State Ionics, 177, 3529-3536 (2007). A. L. Leal-Cruz, "Thermodynamic, kinetics and microstructural study of the decomposition of Na2SiF6- formation of silicon nitrides in Na2SiF6-nitrogen precursordiluent system", Ph. D. Thesis, Department of Ceramics Engineering, CINVESTAVSaltillo, Mexico, (2007). A. I. Kingon, L. J. Lutz and R. F. Davis, Thermodynamic Calculations for the CVD of Silicon Nitride, J. Am. Ceram. Soc, 66, 551-58 (1997). R. Roy, "A Primer on the Taguchi Method", Society of Manufacturing Engineers, Dearborn Michigan (1990) 100.

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

MgAl204/SiC COMPOSITE CERAMIC MATERIAL PRODUCED BY COMBUSTION SYNTHESIS Podbolotov Kirill Borisovich, Diatlova Evgenija Mihajlovna Department of Glass and Ceramic Technology, Belarusian State Technological University Minsk, Belarus. ABSTRACT The paper describes the possibility of producing MgAl204/SiC composite ceramic material by combustion of mixture containing magnesium carbonate, silica, aluminum and carbon. The investigation is made of the processes which proceed during combustion of various mixture compositions. Also the characteristics of obtained materials are determined. The microstructure investigations revealed the presence of fibrous silicon carbide crystals. The material obtained can be used for production of ceramic items for various applications. Keywords: Combustion Synthesis, Spinel, Silicon Carbide, Refractory Materials, Composite. INTRODUCTION The combustion synthesis (CS) or self-propagating high-temperature synthesis (SHS) is one of the efficient methods of producing composite materials. The basic principle of CS is selfpropagating of chemical reaction zone in media that are able to release chemical energy with formation of valuable condensed products1. The CS method does not require high-temperature furnacing of items. This enables significant reduction of production expenses and energy consumption and, consequently, cuts down cost of product. In the process of powder mixture combustion the phase formation occurs exactly in reactionary volume and crystalline compounds formed grow in particles of other compounds. In this case it is possible to obtain composites of crystalline phases such as those of spinel and silicon carbide the formation of which is complicated under usual conditions. MgAl204 magnesium-aluminum spinel is a ceramic material possessing rather high mechanical strength and good resistance to corrosion and radiation. It finds widespread application in the form of lining for inductors and resistance furnaces and as melt filters, hightemperature insulators and other items in metallurgy, machine building, instrument making and chemical industry. Spinel ceramics is extensively used in manufacture of protecting cases for thermocouples and metal melting crucibles. This spinel is also successfully used as refractory material in thermal systems by itself and also in production of magnesia refractory materials ' . In view of the kinetic difficulties encountered during synthesis of spinel according to the MgO + AI2O3 = MgAl204 reaction the CS process has to be carried out at high temperatures4,5. ' A.G. Merzhanov, Solid-Flame Combustion, Chernogolovka: ISMAN (2000), 224 (in Russian). V.L. Balkevich, Technical Ceramics, Stroiizdat, Moscow (1984), 265 (in Russian). 3 E.V. Degtiaryova and I.S. Kaynarovsky, Magnesium-Silica and Spinel Refractory Materials, M: Metallurgy (1977), 169 (in Russian). 4 M. O'Driscoll, Spinel. The Steel and Cement Specialist, Industrial Minerals, 10, 40-41, (2001). 5 I.D. Koshcheyev, Synthesis of spinel from caustic magnesite and dust of silica production, New Refractory Materials, 8, 17-21, (2004) (in Russian).

2

45

MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

There is technology known as low-temperature sol-gel method that is also intended to produce spinel by conducting the reaction between aluminum and magnesium nitrates and citric acid . The fact that formation of spinel and compounds including magnesium and aluminum oxides occurs on combustion of aluminum and magnesium alloys is reported in the works7. However, this method of spinel production is unpractical from the economical standpoint since raw material components are scarce and costly. The production of spinel from available raw materials while reducing power consumption is made possible by using the CS method for mixtures prepared from magnesium-containing components and metallic aluminum. As for the above works attention can be focused on production of spinel when using periclase and aluminum with subsequent sintering at 1250°C, the combustion is sustained by means of magnesium sulfate8. In addition to spinel, a group formed by silicon-carbide based materials appears to be particularly promising. Since production of high-strength molded products from a pure silicon carbide is rather complicated; the refractory materials with high carbide content are manufactured using various binders. Silicon-carbide refractory materials and items made on their basis have comparatively high electric and heat conductivity as well as thermal and abrasive resistance. They are not wetted with nonferrous metals, possess high mechanical strength in cold and heated states and are resistant to acid slags. These properties make them highly suitable for production of various composite materials for metallurgy needs and suggest the use of CS method for this purpose. In some works the combustion synthesis of silicon carbide was investigated using mixtures containing silicon and carbon powders. However, because of low exothermicity of reaction the synthesis was carried out under exposure to microwave irradiation or by passing electric current through mixtures9,!(). When CS processes proceed in mixtures containing silicon and carbon oxides and metallic aluminum the formation of silicon-carbon refractory materials is observed to occur in combination with high-melting corundum and mullite phases11*12. The production of spinel - silicon-carbide refractory materials and items made on their basis has great potential for industrial use owing to high performance characteristics of obtained products. Up to now little attention has been given to this research direction and effective

6

S.K. Behera, Synthesis of magnesium-aluminum spinel from autoignition of citrate-nitrate gel, Materials letters, 58 (9), 1451-1455, (2004). 7 P.F. Pokhil, A.F. Beliayev, Yu.V. Frolov et al., Combustion of Powder-Like Metals in Active Media, Nauka, Moscow (1972), 294 (in Russian). 8 A.B. Ivanov and N.O. Ivanova, Production of aluminum-magnesium spinel under the conditions of SHS , Refractory Materials and Technical Ceramics, 12, 10-12, (1994) (in Russian). 9 Osamu Yamada, Yoshinari Miyamoto, Mitsue Koizumi, Self-propagating high-temperature synthesis of SiC, J. Mater. Res., 1 (2), 275-279, (1986). 1 L.N. Satapathy, P.D. Ramesh, Dinesh Agrawal, Rustum Roy, Microwave synthesis of phasepure fine silicon carbide powder, Mater. Res. Bull., 40,1871-1882, (2005). ll L.G. Abovyan, Activated combustion of S1O2 - Al - C system and synthesis of S1C/AI2O3 composite powders, Physics of Combustion and Explosion, 2, 51-55, (2000) (in Russian). I2 J.H. Lee, C.Y. An, C.W. Won, S.S. Cho, B.S. Chun, Characteristics of Al203-SiC composite powder prepared by self-propagating high-temperature synthesis process and its sintering behavior, Mater. Res. Bull., 35, 945-954, (2000).

46

· Processing and Properties of Advanced Ceramics and Composites

MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

production of spinel- and carbide- based composites requires further elaboration of theoretical and technological aspects of materials formation during combustion synthesis. The paper is concerned with the investigation devoted to CS production of MgAl204/SiC phase composite. INVESTIGATION METHODS Powdered magnesium carbonate, silica, aluminum dust and carbon were used as initial raw materials. Preliminarily weighed components were mixed and sieved. Test samples were semidry-molded in the form of cylinders and plates using polyvinyl-alcohol solution as a binder. The samples obtained were dried in cabinet drier at the temperature of 100 °C up to entire moisture removal. In order to initiate the CS process the dried samples were placed in a furnace heated to the temperature of 800 - 900 °C. Once the samples were warmed up to the propagation of synthesis wave accompanied by bright glow was observed (Figure 1).

Figure 1. Propagation of combustion wave over a sample. The synthesis products were examined using differential thermal analysis (DTA) as well as X-Ray phase, electron diffraction, and elementary analyses. Electron diffraction and chemical analyses were performed using JSM-5610 LV microscope equipped with EDX JED-2201 JEOL system for making chemical analysis. X-ray phase analysis was done using DRON-3 device and diffraction patterns were interpreted by means of JCPDS File. RESULTS AND DISCUSSION Differential thermal analysis (DTA) of as-mixed samples with various contents of silica and aluminum (Figure 2) has indicated that the temperature required for initiating the reaction of exothermal synthesis is 530 °C to 600 °C and depends on the content of components. The general sequence of processes is analogues for all the compositions and is as follows: removal of bound water and decomposition of hydroxycarbonate complexes occur up to the temperature of 500 °C. Then at the temperature of about 550 °C magnesium carbonate is decomposed and aluminum interacts with carbon oxide reducing the latter to carbon. The accomplishment of aluminum oxidation by carbon oxide is followed by the process of spinel formation from aluminum and magnesium oxides. Since oxides are rather active at the instant of precipitation, the kinetics problems are actually absent and the amount of heat released is sufficient for sustaining combustion. Reduction of silica is also observed to proceed in the wave of exothermal synthesis and silicon resulting from the reaction forms silicon carbide with carbon.

Processing and Properties of Advanced Ceramics and Composites

· 47

MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

Figure 2. DTA curves of samples taken from mixtures with various aluminum and silica contents.

48

· Processing and Properties of Advanced Ceramics and Composites

MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

Thus, the phase formation proceeding during combustion can be described by the following reactions: MgC03 = MgO + C0 2 4A1 + 3C02 = 2A1203 + 3C MgO + AI2O3 = MgAl204 3Si02 + 4A1 + 3C = 3SiC + 2A1203

(1) (2) (3) (4)

As is evident from the data of DTA the mixtures with higher aluminum content (the Al-30 %, SiO2-20 % and Al-40 %, SiO2-20 % curves) are characterized by a shift of exothermal peak towards a low-temperature region by 5 °C to 10 °C with increasing its intensity. This is associated with predominance of high-exothermal process of aluminum oxidation, reaction (2). When the silica content is increased (the Al-30 %, SiO2-20 % and Al-30 %, SiO2-40% curves) DTA curves reveal that exothermal peaks shift to high-temperature region by 40 °C to 50 °C and their intensities are decreased.. This is attributed to predominance of a lower exothermal process of silica reduction and formation of silicon carbide, reaction (4). From the above synthesis reactions it is reasonable to expect that silicon carbide would be formed without additional incorporation of carbon into mixture since the latter is formed in the process of carbon dioxide reduction. However this is not the case due to the oxidation of formed carbon with oxygen of air which is present in sample pores or diffused from outer layers. XRD analysis (Figure 3) made at certain aluminum-silica relations of initial mixture shows that the maximum spinel output is observed when the amount of silica is decreased with corresponding increase of magnesium carbonate content.

Figure 3. Diffraction patterns of CS-produced composite ceramic materials.

Processing and Properties of Advanced Ceramics and Composites

· 49

MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

The enhancing of silicon carbide output can be attained by increasing the silica content and evidence of this can be also found in diffraction patterns. However, when content of silica exceeds 40 % its unreacted phase is found in the form of quartz. The presence of corundum (the Al-30 %, SiO2-40 % curve) is conditional upon predominance of silica reduction process with subsequent interaction between silicon and carbon resulting in formation of carbide, reaction (4). As is clear from diffraction patterns the formation of aluminum carbide that occurs at 40 % mixture content of aluminum of is probably associated with some excess of the latter. The aluminum carbide formation results in deficiency of carbon and the process of silicon carbide formation is not completed as demonstrated by precipitation of free carbon phase. Thus, in addition the processes that proceed according to the following reactions take place: 4A1 + 3C = AI4C3 3Si02 + 4A1 = 3Si + 2A1203

(5) (6)

The physicochemical characteristics of CS-produced samples vary widely depending on the composition. The regularity revealed in composition-dependant changing of the sample apparent density and porosity (Figure4) consists in the fact that porosity of samples is increased within the whole temperature range with increasing the magnesium carbonate content. This is caused by loosening of structure under the action of carbon dioxide formed during combustion. The anomalous behavior of this dependence observed at 20 % content of aluminum arises from the fact that this quantity is not sufficient for completing the reaction. Therefore, when the magnesium carbonate content is low the exothermal effect is insufficient for initiating the reduction of silica and the latter remains actually unchanged forming the sample frame with practically constant porosity. However when the magnesium carbonate content is further increased the exothermal effect of combustion becomes sufficient for initiating silica reduction that results in the formation of products the structures of which differ from those of initial frames. As it has been noted above the sample porosity is increased due to structure loosening during combustion.

Figure 4. Dependence of sample apparent density (a) and porosity (b) on composition of initial mixture.

50

· Processing and Properties of Advanced Ceramics and Composites

MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

In general, the samples have the following values of physicochemical properties: the apparent density is 1000 to 1600 kg/m3, the porosity 35 % to 70 % and the compression strength 20 MPa to 55 MPa. The low values of strength are attributed to high porosity of obtained material. The temperature coefficient of linear expansion observed within the range of 20 °C to 900 °C is (6.0 - 8.0)· 106K"' and corresponds to the average for spinel ceramics. The low material electric conductivity equal to 105 - 106 Ohm-cm is caused by the presence of silicon carbide. The microstructure of samples containing less than 40 % of silica the SEM micrographs of which are shown in Figure 5 appears as crystalline spinel aggregates of various shapes as well as silicon carbide fibrous crystals (Figure 5 a). When silica content exceeds 40 % acicular aggregates containing 80 % AI2O3 and 20 % S1O2 are formed (Figure 5 b) due to predominance of silica reduction with precipitation of aluminum oxide (4). This is also substantiated by the above- mentioned data of X-ray phase analysis.

Figure 5. SEM micrographs of the following samples: a - S1O2 content of initial raw mixture is less than 40 %; b -S1O2 content of initial raw mixture is above 40 %. CONCLUSION It is shown that the possibility exists of producing the Mg AI2O4/S1C composite ceramic material by combustion of mixture containing magnesium carbonate, silica, aluminum and carbon. The data of differential thermal analysis have indicated that the onset of combustion is

Processing and Properties of Advanced Ceramics and Composites

· 51

MgAI204/SiC Composite Ceramic Material Produced by Combustion Synthesis

within the range of 500 °C to 600 °C. The processes proceeding on combustion have been studied. It is established that when the content of silica is increased, the process of its reduction by aluminum is predominant with formation of corundum and silicon carbide. Furthermore, when the aluminum content is increased aluminum carbide is formed. The explanation is provided for material porosity dependence on initial composition. The microstructure investigations have revealed the presence of fibrous silicon carbide crystals.

52

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

FINITE ELEMENT ANALYSIS OF SELF-PROPAGATING HIGH-TEMPERATURE SYNTHESIS OF STRONTIUM-DOPED LANTHANUM MANGANATE Sidney Lin and Jiri Selig Chemical Engineering Department Lamar University Beaumont, Texas 77710 U.S.A.

ABSTRACT A two-dimensional finite element model is developed to study the reaction kinetics and heat transfer during the Self-propagating High-temperature Synthesis of Lao.6Sr0.4Mn03, a cathode and interconnect material used in solid oxide fuel cells. Conduction, convection, and radiation are considered in the heat transfer. Both oxygen generation from a solid peroxide reactant and diffusion from the surrounding gas phase are included in the mass transfer. Calculated spatialtemporal temperature profile, reaction generation rate, reaction conversion, and flow pattern of surrounding gas during the reaction are reported in this work. Hot spots are found at the corner near the ignition point shortly after the ignition. The model provided a simple and reliable way to design a large scale production of Lao.6Sr0.4Mn03. INTRODUCTION Fuel cells have a higher efficiency than internal combustion engines and are environmentally friendly because there are no CO2, NOx or SOx emissions generated in the operation. Because of their advantages, fuel cells are a good candidate to be used as a highly efficient and environmentally friendly device for power generation using an alternative energy supply. The development of energy efficient devices has become a demanding challenge in the past few years because of the high fossil fuel price and the uncertainty of its long-term supply. Solid oxide fuel cells (SOFCs) have advantages over other types of fuel cells (PEM, for example) because it can operate on various fuels such as methane, propane, methanol, ethanol, gasoline, or oil derivatives |l ' 2 l PEM is only limited to hydrogen as a fuel and it is also very sensitive to CO and sulfur in the fuel streamf3'. In general, SOFCs have a high tolerance to CO and sulfur in the fuel stream. An SOFC consists of an anode, an electrolyte, and a cathode. Strontium-doped lanthanum manganate (LSM, Lai.xSrxMn03) and its derivatives are the most widely used cathode materials for SOFCs i2]. They have excellent catalytic activities for oxygen reduction. In addition, their good oxygen ion conduction properties at elevated temperatures ' 4| and their thermal expansion coefficient are close to commonly used YSZ electrolyte '3|. Other compounds with a Perovskites structure are used as a material for SOFC cathode. For example, Sr-doped LaFeO.3 and Sr-doped LaCo03 have higher electron conductivity than LSM and are very good ionic conductors, but their coefficients of thermal expansion do not match well with the YSZ electrolyte. Also, Sr-doped PrMn03 has better properties than LSM, but Pr is a very costly material. LSM is traditionally produced by sintering its oxides or carbonates at high temperatures (over 1,200 °C) for several hours, which is an energy intensive process. Other exotic synthesis techniques, such as co-precipitation and sol-gel, which consist of many steps and use costly raw

53

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

materials, are difficult to scale up |5,61. An economical and efficient process is needed for the synthesis of cathode materials to lower the production cost of SOFC's, which has been identified as a major component of SOFC cost17'. Self-propagating High-temperature Synthesis (SHS) was developed in Russia in the late 1960s and has been used to synthesize many ceramic materials including oxides, nitrides, carbides, and metal hydrides. This process is highly exothemiic. Only small amount of ignition energy is needed for the reaction to occur because heat generated from the combustion is sufficient to sustain the reaction. SHS process is conducted at room temperature, which allows the synthesis not to be energy intensive. The fast combustion front movement (1-100 mm/sec) enables a large-scale production in a short period of time. In addition, its fast cooling after the reaction allows the formation of ceramic powders with very fine grains and metastable compositions. The concept of a continuous SHS production of Perovskites to reduce the production cost as well as capital investment has been proven '8|. An understanding of the temperature change during the SHS reaction and the combustion rate is necessary to design a large-scale SHS production. Although there are many published models of SHS, most of them are developed for the syntheses of carbides and simple oxidesf9'10'. To our knowledge, there is no detailed model of SHS of LSM. The goal of this work is to develop a mathematical model using the finite element analysis to help the design of a large batch scale SHS of LSM cathode materials. MATHEMATICAL MODEL Finite element analysis is a technique of solving a system of partial differential equations with irregular boundaries, in which computational domain is divided (meshed) into small but finite elements. Each element in the mesh is characterized by a set of equations which are then solved simultaneously. A simplified model including heat and momentum transfers is used in this work and is described below. Figure 1 is a sketch of the modeled system which consists of a pellet 3 cm long and 2.22 cm (7/8 inch) in diameter placed on a quartz holder 0.39 cm thick. The reaction system is placed inside an oxygen environment.

Figure 1. A sketch of the experimental set-up and dimension of the mathematic domain used in the modeling.

54

· Processing and Properties of Advanced Ceramics and Composites

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

Momentum Balance Weakly compressible Navier-Stokes equations (Eqns. 1 and 2) are used to model the oxygen flow through the reactor to compensate the density change caused by large temperature change during the reaction. dp dt

du p— + pu · Vu = -p + V dt

= ν·(ριι)=0 7(Vu

+ (Vu))-i^7-U(Vu)ll

(1)

(2)

Oxygen enters the reactor in a laminar flow with a parabolic profile specified by equation (3) and mean inlet velocity in x-direction is 0.02 m/s. u = \.5u,„

(r- 0.00625 f

(0.025-.00625)"

(3)

Mass Balance Overall mass balance is specified by equation (4). This mass balance takes into account convective and diffusive mass transfers it also includes a source term which is the reaction rate. 1

dt

- + V(-DiVCl) = Rl-uVCl

(4)

Flow field obtained from momentum balance is used to calculate velocity vectors necessary to calculate the convective mass flux. Oxygen diffusion in the atmosphere is considered as a constant (3 cm2/s at 1,400 K) and the diffusivity in the porous pellet is defined by equation (5): D=D-

(5)

Porosity of our sample is 0.55 and tourtuosity can be taken as 4. Two oxygen sources are considered in our model: the one diffused from surrounding atmosphere and the one form the decomposition of Sr02. Oxygen from both sources is used to oxidize manganese metal to form M112O3 (Eqn. 6). Mn+ 0.75 0 2 -> 0.5 Mn 2 0 3

(6)

From our calculation the decomposition of Sr0 2 (Egn. 7), provides 71 % oxygen needed to complete the oxidation of Mn metal. The balance of oxygen comes from the surrounding atmosphere by diffusion.

Processing and Properties of Advanced Ceramics and Composites

· 55

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

Sr0 2 -» SrO + 0.5 0 2

(7)

The decomposition of Sr02 is considered to be a first order reaction which is proportional to the concentration of Sr0 2 and is shown by equation (8): (8)

dt

where ks,02 is sufficiently large number to provide fast decomposition. Oxygen balance is defined by equation (9): dC() dn -~^-KoCSa) - - f 0.757 *„„ (9) dt dt The constant of 0.75 is used because of the stoichiometry- for each mol of Mn oxidized only 0.75 mol of 0 2 is consumed. Energy Balance Heat transfer mechanisms considered in this work include the convection and radiation heat loss from the pellet surface to the surrounding oxygen and the conduction heat transfer inside the pellet and to the sample holder underneath. Conduction and convection heat transfer for the whole system (pellet, pellet holder and surrounding oxygen) are expressed as equation (10): pCp^

+

u-V ) = V - ( * V 7 > G

(10)

Interior boundary conditions are specified by equation (11):

n · [ ( W - PfJ,ux)-

(k2T2 - p2Cp2T2u2)] = q{}

(U)

In equation (11) the difference of convective and conductive heat transfer between arbitrary domains 1 and 2 is equal to external heat flux qo. Because equation (11) contains only conductive and forced convection heat losses, radiation (qr), natural convection (qnc), and ignition (qig) fluxes are added as an external flux (q0) by equation (12): (},=
(12)

Radiation heat loss from the pellet surface is calculated using the Stefan-Boltzmann equation (Eqn .13): qr = ea{T*-T*)

56

· Processing and Properties of Advanced Ceramics and Composites

(13)

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

In this work, a radiation heat loss is considered by assuming ε = 0.7. Because oxygen flow rate is low, natural convection heat loss from the pellet surface to the surroundings is calculated by: <1,=K(TJ-T.)

(14)

Besides convective and radiative heat flux, ignition heat flux is also specified at the right edge of the pellet by equation (15): qig =3x\0\W/m2);

(t
(15)

For time less than 1.5 seconds, inward heat flux of 3.0x106 W/m2 is imposed onto the ignition area, which is a circle at the center of the right edge with a 0.6 cm diameter. The value of ignition flux assumes 100 % heat flow from graphite igniter to the pellet. Temperatures at the oxygen inlet and the ambient temperature are kept at 300 K (T x = 300 K). Oxygen exiting from the reactor has a convective flux boundary condition. Heat Source Heat source (Q) is expressed by the set of two equations. -Ufr=&p\CpdT

(16)

^- = (\-rjr)'kre^

(17)

Activation energy (Eo), pre-exponent constant (kr), and reaction order (n) are calculated from our experimental data. The above equation is only valid when the oxygen concentration is more than zero, which assumes that the overall reaction rate is equivalent to the reaction rate of manganese oxidation'']'. RESULTS AND DISCUSSIONS Reaction rate (1/s) of formation of LSM at different times after ignition is shown in Figure 2. It can be seen that five seconds after ignition (Fig. 2a), the reaction occurs near the ignition region. At this time, the reaction rate is symmetric along the r-axis. At 10s (Fig. 2b), the reaction rate slows down because the initial ignition heat flux is turned off. Also heat is dissipated by conduction and temperature decreases resulting in lower reaction rate. The reaction front slowly propagates predominately in r-direction until it reaches the top pellet surface (Fig. 2c) and the reaction rate sharply increases (Fig. 2d). This can be explained by the accumulation of reaction heat generated. The heat loss to the oxygen atmosphere by convection and radiation is lower and so is to the reacted pellet which has lower thermal conductivity [2 W/(m s)] than reactant mixture [5 W/(m s)]. Thus, the temperature increases as well as the reaction rate. Reaction does not occur near the pellet holder at t = 13 s because of the low temperature in that region which caused by the pellet holder which acts as a heat sink.

Processing and Properties of Advanced Ceramics and Composites

· 57

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

Figure 2. Reaction rate in 1/s at different times after ignition (5, 10, 13, 15, 20, 25 s) showing the unsteady reaction propagation during SHS. The ignition point is at the mid point of the left surface. The x-axis represents the distance from the ignition surface and the y-axis represents the radius of the pellet. The heat transfer is only limited to positive x-direction where the pellet is still unreacted. The reaction rate increase until reaches steady maximum value of 0.22 1/s and propagates though the rest of the pellet (Figs. 2e and 2f), moving faster in the top half of the pellet then near the

58

· Processing and Properties of Advanced Ceramics and Composites

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

pellet holder. The faster reaction rate in the upper half of the pellet is also observed during our experiments in which the reaction front propagates faster in the upper half of the pellet. Figure 3 shows the temperature profile during SHS reaction at different times (5, 10, 13, 15,20,25 s).

Figure 3. Temperature profiles during SHS at four different times after ignition (5, 10, 13, 15, 20, 25 s) with temperature in Kelvin. Reaction propagates from left to right. The ignition point is at the mid point of the left surface. The x-axis represents the distance from the ignition surface and the y-axis represents the radius of the pellet. Temperature profile follows closely the reaction rate. Initially, there is temperature increases near the ignition region due to the ignition flux (Fig. 3a). Figures 3b and 3c show the cooling which corresponds with the decrease in the reaction rate. The sharp increase in the reaction rate (Fig. 2d) results in a dramatic temperature increase in the same region at the same time (Fig 3d). As the reaction propagates through the rest of the pellet in r and x-directions

Processing and Properties of Advanced Ceramics and Composites

· 59

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

temperature follows similar trend. Surprisingly, highest temperature at a particular time occurs approximately 5 mm behind the reaction front and not at the location of highest reaction rate. This can be explained by the difference in the thermal conductivities of product and reactant mixture. This phenomenon was also observed during thermal imaging studies of LSM fll l Figures 2 and 3 suggest that a hot-spot is being formed after the ignition period; reaction is unsteady for first 15 seconds before a steady propagation is achieved. Temperature near the pellet holder remains lower because of the fast conductive heat transfer from the pellet to the holder. Our calculations indicate that, on average, the heat flux to the pellet holder is one order of magnitude higher than to the surrounding oxygen atmosphere. Reaction temperature in our model reaches up to 1,200 K. In experiments we have recorded similar temperature in the center of the pellet. The rate of propagation in the model is, on average, 0.8 mm/s, while in the experiments an average velocity of 0.5 mm/s is observed. This dissimilarity can be explained by a cracking of the reacting pellet during experiments. Oftentimes the pellet would split into a several smaller disks which reduces the conductive heat transfer and slows down the reaction because of the increased convective heat flux to the oxygen atmosphere. Reaction conversion increases more rapidly on the upper half of the pellet, while the region near the sample holder has a slow increase in conversion due to lower reaction rate (Fig. 4). The whole pellet eventually reaches an almost uniform conversion except near the pellet's outer surface where the pellet remains partially unreacted due to a heat lost to surrounding. This resembles the behavior during experiments where the lower part of the pellet was often unreacted due to the lower temperature during the SHS process.

Figure 4. Reaction conversion during SHS reaction at different times as a function of radial position (with r = 0 being the center axis of the pellet) at a distance of 2 cm from the ignition spot.

60

· Processing and Properties of Advanced Ceramics and Composites

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

CONCLUSIONS A two-dimensional finite element model was developed using data obtained from experiments and literature. Our calculation shows a non-uniform temperature profile and combustion front movement during the synthesis. Hot spot was formed near the top surface of the reacting pellet at the beginning of the reaction. With the propagation of the reaction front into the pellet, the temperature and reaction rate become more uniform. Although the oxygen diffusion resistance is not included in this work, our model can provide good estimates of oxygen flow pattern, pellet temperature profile, reaction rate, and reaction conversion. ACKNOWLEDGMENTS This work was partially supported by Lamar University Research Enhancement Grants under project No. 210381 , Texas Air Research Center (TARC) under contract No. 066LUB0066A, and Texas Hazardous Waste Research Center (THWRC) under contract No. 067LUB0961. NOMENCLATURE C p - specific heat C - concetnratrion D - diffusion coefficient Eo - activation energy ΔΗΓ - heat of reaction hnc - average nature convection coefficient I - inertia k - thermal conductivity kr - pre-exponent ksr02 - rate constant for Sr0 2 decomposition n - reaction order n - normal vector to a boundary p - pressure Q - heat source (heat sink) q - heat flux R - rate of reaction T - temperature T x - ambient temperature Tj - film temperature t - time u - velocity u - oxygen velocity vector Umean - mean inlet velocity in x-direction ε-emissivity η - dynamic viscosity ηΓ - conversion Kdv - dilatational viscosity v - kinematic viscosity p - density σ - Stefan-Boltzmann constant

Processing and Properties of Advanced Ceramics and Composites

· 61

Finite Element Analysis of Strontium-Doped Lanthanum Manganate

REFERENCES l W. Sangtongkitcharoen, S. Vivanpatarakij, N. Loasiripojana, A. Arponwichanop, S. Assabumrungrat, Performance analysis of methanol-fueled solid oxide fuel cell system, Chemical Engineering Journal, 138,436-41 (2007). 2 K.Yasumoto, N. Mori, J. Mizusaki, H. Tagawa, M Dokiya, Effect of Oxygen Nonstoichiometry on Electrode Activity of La].xAxMn3± Cathode, Journal of The Electrochemical Society, 148, A105-11 (2001). 3 Raymond J. Gorte, Recent Developments Towards Commercialization of Solid Oxide Fuel Cells, MChE Journal 51, 2377-81 (2005). 4 E. Krogh Andersen and I. G. Krogh Andersen, Kinetics of Oxidation of Fuel Cell Cathode Materials Lanthanum Strontium Manganates(III)(IV) at Actual Working Conditions: In Situ Powder Diffraction Studies, Journal of Solid State Chemistry, 141, 235-40 (1998). Vuk Uskokovic and Miha Drofenik, Synthesis of lanthanum-strontium manganites by oxalateprecursor co-precipitation methods in solution and in reverse micellar microemulsion, Journal of Magnetism and Magnetic Materials, 303, 214-20 (2006). 6 Cui Xiulan and Liu Yuan, New methods to prepare ultrafine particles of some perovskite-type oxides, Chemical Engineering Journal, 78, 205-9 (2000). 7 Q. Ming, M. D. Nersesyan, J. T. Richardson, Dan Luss, A New Route to Synthesize La]. xSrxMn03, Journal of Material Science, 35, 3599-3606 (2000). 8 Sy-Chyi Lin, James T. Richardson, Dan Luss, Continuous Synthesis of YBa2Cu306+x by Thermal Explosion in a Rotary Kiln, Physica C, 260, 321-26 (1996). 9 B. B Kniha, B. Formanek, 1. Solpan, Limits of Applicability of the "Diffusion-controlled Product Growth" Kinetic Approach to Modeling SHS, Physica B, 355, 14-31 (2005). 10 Alexander S. Mukasyan and Karen S. Martirosyan, Combustion of Heterogeneous Systems: Fundamentals and Applications for Materials Synthesis, 1-39 (2000), ISBN: 8178952696. 11 Maxim V. Kuznetsov, Ivan P. Parkin, Daren J. Caruana, Yuri G. Morozov, Combustion Synthesis of Alkaline-earth Substituted Lanthanum Manganites; LaMn03, Lao.6Ca0.4Mn03 and Lao.6Sro.4Mn03, Journal of Materials Chemistry, 14, 1377-82 (2004).

62

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Reaction Forming and Polymer Processing

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

COMPARISON OF BULK AND NANOSCALE PROPERTIES OF POLYMER PRECURSOR DERIVED SILICON CARBIDE WITH SINTERED SILICON CARBIDE Arif Rahman1, Suraj C. Zunjarrao1 and R. P. Singh12 1

Mechanics of Advanced Materials Laboratory School of Mechanical and Aerospace Engineering Oklahoma State University Stillwater, OK, USA 2

School of Mechanical and Aerospace Engineering Oklahoma State University Tulsa, OK, USA ABSTRACT Polymer precursor derived silicon carbide ceramics made by polymer infiltration and pyrolysis (PIP) technique offer various advantages over conventional processing, such as near net shape fabrication, relatively low temperature processing and the ability to tailor the microstructure. In this work, a comparison has been conducted between properties of polymer precursor derived silicon carbide and commercially available sintered silicon carbide. Silicon carbide samples are fabricated by pyrolysis of allylhydridopolycarbosilane at different temperatures. Average grain size is determined using XRD. Nanoscale mechanical properties in terms of modulus and hardness and bulk scale characterization in terms of flexure strength have been measured. Nanoscale as well as bulk properties of commercially available silicon carbide were higher than PIP derived SiC, however, for the latter some improvement in properties is observed with increasing processing temperature. INTRODUCTION Properties of silicon carbide such as high strength, modulus and creep resistance [1], along with its retention of properties at high temperatures made it a potential candidate for research related to various applications. Its resistance to oxidation up to 1600CC and very high dissociation temperature of 2850CC [2] along with excellent thermal conductivity, 31 W/m °C and a very low coefficient of thermal expansion, 4.7 x 1CT6/ °C at 1200°C [3] makes it favorable for thermal applications. Moreover, nano-crystalline ceramics show great potential for unique properties like super hardness and very high toughness [4-7] which has generated interest into the nanoscale properties of ceramics. These unique properties are governed primarily by the underlying microstructure and improvement at a significant scale can be achieved with appropriate tailoring. Different techniques have been used over the years to fabricate silicon carbide and powder sintering is one of the most common of all. It includes powder preparation, powder processing, powder consolidation, and sintering. Basic requirements for this kind of fabrication process is high pressure and/or high temperature which limit the kind of materials that can be produced. For example, silicon carbide fibers degrade at 1200CC [81 which makes fabrication of SiC fibers reinforced ceramic matrix composites extremely difficult using sintering. Plasma-activated sintering is another kind of sintering which can be used to fabricate SiC ceramics with better mechanical properties under lower temperature with shorter sintering time than traditional sintering [9]. It involves plasma generation, resistance heating and pressure application. However, lower sintering

65

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

temperature and shorter sintering time impedes grain growth which makes yielding of finer microstructure difficult [9]. Reaction processing is one of the alternatives that has been widely used instead of sintering [10]. This method involves a solid-liquid reaction for the synthesis of ceramic compound, and offers the capability to fabricate refractory ceramics at lower cost compared to conventional sintering. However, the presence of residual, free silicon limits the maximum temperature to below 1400°C at which SiC can be used. Moreover, this free silicon also reduces the resistance to corrosion and oxidation as compared to stoichiometric silicon carbide. Among the non-powder based processing techniques, chemical vapor deposition (CVD) involves the deposition of a solid material on an activated or heated surface by reaction with a gaseous phase. Control over the processing temperature is critical for CVD as this regulates the process thermodynamics and kinetics [11]. Challenges in CVD include gas phase nucleation and homogenous reaction occurring due to supersaturation of the reactive species in the gaseous phase. Also control of microstructure and residual stresses is a matter of concern in CVD process. Chemical vapor infiltration (CVI), another non-powder based processing, involves infiltration of porous structures or performs with inorganic vapors. CVI is widely used for fabrication of silicon carbide matrix composites reinforced by continuous silicon carbide fibers. However, the CVI processing is usually very slow due to low diffusion rates. For this investigation, another non-powder based process, polymer infiltration and pyrolysis (PIP) has been used. Compared to other fabrication processes, this, method is simpler and involves thermal treatment of organometallic compounds under an inert atmosphere to condense into inorganic materials. Polymer precursor is used as starting material for this process which undergoes polymer to ceramic conversion when heated above 800°C. Using the PIP technique allows some advantages such as near net shape fabrication, comparably low temperature and almost pressure less fabrication. From different available SiC polymer precursors we have used allylhydridopolycarbosilane (AHPCS, Starfire Systems Inc., Malta, New York) which is an ultra high purity precursor that yields a near stoichiometric ratio on complete pyrolysis [12]. The main challenge of using polymer precursor is the increase in density observed while conversion occurs from polymer precursor to ceramic. To address this issue, multiple polymer infiltration and pyrolysis cycles are required to achieve suitable densification. The focus of this paper is to characterize mechanical and physical properties of AHPCS derived SiC for a range of processing temperature and compare with commercially available sintered silicon carbide. Mechanical properties in terms of biaxial flexure strength was determined using a ring-on-ring test. Since, at higher processing temperature, AHPCS derived SiC develops cracks and porosity due to volume shrinkage and development of hydrogen gas [13-16], to get the true local property of the silicon carbide nanoindentation was conducted. Physical properties in terms of density, porosity and thermal conductivity were determined. Crystallite sizes, as functions of processing temperature, were characterized using x-ray diffraction. Some of the properties of commercially available sintered silicon carbide were determined using the same process as used for AHPCS derived SiC and the rest of them were obtained from the manufacturer's website. Finally, a comparative study of different properties for both AHPCS derived SiC and commercially available sintered silicon carbide is presented. MATERIAL FABRICATION Fabrication of SiC as a bulk sample from AHPCS is tricky and lengthy since shrinkage occurs while precursor converts to ceramic SiC. Using reinforcing agents such as SiC whiskers

66

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

or fibers helps in this case as these fillers substitute a considerable volume in the composites and do not change in volume significantly through out the process. But, this investigation involves only the characterization of SiC derived from AHPCS. So, reinforcing agents such SiC fibers or whiskers were not used. However, the fabrication process adopted is similar to the process used for fabrication of ceramic composites using polymer infiltration and pyrolysis.

Figure 1: Schematic of PIP process for fabrication of bulk samples for bulk characterization At first, SiC powder was prepared by pyrolyzing AHPCS to 900°C, 1150°C, 1400°C and 1650°C with a heating rate of l°C/min till 650°C and held at that temperature for 10 minutes, then 3°C/min till the desired final temperature. All specimens were held at the final temperature for 1 hour to ensure thermal equilibrium. Due to the release of hydrogen gas during the process, the acquired materials contained large voids. These were crushed into powder by a planetary ball mill (PM-100, Retsch GmbH, Haan, Germany) in a tungsten carbide bowl (WC) with 10 mm WC balls for 12 min at 300 rpm. These powders were used as reinforcing fillers in the next step of the fabrication process. The powders were then mixed with a small amount of polymer precursor (3% by weight of the milled powder). Then, the mixture was compacted using a steel ram-cylinder setup into short cylinders of dimensions of 25.4 x 15 mm. Thin discs of a graphite sheet were used at the top and bottom of the powders being compressed. These discs prevented the powders from sticking to the ram or the cylinder. To form plugs, hand compaction with a maximum load of ~445 N was used. The resulting plugs were then heated to the final temperature (900°C, 1150GC, 1400°C or 1650CC) using the same heating cycle. The resulting samples had a good dispersion of SiC particles in an amorphous silicon carbide (#-SiC) matrix; however, due to the release of the hydrogen gas during polymer pyrolysis pores were present in the plugs. Multiple PIP cycles were conducted to minimize porosity and to increase the material density. Figure 1 shows the schematics of the complete PIP processing used in the current study. The infiltration was carried out under vacuum with a repeated 1 hour cycle for 4 hours and 1 minute purge in between of each cycle. The plugs were cut into 1 mm thick discs using a precision sectioning saw (Isomet 1000, Buehler, 202 Lake Bluff, Illinois, USA) after the 3rd PIP cycle. The discs were subjected to infiltration and pyrolysis till the 8th cycle. Work by Ozcivici et al. [17] with polymer derived ceramic composites using this polymer system has showed that 8 infiltration cycles are sufficient to obtain the maximum achievable density. These discs were tested for biaxial flexure strength using

Processing and Properties of Advanced Ceramics and Composites

· 67

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

ring-on-ring test. MECHANICAL CHARACTERIZATION Bulk Characterization Bulk mechanical characterization was done using ring-on-ring (RoR) biaxial tests to achieve flexural strength for SiC samples. Figure 2 shows the schematic of the RoR fixture, which is an axisymmetric test done using two concentric rings of different size. The concentric rings in the current setup were made up of stainless steel and had bullnose edges with a radius of 0.3125 mm towards the loading side. This configuration employed a support ring diameter of 19.05 mm and the loading ring diameter of 6.35 mm, which insures valid fracture mode of the samples [18.].

Figure 2: Schematic of fixture for testing biaxial flexure properties using RoR. The specimens were loaded in the RoR fixture using a table-top test frame (Instron® 5567, Instron Corporation, Norwood, Massachusetts, USA). Displacement controlled loading at a rate of 0.5 mm/min. was used and the peak load at failure was recorded. The flexure strength was then determined from the peak load at failure as per Eqn. 1 [19], 3P

( 1

- ^ 2 22R

r 2 )

+

( vi

+

,)in^ ' r\

(i)

where v is the Poisson ratio of the specimen and was assumed to be 0.20 for SiC, a is the radius of the support ring, r is the radius of the load ring, and R and t are the radius and thickness of the disc specimen, respectively. Table 1 lists the biaxial flexure strength obtained for the bulk samples composed purely of SiC derived from AHPCS and fabricated using the PIP process. Though, strength of SiC processed at 1650CC could not be determined due to the oxidation of the pellet, the rest of the samples followed an interesting trend of increasing strength with increasing processing temperature. Table 1: Biaxial strength of the SiC-SiC composites fabricated by using PIP route and tested using RoR biaxial flexure test No. Material Biaxial Strength (MPa) Pyrolysis Temp. ( C ) 1 2 3 4

68

SiC-900 SiC-1150 SiC-1400 SiC-1650

55.30 ± 1.98 79.58 ±8.11 85.26 ± 5.28 Oxidized

900 1150 1400 1650

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

Nanoscale Characterization Nanoindentation was performed on silicon carbide samples derived from AHPCS pyrolized to 900°C, 1150°C, and 1400°C with a hold time of 1 hour to characterize mechanical properties in terms of hardness and modulus. Discs of SiC processed at different temperatures used in ROR test were mounted in epoxy and polished using a Ecomet 3 polisher (Buehler, Lake Bluff, Illinois, USA) to a mirror finish. Nanoindentation was done using a berkovich tip with a peak load of 25 mN.

Figure 3: Load displacement plots obtained during indentation of SiC processed at different temperatures for a hold time of lh. Peak indentation load of 25 mN.

(a) Hardness of AHPCS derived SiC

(b) Modulus of AHPCS derived SiC

Figure 4: Hardness and modulus determined by nanoindentation for SiC derived from AHPCS heated to 900°C, 1150°C, and 1400°C, as a function of processing temperature. Figure 3 shows load-displacement curve for 25 mN peak load for samples processed at different temperatures with a hold time of 1 hour. From figure 3, it was interesting to observe that

Processing and Properties of Advanced Ceramics and Composites

· 69

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

material processed at 900°C had the highest indentation depth and indentation depth decreased for samples processed at higher temperature. From this it can be interpreted that as processing temperature increases mechanical properties of SiC changes and makes it harder. Figures 4(a) and 4(b) show hardness and modulus of SiC samples as a function of processing temperature. Error bars show standard deviation. Figure 4(a) shows three points of hardness values as a function of processing temperature. Similarly, figure 4(b) shows three points of modulus values as a function of processing temperature. Material processed at 1400CC had the highest hardness as well as modulus value compared to materials processed at 900°C and 1150°C. From this it can be interpreted that increase in processing temperature results in better hardness and modulus value when it is held for 1 hour at final temperature. PHYSICAL CHARACTERIZATION Density and Porosity The density and porosity of the silicon carbide samples for different processing temperatures was determined after 8i/l cycle of infiltration and pyrolysis. The Ultrapycnometer 1000 (Quantachrome instruments, Pittsburgh, PA, USA) was used for measuring skeletal and true density for all the samples. For purging helium was used in the instrument. For skeletal density, ps, the sample was dried in a drying oven set at 120°C for 12 hours and then cooled to room temperature. The dry mass of the sample was then recorded using a high-resolution analytical balance. Then the sample was placed inside a measurement cell of the pycnometer to start the density measurement process. The pycnometer was programmed to purge with helium for 20 minutes and take average of 3 out of 5 readings with standard deviation less than 0.010%. For True density, discs obtained from different samples were crushed into powder by a planetary ball mill (PM-100, Retsch GmbH, Haan, Germany) in a tungsten carbide bowl (WC) with 10 mm WC balls for 12 min at 300 rpm. Then using the Ultrapycnometer 1000 the true density, pt, for the samples were determined using the same procedure mentioned above.

(a) Density of AHPCS derived SiC

(b) Porosity of AHPCS derived SiC

Figure 5: Density and porosity for SiC derived from AHPCS heated to 900°C, 1150°C, and 1400°C, as a function of processing temperature.

70

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

Buoyancy method [201 was used for determining porosity of the samples using a density measurement kit along with a high-resolution analytical balance. The sample was evacuated and saturated with high purity distilled water and the open pores were filled with the saturation liquid. This is done using a four-hours evacuation cycle with intermittent purging at every hour to release trapped air. The mass of the sample in saturation liquid (water), m2, was determined using the density determination kit. The temperature of the saturation liquid was recorded to correct the density variation of water as a function of temperature. The wet mass, ra3, of the sample was then determined by weighing in air. Before weighing, a kimwipes (cleaning tissue) was used to remove any liquid that remained on the surface of the sample and this was done quickly to avoid loss of mass due to evaporation. The sample was then dried in a drying oven set at 120° for 12 hours and cooled to room temperature to remove the saturation liquid from the sample and the dry mass, mi, was recorded. The bulk density, pb, was calculated as, m

i

x Pfi (2) m3 - m-2 where, mi is the mass of the sample in air, m2 is the mass of the saturated sample in the fluid medium (water), m3 is the mass of the saturated sample in air and p// is density of the medium used (water) at the recorded temperature which was taken from a table of density for water at different temperature (DensityKit-Sartorius). The open porosity, πα, in volume % was calculated as,

Pb =

τη3-ηιΛ

x 100 m.3 — πι·2 The apparent porosity, π(, in volume % was calculated as, πα —

Pt - Pb Pt

x 100

(3)

(4)

Where, p, is the true density determined using the Ultrapycnometer 1000. Finally, the closed porosity, TTC, in volume % was calculated as, 7TC =

7Γ* -

7Γα

(5)

Figure 5(a) and 5(b) show density and porosity of silicon carbide as a function of processing temperature. From figure 5(a) it is evident that skeletal density as well as true density increases as the processing temperature increases which indicates conversion of amorphous to crystalline structure as the temperature increases. Figure 5(b) shows decrease in closed porosity with increasing processing temperature, although open porosity increases. Decrease in closed porosity is also a factor which explains increase in density as processing temperature increases. Thermal Conductivity The thermal conductivity, λ, was measured for cylindrical samples of φ 25 x 5 mm, using a steady state axial flow setup [21] shown in Fig. 6. Eight T-type high precision copper-constantan thermocouples were used to measure the heat flow of eight points on two copper rods. Thick cylindrical teflon insulation was used to prevent heat loss through radial conduction from the setup. Temperature on both hot and cold side were controlled using Omega CN132 temperature/process

Processing and Properties of Advanced Ceramics and Composites

· 71

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

Figure 6: Axial heat flow setup for measuring thermal conductivity controller and analog controlled Cole Palmer chiller, model: 12750-00 respectively. To minimize the thermal resistance between the copper rods and the SiC samples, thin layer of high thermal conductivity paste (Omegatherm 201) was used. After ^24 hours, steady state condition was achieved and a Fluke 53 series II digital storage thermometer was used to record the temperature at eight different points. The data points on both hot and cold side followed a linear trend.

Figure 7: Thermal conductivity of SiC as a function of processing temperature To determine the temperatures at the sample interfaces, linear curves on hot and cold sides were extrapolated. These temperatures were considered as Thot and Tco/^. Heat flow at the sample interfaces were calculated using equation 6 and represented as Qhot and Qcoid' Finally, the thermal conductivity of the sample was determined using the Fourier law of heat conduction as shown in

72

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

equation 7. Γ

ΙΑ(« Q = -XA{-) Λ ^sample

~

[Uihot ' Wcold)*-1 sample ~ . Τψ ΖΓτ \ ^^.sampif.\ -* hot J- cold)

(6) (~. ^ '

Figure 7 shows thermal conductivity of silicon carbide processed at different temperatures. Samples processed at higher temperature show higher thermal conductivity which indicates conversion of silicon carbide from amorphous to crystalline structure at higher processing temperature. MICROSTRUCTURE CHARACTERIZATION To study the crystallite sizes, x-ray diffraction (XRD) studies were performed on the samples. Powder samples were prepared by wet milling in a planetary ball mill (PM-100, Retsch GmbH, Haan, Germany) for 4 hours in ethanol and then mounted on glass slide. Powder diffraction patterns were collected using Scintag PAD-X automated diffractometer with a CuKQ radiation (λ = 0.1540 nm) using a scanning rate of 0.5° per min and operating at 45 kV and 25 mA. Figure 8

Figure 8: Powder diffraction patterns of SiC derived from AHPCS heated to 900CC, 1150°C, 1400°Cand 1650°C. shows the XRD patterns obtained for various samples; data is offset to aid comparison. From the figure, greatly diffused peaks at 900°C indicates amorphous SiC and the peak intensity increases with processing temperature increase. SiC peaks increases gradually at 2Θ values of 35.7°, 60.2° and 72.0° which indicates formation and growth of nano-crystalline domains. Small peaks for residual tungsten carbide (WC) are also observed in the pattern. Among these peaks for WC, one peak lies very close to the SiC peak at 35.7°, but these peaks are attributed to SiC as WC peaks at 35.6° and 48.3° are expected to be of the same intensity according to JCPDS (ICCD 29-1131). The Debye-Scherrer equation was used to estimate the crystallite size from the peak broadening [22]. Peak broadening was determined using XF1T program in terms of full width at half-maximum (FHWM) [23]. Instrument broadening was accounted for determining the crystallite sizes at different temperatures. The crystallite sizes were found to be about 3.65 nm, 5.02 nm and 11.03 nm at 1150CC, 1400°C and 1650°C, respectively.

Processing and Properties of Advanced Ceramics and Composites · 73

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

COMPARISON OF PIP DERIVED SiC WITH SINTERED SILICON CARBIDE For comparison, Hexoloy® SA silicon carbide received from Saint-Gobain Ceramics (23 Acheson Drive, NY 14303) was used as commercially available sintered SiC. Hexoloy SA is claimed to be a sintered form of alpha silicon carbide. Table 2 shows different properties of sintered silicon carbide as well as SiC derived from AHPCS using PIP technique. It should be noted that mechanical properties of precursor derived SiC were comparable with those of Hexoloy® SA SiC sample. Hexoloy samples showed better biaxial strength and higher thermal conductivity as compared to polymer derived SiC, however, for the latter increase in strength as well as thermal conductivity was observed as a function of processing temperature. Table 2: Comparison of different properties of AHPCS derived SiC using PIP technique and Hexoloy® SA SiC. (* These specifications were obtained from manufacturer's website) Pronerty Hexolov® SA SiC AHPCS derived SiC Property Hexoloy SA ML. 9()()cc U5QGC HQQ0C Biaxial Strength (MPa) 263±75.67 55.3±1.98 79.58zb8.il 85.26±5.28 23.24 Hardness (GPa) 22.036 28.57* 24.66 226.691 Modulus (GPa) 410* 183.039 249.49 Density (gm/cc) 3.1* 2.43 2.5 2.73 Closed Porosity (%) 1.93 4.15 1.84 1.05 Thermal Conductivity (W/m-K) 65.6 2.39 3.83 12.03 Average Grain Size (nm) 4000-10000* 3.65 5.02 — CONCLUSIONS Silicon carbide derived from allylhydridopolycarbosilane (AHPCS) using polymer infiltration and pyrolysis (PIP) technique at different temperatures were characterized in terms of strength at bulk scale and in terms of modulus and hardness at nano-scale. Physical properties in terms of density, porosity and thermal conductivity were also determined. From ring-on-ring test on AHPCS derived samples as well as Hexoloy® SA sample, it was observed that Hexoloy sample showed better strength, however, for the former there was ~44% and ~54% increase in biaxial strength from SiC processed at 900°C to 1150°C and 1400°C, respectively. From nanoindentation, it was observed that mechanical properties were influenced by increasing processing temperature for AHPCS derived SiC. The highest values of hardness were observed to be around 25 GPa for material processed at 1400°C. Hardness value of Hexoloy sample was higher than AHPCS derived SiC, however, there was 12% increase in hardness value for material processed at 1400CC from 900°C. There was a 37% increase in modulus from material processed at 900°C to 1400°C. Though modulus of AHPCS derived SiC was lower than Hexoloy sample, improvement was achieved with increasing processing temperature. Density and thermal conductivity of AHPCS derived SiC followed an increasing trend with increasing processing temperature and great improvement in thermal conductivity can be obtained by complete nano-crystallization of a-SiC to 6-SiC [24,25] . These properties were higher for Hexoloy sample as it is formed from a-SiC. From x-ray diffraction, it was observed that with increasing processing temperature crystallite sizes increases. Average grain size for Hexoloy sample was in the range of 4-10 micron whereas grain size for AHPCS derived SiC were in range of 3-12 nm. Thus, changes in properties of SiC derived

74

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

from AHPCS with increasing processing temperatures were studied and compared with Hexoloy® SASiC. ACKNOWLEDGEMENT The authors wish to thank DOE for financial support on this project (Award No. DE-FC0705ID14673). We would also like to thank Dr. Hongbing Lu for letting us use the nanoindentation facility at polymer mechanics lab, Oklahoma State University. We are also thankful to Nitin Daphalapurkar, Oklahoma State University for his help and expertise in nanoindentation. REFERENCES [1] Tredway, W., 1998. pp. 1275-1275.

"Materials science - toughened ceramics".

SCIENCE, 282(5392),

[2] Schick, H. L., 1966. "Thermodynamics of certain refractory comounds". Academic Press, 1, pp. 1—402. [3] Munro, R., 1997. "Material properties of a sintered alpha-sic". JOURNAL OF PHYSICAL AND CHEMICAL REFERENCE DATA, 26(5), pp. 1195-1203. [4] Szlufarska, I., Nakano, A., and Vashishta, R, 2005. "A crossover in the mechanical response of nanocrystalline ceramics". Science, 309(5736), pp. 911-914. [5] Veprek, S., Nesladek, P., Niederhofer, A., Glatz, F., Jilek, M., and Sima, M., 1998. "Recent progress in the superhard nanocrystalline composites: towards their industrialization and understanding of the origin of the superhardness". Surface & Coatings Technology, 109(1-3), pp. 138-147. [6] Zhao, Y., Qian, J., Daemen, L„ Pantea, C , Zhang, J., Voronin, G., and Zerda, T., 2004. "Enhancement of fracture toughness in nanostructured diamond-sic composites". Applied Physics Letters, 84(8), pp. 1356-1358. [7] Liao, F , Girshick, S., Mook, W., Gerberich, W., and Zachariah, M., 2005. nanocrystalline silicon carbide films". Applied Physics Letters, 86(17).

"Superhard

[8] Clark, T. J., Arons, R. M., Stamatoff, J. B., and Rabe, J., 1985. "Thermal degradation of nicalon sic fibers". Ceramic Engineering and Science Proceedings, 6, pp. 576-588. [9] Jin, H.-Y., Ishiyama, M., Qiao, G.-J., Gao, J.-Q., and Jin, Z.-H., 2008. "Plasma active sintering of silicon carbide". MATERIALS SCIENCE AND ENGINEERING A-STRUCT URAL MATERIALS PROPERTIES MICROSTRUCTURE AND PROCESSING, 483(Sp. Iss. SI), JUN 15, pp. 270-273. [10] WASHBURN, M., and COBLENZ, W., 1988. "Reaction-formed ceramics". CERAMIC SOCIETY BULLETIN, 67(2), pp. 356-&.

AMERICAN

[11] Pierson, H. O., 1999. Handbook of Chemical Vapor Deposition (CVD): Principles, Technology and Applications. Noyes Publications.

Processing and Properties of Advanced Ceramics and Composites

· 75

Comparison of Polymer Precursor Derived Silicon Carbide with Sintered Silicon Carbide

[12] Interrante, L. V., Whitmarsh, C, Sherwood, W., Wu, H.-J., Lewis, R., and Maciel, G., 1994. "High yield polycarbosilane precursors to stoichiometric sic. synthesis, pyrolysis and application". Proceedings of the 1994 MRS Spring Meeting, Apr 4-8 1994, Son Francisco, CA, USA, 346, pp. 593-603. [13] Moraes, Kevin, V., and Interrante, Leonard, V., 2003. "Processing, fracture toughness, and vickers hardness of allylhydridopolycarbosilane-derived silicon carbide". Journal of the American Ceramic Society, 86(2), pp. 342-346. [14] Zheng, J., and Akinc, M., 2001. "Green statejoiningof sic without applied pressure". Journal of the American Ceramic Society, 84(11), pp. 2479-2483. [15] Berbon, Min. Z., Dietrich, Donald, R„ Marshall, David, B., and Hasselman, D., 2001. "Transverse thermal conductivity of thin c/sic composites fabricated by slurry infiltration and pyrolysis". Journal of the American Ceramic Society, 84(10), pp. 2229-2234. [16] Lewinsohn, A. C, Jones, H. R., Colombo, R, and Riccardi, B., 2002. "Silicon carbide-based materials for joining silicon carbide composites for fusion energy applications". Journal of Nuclear Materials, 307-311(2 SUPPL), pp. 1232 - 1236. [17] Ozcivici, E., and Singh, R., 2005. "Fabrication and characterization of ceramic foams based on silicon carbide matrix and hollow alumino-silicate spheres". Journal of the American Ceramic Society, 88(12), pp. 3338-3345. [18] Wereszczak, A. A., Swab, J. J., and Kraft, R. H., 2003. "Effects of machining on the uniaxial and equibiaxial flexure strength of cap3 ad-995 AI2O3". ARL Technical Report. [19] Ovri, J. E. O., 2000. "A parametric study of the biaxial strength test for brittle materials". Materials Chemistry and Physics, 66(1), pp. 1 - 5. [20] "ASTM-C830-00". American Society for Testing and Materials, West Conshohocken, PA. [21] ASTM, 1225. "Standard test method for thermal conductivity of solids by means of the guarded-comparative-longitudinal heat flow technique". [22] Cullity, B. D., 1978. Elements of X-ray Diffraction. Addison-Wesley Publishing Co. Unc, London. [23] Cheary, R. W., and Coelho, A. A., 1996. "Programs xfit and fourya, deposited in ccp 14 powder diffraction library, engineering and physical sciences research council, daresbury laboratory, warrington, england". [24] A, Solomon, A., Julien, F., Sang-Gyu, L., Sarma, K., Shripad, R., Ryan, L., L, Holman, P., and Kevin, McCoy, J., 2004. "The polymer impregnation and pyrolysis method for producing enhanced conductivity lwr fuels". American Nuclear Society, pp. 146 - 155. [25] Verrall, R. A., Vlajic, M., D., and Krstic, V. D., 1999. "Silicon carbide as an inert-matrix for a thermal reactor fuel". Journal of Nuclear Materials, 274(1-2), pp. 54 - 60.

76

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

PROCESS DESIGN AND PRODUCTION OF BORON TRICHLORIDE FROM NATIVE BORON CARBIDE IN LAB-SCALE D. Agaogullaria'*, I. Duman a a Istanbul Technical University, Faculty of Chemical and Metallurgical Engineering, Department of Metallurgical and Materials Engineering, Istanbul, Turkiye. ABSTRACT Turkiye owns the biggest boron mine reserves of the world, so it is a necessity to research the possibilities of obtaining high-tech products of boron. This study is oriented to provide one of two main inputs, boron trichloride (BCI3) gas and carbon/tungsten core, for production of boron fiber manufactured by well-hidden hot filament CVD method. In this study, the production of BCI3 from native boron carbide (B4C) was aimed. In the production process, two different originally designed set-ups were used. In the set-ups, the reaction occurred by passing chlorine gas (Cb) through B4C layer (10-110 g) which has different heap heights. The reactor was a vertical quartz tube heated up to 850 °C by a tube furnace with silicon carbide (SiC) heating element. Chlorine as reaction gas and nitrogen as purging gas were dried in two different columns with calcium chloride (CaCl2) and molecular sieve package. Flow rates of Cl2, pressure and temperature changes in the set-ups were controlled respectively by rotameter, mass-flow controller (MFC), mercury U-tube manometer, thermocouple and thermometers which were placed in different zones. The product gases were separated in consecutive distillation columns and waste gases were conveyed to a rashing ring packaged gas scrubber. In pipelines, traps and condensers, various metal (i.e. Fe, Al, W, Si, and Cr) halides depending on the impurities of B4C were detected in different phases. The products in form of gas, liquid and solid were analyzed with FTIR, AAS, XRD and SEM/EDS. Production efficiencies (up to 85 %) and consumption rate of CI2 (up to 91 %) were assigned. Keywords: Boron Carbide, Boron Trichloride, Vertical Tube Furnace, Chlorine, FTIR INTRODUCTION Turkiye produces raw boron products (ulexit, tincal, colemanite, hydroboracite and szaybelite concentrates), refined boron products (calcinated tincal, calcinated ulexit, calcinated colemanite, boric acid, borax pentahdyrate, borax decahydrate, sodium perborate tetrahydrate, sodium perborate monohydrate and anhydrous borax) and special boron products (boron carbide and boron nitride)1. However, every step towards high-tech products provides a huge added value to raw and refined materials. The aim of this study is to produce the boron trichloride gas (BC13) which takes part in the special boron chemicals group, from native boron carbide. It is generally used for the production of boron fibers by chemical vapor deposition, for derivation of high performance fuel and rocket fuel, for synthesis of boron hydrate and boron nitrate, for dehydration and polymerization reactions. It is also used in electronic industry and in purification of refractory metals2 \ Boron trichloride gas is produced from raw materials such as boron oxide (B203), borides (TiB2 and CaB6), borates (Ca2B6On, NaCaB509 and Na2B407), borate ester (B(0CH3)3) and boron carbide (B4C). If the raw material includes oxide, the performance of BCI3 gas is affected in a negative way and remarkable amount of phosgene (COCl2) impurity which is very difficult

77

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

to remove occurs (depending on the reaction temperature). Even if B4C, the most favorable raw material (theoretically oxygen-free), is used for the production, little amounts of COCI2 are generated simultaneously because of unavoidable humidity in reaction medium or non-pure CI2 gas. However, it is essential to take some additional purification steps for BCI3 used in special applications4. In 1954, C. Marks produced BCI3 gas from B4C powder by using horizontal graphite reactor heated to 400 °C5. In 1962, R. C. Davis et al. obtained BC13 with 97 % efficiency from 44 μπι-2 mm particle sized B4C by continuous chlorination (0.03-0.9 m/s CI2) in fluidized-bed furnace, at 700-1200 °C6. W. P. Thompson produced BC13 in 1964, by external heating of Cl2 gas (5.3 g/min) and B4C in a borosilicate glass tube up to 790 °C, using RH actuator and so he achieved 98 % efficiency7. In 1973, the production of BCI3 was realized by G. Kratel and G. Vogt in a different way, by addition of borides of the first, second and third main groups into B4C chlorinated with 0.8 kg/h rate at 1000-1200 °C and they obtained 99 % production efficiency8. In 1980, G. Anmelder used chlorination catalysts such as N1G2.6H2O, C0CI2.6H2O, FeCl2.4H20 and CuCl2.H20 for the production of BC13 in a quartz reactor at 650-1000 °C. However he did not provide the complete conversion of B4C to BCI3 . A. C. Jones et al. produced BCI3 in 1997, by passing 3-8 kg/h Cl2 through 4-8 mm grained B4C placed in a quartz reactor heated up to 800 °C with RF. They described their method as an inconvenient way for industrial applications because they achieved maximum 80 % efficiency10. In this study, the production of BCI3 gas was realized in originally designed set-ups by continuous chlorination of native B4C powder in a vertical quartz reactor externally heated and the effects of reaction temperature, B4C amount and B4C heap height on the process efficiency were examined. EXPERIMENTAL The raw material was B4C produced from native boron resources in BM Boron Technologies in Kayseri, Turkey. The average particle size of B4C was determined as 4.647 μιη [d(0.1) = 0.887 μιη, d(0.5) = 3.129 μπι, d(0.9) = 10.861 μπι]. B4C and graphite phases of the powder were confirmed by XRD (respectively, JCPDS Card No: 35-0798, 41-1487). The weight percentages of the carbon, oxygen, sulfide and nitrogen impurities in B4C are represented in Table I: Carbon content of the powder is ideal according to the boron-carbon phase diagram but oxygen content is very high compared with other powders. The content of metallic impurities in the powder are given in Table II: Al, W, Si and Fe are higher in percentage. The amount of boron in B4C powder was analyzed as 67.87 % by weight. The other starting material, Cb gas in 99.99 % purity, was assured from Istanbul Water and Sewerage Administration. The purging gas, N2, was supplied from BOS in purity of 99.998 %. Molecular sieve (Applichem) which was used as nitrogen dryer has 0.3-0.4 nm particle size. Also, CaCl2 (Tekkim, purity > 98 %) was used as chlorine dryer. 80 % humidity absorbencies of molecular sieve and CaCb were respectively greater than 20 % and 25 %. The ethanol (Tekkim, 96 % purity) acted as cooling medium in quartz heat exchangers. The metallic impurity analysis of B4C powder was performed by Perkin Elmer 1100 B Atomic Absorption Spectrometer (AAS) and confirmed by CAMECA SX-100 Electron Probe Micro-Analyzer (EPMA). Non-metallic impurities were analyzed with ELTRA CS 800 Carbon Sulphur Determinator and ELTRA ONH 2000 Oxygen Nitrogen Hydrogen Determinator. The particle size distribution of B4C was identified by Malvern Mastersizer 2000 Particle Analyzer. The phase analysis of B4C powder and reaction products were performed by PHILIPS PW 3710

78

· Processing and Properties of Advanced Ceramics and Composites

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

Rigaku XRD with Cu Ka radiation in the 2Θ range of 0-80° at a rate of 5°/min. The grain morphologies of the reacted B4C and by-products were examined by JEOL JSM T330 scanning electron microscope (SEM). The production process of BC13 from native B4C was comprised of two different set-ups originally designed. The process in the first set-up was essentially the try-out process. It contained Kofloc 3440 mass flow controller (MFC) and MR-5000 screen system for Cl2 and N2 gases, Aalborg rotameter, mercury U-tube manometer, quartz reactor, silicon carbide resistance and programmable Protherm tube furnace, thermocouple, two pieces of quartz spiral heat exchangers, two pieces of Polyscience cryostats (up to -30 ve -40 °C), two pieces of quartz gas-wash-bottles, ultra-freezer (up to -90 °C), PTFE spiral cooler and Supelco Tedlar gas bags (5, 10 and 25 L). Also, the pipeline between the exit of the furnace and first spiral heat exchanger was covered with alumina sheaths for isolation considering that the highest temperatures prevail in this region. In order to decide the isolation, the temperature was measured with 10 minutes intervals at three different points, i.e. unmantled zone, top cover of quartz reactor and lead-in of pipeline. The maximal temperature didn't exceed 200 °C, 105 °C and 35 °C respectively. Therefore, the isolation was demounted in the second experimental set-up. Table I The Non-metallic Impurities in B4C Powder Elements

Weight %

C

19.9207 0.2214 6,8165 2.4285

s

0 N

Table II. The Metallic Impurities in B4C Powder Cr Elements Ni Mn Cu Co Fe Weight % 0.0032 0.014 0.24 0.0041 0.053 0.015 Elements Weight %

Zn 0.008

Pb 0

Ag 0.0099

Al 1.073

Elements Weight %

Ca 0.083

K 0.0047

Na 0.011

Si 0.45

Mg Cd 0.0063 0.0016 Ti 0.22

W 0.55

The process in the second set-up was more complicated than the first; it was also included N2 and Cl2 dryer columns, thermometers, two pieces of gas sampling chambers and rashing ring packaged gas scrubber. In both systems, furnace, distillation units and ultra-freezer were connected with special designed PTFE, borosilicate glass and quartz pipes and fittings to prevent gas leakage. In this manner, the second process was the improved version of the first process to obtain high efficiency. In order to remove the humidity or oxygen content in the systems, it is necessary to purge with inert N2 gas with low flow rates up to 100 ml/min throughout 5-12 hours. In the first system, after N2 purging, Cl2 gas was passed through MFC and also rotameter to observe the flow oscillations. The pressure change in the both system was controlled by mercury U-tube manometer with bypass flow of Cl2. The reaction was carried out

Processing and Properties of Advanced Ceramics and Composites

· 79

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

by passing the Cl2 gas through a heap of B4C powder stacked on a quartz frit placed in the point of 2/3 of the vertical quartz reactor (d,: 82 mm, d0: 110 mm) that provides adequate heating of Cb gas. The internal and external parts of the furnace are sealed by cylindrical blocks of alumina fiber, teflon gaskets and steel covers. The quartz reactor was externally heated by SiC resistance tube furnace up to 650-850 °C. Also, the inner temperature of the furnace was controlled by pyrometer with a protective alumina sheath. After the reaction, it is essential to separate the main product gas, BCI3, from the gas impurities such as free CI2, CCU and CO. In the first set-up, first and second step distillations were realized by quartz spiral heat exchangers with respectively oil (constant at 85 °C) and ethanol (constant at 14 °C) medium. It was considered that the temperature of the first column was held enough high to provide the conveyance of gas mixture to the second column without any loss. Moreover, the temperature of the second column was fixed at 14 °C for removal of CCU from the gas mixture, so the second column was charged with distillation. Quartz gas-wash-bottle cooled by the same water circulated at the second cryostat was placed at the end of the second distillation column for holding CCU liquid. The other gas products were carried to the PTFE spiral essential for cooling the gas mixture and than to the quartz gas-wash-bottle situated in the ultra-freezer. Also, there was a by-pass flow valve in the pipeline between gas-wash-bottle and ultra freezer for sampling and purging the system when the reaction was ended. BCI3, COCI2 (depending on the temperature) and unutilized CI2 were collected in the quartz gas-wash-bottle. For discharge of probable CO, a gas bag outside of the freezer was connected to this gas-wash-bottle. Process flow diagram of the first set-up is illustrated in Figure 1.

Mercury U-Tube Manometer

Chlorine

4

r

Mass Flow Controller — t \ (MFC)

1

Rota

Quartz Tube "Ί Heating Element Γ

♦

Nitrogen

Pyrometer with Dewar Flask

1. Quartz Spiral Heat ExchangetCoil)

Gas-Wash-Bottle Cooled by Water

2 Quartz Spiral Heat Exchanger (ethanol)

Ultra Freezer (- 90 °C) Gas Bag

l·*—

Quartz Gas-Wash-Bottle

4 —

PTFE Spiral

U-

1

Figure 1. Process flow diagram of the first set-up for BC13 production In the second set-up, the flow of purging gas, N2, applied before the chlorination reaction was controlled by rotameter and dried by passing through molecular sieve package column. The flow of reaction gas, Cl2, was regulated with MFC and dried in CaCb package column. After the chlorination reaction occurred as same in the first set-up, the product gases were carried to the first quartz heat exchanger with ethanol-water medium which is constant at -2 °C and to the thermometer chamber to control the temperature change. It was essential to take the sample of

80

· Processing and Properties of Advanced Ceramics and Composites

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

product gas with bypass flow. The second step cooling was realized by using quartz heat exchanger with ethanol-water medium which is constant at -10 °C. The cooled gases were conveyed to the quartz gas-wash-bottle filled with PTFE rashing rings and glass beads (φ 3 mm). The temperatures of the heat exchangers were held below 0 °C in order to separate BCI3 and COCI2 as liquid from the gas mixture. However, the temperature in the pipeline positioned between the exit of furnace and first heat exchanger was approximately sufficient to hold CC14 as liquid. The rashing rings and glass beads were present to expand the cooling surface. After distillation step, the samples of undistilled gases (free Cl2 and CO) were taken with bypass flow subsequent to temperature control and gases were conveyed to the gas bags. At the end of the reaction, the valves placed on the different zones of pipeline were opened and waste gases were carried to a rashing ring packaged gas scrubber with NaOH solution. Process flow diagram of the improved second set-up is demonstrated in Figure 2. Mercury U-Tube Manometer Mass Flow Controller (MFC) Molecular Sieve Package Column

Quartz Tube Furnace with SiC Heating Element

t

Pyrometer with Dewar Flask

Sampling Chamber

Nitrogen

Gas Bag

H

CaCl: Package Column

L-

Thermometer

1 Quartz Spiral Heat Exchanger (ethanol-water)

Sampling Chamber Quartz Gas-Wash Bottle

2. Quartz Spiral Heat Exchanger (ethanol-water)

Figure 2. Process flow diagram of the improved second set-up The gaseous samples were analyzed by Bruker Alpha-T Fourier Transform Infrared Spectrometer (FTIR) by using a glass gas cell with NaCl windows on each end. The gas cell was evacuated and charged with gas sample taken with bypass flow from two different points. In 1977, J. A. Merritt and L. C. Robertson analyzed the dissociation products by a spectrophotometer by using stainless steel cell with NaCl windows on each end, in their patent related to removal of COCb impurity from BC13 by laser radiation11. In 1979, H.R. Bachmann et al. used an IR absorption cell with NaCl windows for IR spectroscopy analysis of products containing BC13, COCl2 and CO, in the study of decomposition of COCl2 sensitized by BCI312. The parameters such as reaction temperature, amount of B4C and B4C heap height were chosen considering their effects on process efficiency. The amounts of reactants (B4C and CI2) were calculated stochiometrically according to the reactions (1) and (2). Previously, A.C. Jones et al. produced BCI3 at 800 °C with progress of reaction (2)10.

Processing and Properties of Advanced Ceramics and Composites

· 81

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

In the experiments 1 and 2, B4C powders were chlorinated at 800 °C and 750 °C respectively, with the constant Cl2 flow rate (430 ml/min) at the same reaction time (1.5 h). The heap heights of the powders were 1.8 mm and 18 mm. Thus, in the second experiment the amount of B4C (110 g) was ten times greater than the first. Also, the bulk density of B4C powder was approximately 1.16 g/cm3. B
(1)

BAClk)+eCl2{K)^4BCl3{g)+Cik)

(2)

RESULTS AND DISCUSSIONS First experimental set-up was not completely appropriate for production of BCI3 due to some deficiencies. It was observed that boric acid (H3BO3) formed in the pipelines according to the reaction (3) and it was confirmed by XRD analysis (JCPDS Card No: 73-2158). BCl3 + 3H20 -> H3B03 + 3HCI

(3)

After the reactions at first set-up, a bark-like yellow layer (grained 3-4 μπι) formed at the bottom zone of the reactor. SEM image and EDS analysis of this layer is shown in Figure 3. This impurity may be formed by heating of cylindrical blocks of alumina fiber which were used for keeping straight the protection tube of thermocouple in the reactor. However, this desublimated layer had no effect on the product gases.

Figure 3. (a) SEM image of the bark-like layer formed in the first set-up, (b) EDS analysis of the layer from region a, (c) Weight percentages of the elements in the layer.

82

· Processing and Properties of Advanced Ceramics and Composites

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

Due to the examined gas flow rates, the first set-up was totally insufficient in cooling and separating the gas mixture. The oil heating to avoid the condensation of CC14 proved false. The second spiral cooler was inadequate for liquefying of the same gas. So, the whole gas mixture went to the ultra freezer and was condensed at the end point. Also, the Cl2 flow rates were chosen too high (800-1000 ml/min) for less amounts of B 4 C, so that a high amount of free Cl2 were condensed in the gas-wash-bottle. In the second experimental set-up, the individual constituents in the gas mixture were successfully separated. In the experiment 1 at 800 °C, boron chlorination efficiency was found 63 % with a chlorine consumption rate of 75 %. Boron chlorination efficiency was calculated with the formula shown in equation (4). All boron amounts were determined by wet analysis of solid B 4 C. BCE[%]-.

Boron (input) [g\- Boron (remaining) [g] Boron (input) [g]

xlOO

(4)

XRD analysis of the powder before and after the experiment 1 is shown in Figure 4. The decrease in the concentration of B4C and the increase in the amount of graphite are obviously seen.

after reaction

jjiJLjUu

before reaction

UK

40 000

2Θ (degrees) Figure 4. XRD patterns of B 4 C powder before and after experiment 1, (*) B 4 C and (°) C. Assuming that the charge was 100 % B 4 C, the amount of Ch has been chosen as 114 g, according to the reaction (1). In regard of reaction (2) which is dominated at elevated

Processing and Properties of Advanced Ceramics and Composites

· 83

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

temperatures (> 500 °C), there was a chlorine excess of 29 g (i.e. 25 %). The impurities share approximately 15 % of the fed chlorine in form of MeCl, MeCl2, MeCl3, MeCl4 and MeCl6. The relatively low chlorination efficiency of boron is then explainable only due to the low heap height of the bed through which the unutilized chlorine gas flows. In the experiment 2 at 750 °C, the boron chlorination efficiency was 84 % with a chlorine consumption rate of 91 %. Only 10.7 g Cl? of given 114 g was unutilized by means of higher heap of the reaction bed and longer retention time of Ch. This improvement can be interpreted also in regard of the well-sloped gradient at the beginning of the reaction between less chlorine and much boron. FTIR spectrums of the gas samples produced in experiment 2 are shown in Figure 5. The product gas shows the same characteristic peaks with the reference one. In Figure 5 (b) and (d), the characteristic peak of BCI3 disappeared. This means that there is a successful gas capturing in the gas-wash-bottle with glass beads and PTFE rashing rings. It takes approximately 35 minutes that CI2 gas contacts with the B4C layer. So, the absorbance of the peak is very low at the 5th minute after the first contact. Also, at 1190 cm"1 in Figure 5, there is a peak that presents in all samples. This peak probably belongs to a gaseous compound which is much more volatile than BCI3. In Figure 5 (a)-(d), very weak peak of CCI4 is detectable at around 800 cm"1, although it seems not probable that CC14 slips away from the second heat exchanger and reaches the gas sampling chamber. This can be explained due to the relatively higher vapor pressure (400 mbar at -10 °C) of BCI3. However, the vapor pressure of CCI4 at the same temperature is only 26 mbar. It can be dragged by BCI3 in very small but detectable amounts. This case is in the favor of reaction (1) but it is unexpected according to reaction (2).

Figure 5. FTIR spectrum of the gas samples produced in experiment 2, (a) taken from 1. sampling chamber after 40 minutes, (b) from 2. sampling chamber after 50 minutes, (c) taken from 1. sampling chamber after 80 minutes, (d) from 2. sampling chamber after 90 minutes and (e) sample of gas mixture containing 95 % BCI3 and 5 % N2, (reference).

84

· Processing and Properties of Advanced Ceramics and Composites

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

The SEM images of the B4C powder before and after experiment 2 are illustrated in Figure 6. There are not very exact differences between (a) and (b). However, there is reduction in particle size and some white clusters disappear. EDS analyses of the powders from the pointed regions are represented in Figure 7. The EDS spectra indicating the powder before and after the chlorination reaction show the rising graphite peak. This case is contrary to the reaction (1), but in favor of reaction (2). So, the rising graphite peak in Figure 4 and 7, also the weak peak of CC14 in Figure 5 give us pause to think that both reactions (1) and (2) occur partially and simultaneously.

(a)

(b)

Figure 6. SEM images of B4C powder (750X), (a) before reaction (b) after reaction.

Processing and Properties of Advanced Ceramics and Composites

· 85

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

Figure 7. EDS analysis of B4C powder, (a) before reaction, from a region (b) after reaction, from b region. In the second experimental set-up, different type of impurities formed as orange-reddish dendrites at the cold bottom region, dark-colored precipitates at the top region and white precipitates at the quartz cover of the reactor. The SEM images of the impurities are illustrated in Figure 8. The EDS analysis from region a, b and c pointed at Figure 8 is shown in Figure 9. Although EDS determined all the elements in the impurities, XRD analyzes only the dominant phases in (a) as A1C13.6H20 (JCPDS Card No: 44-1473); in (b) and (c) as A1C13.6H20 and H3BO3 (JCPDS Card No: 75-1127).

86

· Processing and Properties of Advanced Ceramics and Composites

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

Figure 8. SEM images of the impurities, (a) orange-reddish dendrites at the cold bottom region (b) dark-colored precipitates at the top region (c) white precipitates at the quartz cover of the reactor.

Processing and Properties of Advanced Ceramics and Composites

· 87

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

The orange-reddish dendrites include W, Al and Si; dark-colored precipitates include W, Al, Fe and Si; white precipitates include Al, Si, Cr and Fe metallic impurities. Most of the

88

· Processing and Properties of Advanced Ceramics and Composites

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

metallic impurities in the powder form also volatile chlorides with relative high densities. They went through the quartz frit down to the bottom of the reactor. This can be called as a "self-cleaning" effect. CONCLUSION Production of BC13 from native B4C was tested in two different originally designed set-ups. First set-up was not appropriate for the efficient production of BCI3. Because the spiral exchangers were insufficient in length, temperature and contact surface; Cl2 flow rate was too high for chosen amount of B4C. So, high amount of free Cl2 went to the ultra freezer with produced BCI3 and a yellowish-green liquid was obtained as a mixture of two gases. The second set-up was enough to produce BCI3 with boron chlorination efficiency of 84 %. The second set-up was superior in capturing of BC13. However, CCU in very small amounts dragged with BCI3. Rising of free graphite peak and dragging of CCU proves that both reactions, (1) and (2), were occurred simultaneously. For production of BCI3, the B4C powder does not need to be extremely pure. Most of the metallic impurities in the powder form volatile chlorides but they went down to the bottom of the reactor. Thus, there is a self-cleaning effect. Consequently, the experiments prove that the heap height of the bed elongates the solidgas contact time and provides effective chlorine utilization. These facts have to be considered for deciding to work in fix-bed or fluidized-bed reactor. ACKNOWLEDGMENTS This work has been supported by The Scientific and Technological Research Council of Turkey within with the project number of 106M087. REFERENCES 'Boron Report, The National Boron Research Institute of Turkey, BOREN, Ankara (2004). F. Habashi, Handbook of Extractive Metallurgy, Vol IV, WILEY-VCH, 1986-2021 (1997). 3 K. Othmer, Encyclopedia of Chemical Technology, Vol IV, Wiley-Interscience Publication, John Wiley&Sons, 129-132 (1978). 4 R.C. Hyer, S.M. Freund, A. Hartford and J.H. Atencio, Selective Removal of Phosgene Impurity from Boron Trichloride by Photochemical Dissociation, Journal of Applied Physics, Vol 52, 6944-6948 (1981). 5 C. Marks, Process for the Manufacture of Boron Nitride, United Kingdom Patent, No: 711254 dated 30.6.1954. 6 R.C. Davis, J.N. Haimsohn and J.T. Bashour, Manufacture of Boron Trichloride, united States Patent, No: 3025138 dated 13.3.1962. 7 W.P. Thompson, Improvements in and Relating to Mineral Active Carbons and to a Process for Their Preparation, united Kingdom Patent, No: 971943 dated 7.1.1964. 8 G. Kratel and G. Vogt, Process for Manufacturing Boron Halides, United States Patent, No: 3743698 dated 3.7.1973. 9 G. Anmelder, Verfahren zur Kontinuierlichen Darstellung von Borhalogeniden, German Patent, No: 2826747 dated 3.1.1980. 10 A.C. Jones, G. Williams and A.B. Leese, Process for Producing Boron Trichloride, United Kingdom Patent, No: 2304104 dated 12.3.1997. 2

Processing and Properties of Advanced Ceramics and Composites

· 89

Process Design and Production of Boron Trichloride from Native Boron Carbide in Lab-Scale

11 J.A. Merritt and L.C. Robertson, Removal of Phosgene Impurity from Boron Trichloride by Laser Radiation, United States Patent, No: 4063896 dated 20.12.1977. 12 H.R. Bachmann, H. Noth and R. Rinck, Infrared Laser-Specific Reactions Involving Boron Compounds III: Decomposition of Phosgene Sensitized by Boron Trichloride, Journal of Photochemistry, 10, 433-437 (1979).

* Corresponding author. Tel.: +90 2122856893; Fax: +90 2122853357 E-mail address: [email protected] (Duygu Agaogullari).

90

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Sintering and Hot Pressing

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

SPARK PLASMA SINTERED ALUMINA-ZIRCONIA NANO-COMPOSITES BY ADDITION OF HYDROXYAPATITE

'college of Science and Technology, Nihon University, 7-24-1, Narashinodai, Funabashi. Chiba, 274-850 1, Japan 2 ~ o l l e gof e Science and Technology. Nihon University, 1-8-14, Kandasurugadai, Chiyoda, Tokyo, 101-8308, Japan 3 ~ o l l e gof e Materials Science and Technology, Xi'an University ofTechnology, 5 South Jinhua Road, Xi'an Shaanxi, 710048, China ABSTRACT Zirconia-alumina composites with addition of different volume fraction of hydroxyapatite (HA) were fabricated successfully using spark plasma sintering (SPS). The densification behavior, microstructure and mechanical properties of composites are investigated as a function of sintering temperature and HA content respectively. The sintering temperature has a significant effect on the final densities achieved in the ZrOz-Al2OdHA compacts. The addition of HA has a barrier effect on diffusion between grains of Z r 0 2 and A1203 and thus limit the grain growth of ZrOz and A1203. Sintering the Zr02-AI2O3/HAcomposites at 1400 OC led to the decomposition of HA in the samples. Flexural strength, fracture toughness and Vickers hardness values increase with increasing sintering temperature, and show decreasing trend with increasing of HA content at the same temperature. They compared well with densities obtained at different sintering temperature. The maximum flexural strength, fracture toughness and Vickers hardness of 967.1MPa, 5.27 M ~ a . m l "and 13.26 GPa were achieved for ZrO?-Al?O, composite respectively. Flexural strength. fracture toughness and Vickers hardness values of the Zr02-A1203/HA composite fell within the value range of dense HA and of Zr02-AI20i composite.

Hydroxyapatite (HA, Calo(P04)6(OH)z)is a kind of calcium phosphate bioceramic materials. Owing to its chemical and structural similarity with natural bone mineral, it has a unique capability of binding to the natural bone through biochemical bonding, which promotes the interaction between host bone and graft material' I . :-I. Densified HA has attracted extensive attention for its excellent biocompatibility, and has been widely used for bone substitute and grafting; but the low mechanical properties restricted its application for load-bearing implants. The fracture toughness of the monolithic HA ceramics . with 2-12 am"' for human bone':. -1:. Investigations does not exceed 2 M ~ a m " ~coinpared aimed at broadening the medical application potential of implant materials based on HA are carried out in scientific research institutes around the world. The investigations mainly focus on the two aspects: one approach is to use HA as a coating on a strong metallic or ceramic substrate such as zirconia or titanium alloy. One of the limitations of this approach is susceptibility of the HA coating to de-bond from the substrate"!. Another attractive approach is to maintain the biocompatibility of HA and improve its mechanical properties by introducing thc concept of fabricating composites either with HA as substrate[(.. -!or with HA as additive[:. "1.

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

Yttria stabilized zirconia (Zr02(Y203)), alumina (A1203) and their composites have been considered as substrate or reinforcement phases for use in implants, such as prostheses and dental materials due to their excellent biocompatibility as well as their desirable material properties, such as strength, chemical stability, and wear resistance . The aluminatoughened zirconia (ATZ) Bio-Hip*', developed by Metoxit AG (Thayngen, Switzerland), has a bend strength of up to 2000MPa , indicating that it can withstand loads that are nearly twenty times greater than densified HA. But as it is well known, Zr0 2 , ΑΙ2Ο3, and their composites are classified as bioinert materials. Therefore, in order to increase the biocompatibility, calcium phosphate such as HA are considered to be the most suitable bioactive material to be used as an additive. Shen et al. reported that the bend strength and the fracture toughness of HA-50 vol% Zr02 composite was 440 MPa and 2.5 MPa.m1/2, respectively . Miao et al. reported that the HA-40 wt.% Zr02(Y203) composites sintered at 1200 °C showed 200 MPa bend strength However, few detailed reports on the microstructure change and mechanical properties of HA added ΖΓ02(Υ20 3 )-Α1 2 03 composites have been published. Spark plasma sintering (SPS) processing of monolith and composite ceramics materials has been recognized to offer a number of advantages over conventional sintering approaches. Increase of densification rates at relatively low sintering temperatures, makes the SPS fabricating technique significantly faster than the conventional process, which means relatively shorter sintering time and less energy consumption. It was concluded by some, that SPS has the potential for enhanced densification and suppressed grain growth due to a fast heating rate and apparent low firing temperature. Based on the advantages mentioned previously, the application of SPS processing in this study may solve the serious problem of extensive reaction between HA and Zr0 2 to form tricalcium phosphate (TCP), and avoid the fully stabilized Zr0 2 . The forming of TCP would lead to the serious reduction in the biocompatibility of HA, and fully stabilized ZrC>2 would cause the decreasing of strength and toughness, which mainly depends on the phase transformation from the tetragonal phase to the monoclinic phase The aim of this study was to fabricate Zr02(Y203)-Al203 composites with addition of different volume fraction of HA at various sintering temperatures using spark plasma sintering. The mechanical properties and densification behavior of composites are investigated as a function of HA content and correlated with the compositional and structural variations. And this will be the foundation of the final aim; to fabricate the functionally graded materials (FGM) based on ATZ composites. EXPERIMENTAL PROCEDURES Preparation and Characterization of Powders The starting materials used in this study were two kinds of powders. One was a commercially available nano-composite powder, 3-mol% yttria-stabilized zirconia (3-YSZ) reinforced with 20 wt% alumina (Tosoh Co., Japan, average particle size 20μηι, specific surface area 13 m2/g and crystallite size 28nm) was used in the as-received state, which is denoted as TZP-3Y20A in this study. Another commercially available hydroxyapatite powder (HA, Taihei Chemical, Japan) was used for this work. Appropriate quantities of TZP-3Y20A and HA powders were slurry mixed in acetone, and ball milled for 24 h using zirconia balls to obtain a homogenous mixture with a composition ranging from 10-50% volume fraction of HA. Acetone was then evaporated and the powder was dried, crushed and sieved through a 500 mesh sieve (25

94

·

Processing and Properties of Advanced Ceramics and Composites

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

μιη granules). To determine the phase composition and purity, X-ray diffraction (XRD) was performed on the starting powders as well as the powder samples obtained by grinding the sintered samples. The powders were mounted on a borosilicate glass slide and scanned on a RINT 2000 X-ray diffractometer (Rigaku, Japan) with a Cu-Ka radiation source at 30 mA and 40 kV with a scan speed of 0.5°/min and steps of 0.02°. The morphology of the powder precursor was also examined both on a SSX-550 scanning electron microscope (SEM; Shimadzu, Japan) equipped with an energy dispersive X-ray spectrometer (EDX) attachment, and a transmission electron microscope (TEM; JEOL-JEM3010, Japan). Sample Preparation For sintering, the powders were loaded into a cylindrical graphite die and uniaxially pressed into green compacts with dimensions of 56 mm x 11 mm x 2 mm by using a hydraulic pressure of 10 MPa. Prior to this step, the interior surface of the die was sprayed with a layer of boron nitride to lubricate and prevent diffusion between the graphite and compact at high temperatures. The green compacts were sintered in an SPS-3.2MK-IV system (Sumitomo Coal Mining, Japan). The sintering was performed in the temperature range of 1000-1400°C with steps of 100°C at a heating rate of 200°C/min. The temperature was measured by means of an optical infrared thermometer focused on to the graphite die surface. The sintering pressure was set at 44.6 MPa. The vacuum level of the chamber was kept below 10 Pa during sintering. After heating at the desired temperature for 8 min, the power was turned off, and the sample was cooled in the chamber to less than 300°C at a cooling rate of 100°C/min. Characterization of Sintered Samples and Mechanical Testing The density of the sintered samples was determined by using Archimedes' method with distilled water. The composites relative density was determined using a theoretical density of 3.94 g/cm3 for alumina, of 6.00 g/cm3 for 3-YSZ and 5.50 g/cm3 for TZP-3Y20A (density supplied by the manufacture). Five measurements were conduct to obtain an average value. The crystal phases in the sample were identified by using XRD referenced to the standard ICDD PDF cards available in the system software. In addition, the microstructure evolution of the dense samples, including the fracture surface, with respect to various sintering temperatures was examined by using SEM. The surfaces prepared for examination in the SEM were coated with gold using a HITACHI E-1030 sputter coater operated at 15 mA for 30 s. The grain size was estimated from the full width at half maximum (FWHM) by the Scherrer equation|l6| and confirmed by TEM micrographs and by the line intercept method with SEM micrographs. Three-point bending tests were conducted on an Instron-5500R tensile tester (Instron Corp., Canton, MA, USA) with a cross-head speed of 0.5 mm/min with an inner span of 30 mm. Test samples with dimensions of 56 mm x 11 mm x 2 mm were ground and polished for the three-point bending test. Three samples sintered under the same conditions were prepared for the bending test in order to obtain an average value. The flexural strength, σ, was calculated using _ 3PL a

~2bd\

(1)

Processing and Properties of Advanced Ceramics and Composites

· 95

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

where P is the applied load, and L, b, and d are the span length, the width, and the thickness of the specimen, respectively. The Young's modulus E was calculated using E=

L3 (/>-/>) 4&/3 " ( & - $ ) ^

(2)

where P, (/ = 1 and 2) is the applied load, and Si (/ = 1 and 2) is the displacement, and L, £, and d are the same as equation (1). Fracture toughness measurements were conducted by using an indentation fracture (IF) method on a Vickers microhardness tester (Shimadzu HSV-30). Prior to indentation, the crosssectional surface of the samples was polished to a 3 μιτι surface finish. In order to evaluate the fracture toughness of the sintered samples, the median lengths of the radial cracks at the four corners of the indentation trace were measured using a Vickers microindenter at a load of 98 N applied for 20 s. Average hardness and radial crack length values were determined from 10 indentations made for each sample. Fracture toughness, Kic, was calculated based on the median crack equation: / O =0.016·

E

1 c3,2 L

J

,

where C is the radial crack length, and Hv is the Vickers hardness.

(3 )

RESULTS AND DISCUSSION Characteristics of powders The TEM image shown in Figure 1 (a) is the nano-powder of HA, it can be seen from the image that the HA particles show a needle-like morphology, the particle size of the short axis is 20 nm and shows a narrow distribution. Figure 1 (b) shows the micrograph of TZP-3Y20A/HA10vol% after ball milling for 24h. Most of the particles have a uniform size and the TZP-3Y20A particles are smaller than 100 nm. It can be observed that agglomeration and close bonding formed between the particles. The XRD pattern shows that the TZP-3Y20A composite powder consists of a mixture of monoclinic and tetragonal phases of Zr0 2 and α-Α1203 powders, as shown in Figure 2. Average crystal size for Zr0 2 was calculated to be 24.5 nm, and that for αAI2O3 was 86.4 nm using the Warren-Averbach method after correcting for instrument broadening.

96

· Processing and Properties of Advanced Ceramics and Composites

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

Figure 1. TEM micrograph of the (a) HA powder and (b) TZP-3Y20A/HA-10 vol.% powder. ΤΖΡ-3Υ20Λ/ ΗΛ powder t: tetragonal ZrO; m: monoclinlc ZrO ; α: α-ΑΙ : 0,; Η: hydroxyapatite

2-Theta [degree]

Figure 2. XRD of TZP-3Y20A/HA-10 vol.% powder and TZP-3Y20A powders. Microstructure and densification behavior

Figure 3. Fractured surface images of (a) TZP-3Y20A composites sintered at 1400 °C and (b) pure HA specimen sintered at 1000°C.

Processing and Properties of Advanced Ceramics and Composites

· 97

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

Figure 3 (a) shows the fracture surface of TZP-20A nano-composite without HA sintered at 1400 °C. Compact microstructure and small grain size (about 296 nm) are well illustrated in this micrograph. The fracture proceeded mostly in an intergranular pattern and the fracture surface was very rough. Figure 3 (b) shows the fracture morphology of pure HA specimen sintered at 1000 °C, which was much different from that of TZP-3Y20A, with compact structure and the fracture mostly occurring in a transgranular pattern.

Figure 4. SEM micrographs of fractured surface of TZP-3Y20A/HA composites containing various volume fraction of HA after sintering at 1400 °C; (a) HA-10 vol.%; (b) HA-20 vol.%; (c) HA-30 vol.%; (d) HA-40 vol.%; (e) HA-50 vol.%. Figure 4 (a)-(e) show the microstructures of the TZP-3Y20A nano-composites containing different amounts of HA sintered at 1400°C. When 10 vol.% HA was added, the microstructure changed and exhibited a different morphology in comparison with samples without HA, as shown in Figure 5 (a). The large dark grains (about 300 nm) of HA appear in the microstructure, incompact morphology can be observed, and the grain size (125 nm) is smaller

98

· Processing and Properties of Advanced Ceramics and Composites

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

than those without HA in Figure 3 (a). When 20 and 30 vol.% HA were added (Figure 5 (c)-(d)), nearly no changes occurred for the grain size of TZP-3Y20A. HA grains were contact with each other in the local area. As HA content was increased up to 40 and 50 vol.%, the amount of contact between HA grains increased and agglomeration occurred through the continual pores fabricated by ZrO: and AI2O3 grains. The fine grains of Zr02 and AI2O3 embedded in HA grains can be observed, as shown in Figure 4 (e). It was found that a continuously porous skeleton was fabricated by fine grains of ZrC>2 and AI2O3, while HA acts as material filling in the pores, this structure is beneficial to homogenous distribution of bioactive HA. This framework structure is considered as an important factors in improving the biocompatibility of ceramics for implants because it is closely related to cell attachment, growth behavior, and bone strength between the tissue and artificial implant in the human body[17]. The grain size and porosity keep steady, by contrast to distinct densification behavior of TZP-3Y20A composites as shown in Figure 3 (a). This means that the addition of HA has a barrier effect on diffusion between grains of ZrC>2 and AI2O3 and thus limit the grain growth of ZrC>2 and AI2O3.

Figure 5. Density of TZP-3 Y20A/HA composites containing various volume fraction of HA as a function of sintering temperature. The densities of TZP-3Y20A/HA composites containing various volume fraction of HA as a function of sintering temperature are shown in Figure 5. For TZP-3Y20A composites, the density increased steadily with temperature and a maximum of 97.8% of the theoretical value (5.5 g/cm3) is obtained after sintering at 1400°C. When different volume amounts of HA were added, the density of TZP-3 Y20A/HA composites as a function of sintering temperature showed similar variation in trend compared to the TZP-3Y20A composites. The densities increased with increasing temperatures, but showed a decreasing trend when the sintering temperature was higher than 1300°C. It was also found that the densities decreased with increasing the volume amounts of HA. This is due to the density of HA (3.16 g/mm3) is lower than the density of TZP3Y20A (5.5 g/mm); the density of the sample with high HA content is lower than those with low HA content. The theoretical densities (T.D.) for different TZP-3Y20A/HA composites shown in Figure 6 are calculated by the rule of mixtures. Phase stability

Processing and Properties of Advanced Ceramics and Composites

· 99

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

Figure 6 shows the X-ray diffraction patterns of the TZP-3Y20A/HA-50 vol.% composites sintered at the temperature range of 1000-1400°C. There are reports in the literature on the promotive effect of ZrC>2 on the thermal decomposition of hydroxyapatite, even at lower temperatures . In the presence of zirconia even more OH" ions are lost due to the reaction between oxyhydroxyapatite and ZrC>2 according to Eq. (2) Ca»{PO<)6{OH)n :..a + ZrOr -> 3[Ca>(PO^] + CaO(ZrO) + (1 -x)H.O (2) The decomposition of oxyhydroxyapatite in the presence of ZrC>2 is reported to take place at a surprisingly low temperature (~950°C), taking into account that the oxyhydroxyapatite formed according to Eq. (1) is stable up to around 1400°C in air. That is to say that a decomposition temperature of 950°C is well below that required for densification of Zr02/HA composites by conventional processes such as pressureless sintering and hot pressing. By using the SPS technique, however, the deleterious reactions describe above could be avoided . The XRD patterns illustrate that sintering the composites at 1400°C led to the appearance of TCP in the samples. The results explain the reason why the densities decrease when the sintering temperature is increased to 1400°C, as shown in the Figure 5. This is because the density of TCP is 3.00 g/cm3 which is lower than that of HA . All these results could be attributed to the rapid sintering speed of SPS. The high heating rate and very short dwelling time prevented the reaction between ZrC>2 and HA. TZP-3Y20A/HA-50 vol.% composites

A

A A A 1

i i ' i 40

1

i

ιιΛ i

i ■ i ' 45

^w50

55

2-Theta [degree]

Figure 6. X-ray diffraction patterns of the TZP-3Y20A/HA-50 vol.% composites sintered at different temperatures; (a) TZP-3Y20A/HA powder, (b) 1000°C, (c) 1100°C, (d) 1200°C, (e) 1300°C, (0 1400°C, T: tricalcium phosphate (TCP; Ca3(P04)2). Mechanical properties Figure 7 shows the variations of the flexural strength of the TZP-3 Y20A/HA composites containing different volume fractions of HA with sintering temperature. The flexural strength increases with increasing sintering temperature and is strongly affected by the amount of HA content. In samples without HA, the flexural strength of 186.95 MPa was obtained at a temperature of 1000°C. the flexural strength increase continuously with increasing temperature. The maximum flexural strength of 967.1 MPa was attained at 1400°C, corresponding to a relative density of 97.8%, which is close to full density. The flexural strength is higher compared to those

100

· Processing and Properties of Advanced Ceramics and Composites

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

reported for similar compositions processed differently; 450 MPa by isostatically pressed and sintered and a little lower than 1.1 GPa by SPS[ When 10 vol.% of HA was added to the TZP-3Y20A powder and sintered under the same conditions, the flexural strength decreased compared to TZP-3Y20A without HA additions. The curve also shows that with increasing temperature in the range of 1000-1300°C, the flexural strength increased from 144.4 to 374.8 MPa, then decrease to 316.4 MPa when the sintering temperature increased to 1400°C. As the HA content was increased up to 20-50 vol. % , the flexural strength showed a similar trend as the samples of 10 vol. % HA content, and the flexural strength decreased steadily with increasing HA content at the same temperature. In other words, the flexural strength increased steadily with decreasing HA content. For example, the flexural strength increased from 117.5 to 640.0 MPa with the HA content decreasing from 50 to 0 vol. % at 1300°C. Another distinct feature in Figure 7 is the flexural strength decreased with the temperature increase to 1400°C for different HA content samples. This phenomenon could be due to decomposition of HA at 1400 °C as shown in Figure 6.

Figure 7. Influence of sintering temperature on flexural strength of TZP-3Y20A/HA composites containing different volume fraction of HA. The maximum flexural strength for TZP-3Y20A/50 vol.% HA is 117.5 MPa sintered at 1300°C, which is lower than the maximum flexural strength for pure HA of 131.5 MPa sintered at 950°C studied previously . When the HA content decreased to 40 vol. %, the maximum flexural strength increased to 211.6 MPa, which is higher than the strength of pure HA. Based on the results obtained, the flexural strength is increased with increasing TZP-3Y20A content, on the other hand, bioactivity is achieved by addition of HA at the expense of the flexural strength, if the TZP-3Y20A composite is considered as the substrate.

Processing and Properties of Advanced Ceramics and Composites

101

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

Figure 8. Vickers indentation observed in samples (a) TZP-3Y20A (b) TZP-3Y20A/50% HA for fracture toughness evaluation. Hardness and fracture toughness were measured in order to determine the influence of the sintering temperature and the content of HA phase on the mechanical characteristics of composite materials. Hardness and particularly fracture toughness belong to the most important parameters used in the characterization of ceramic materials, for which toughness is of primary important. The typical Vickers indentation with cracking at the tips for fracture toughness measurement is shown in Figure 8. To measure the fracture toughness accurately, the load of the hardness testing was increased from 9.8 N to 98 N in order to observe distinct crack generated at the tip of the indentation. Plastic deformation was found at the indentation edges of Figure 8 (a), which indicates that the as-sintered TZP-3Y20A composites have better fracture toughness. The indentation of Figure 8 (b) shows a different morphology for TZP-3Y20A/50% HA composite. Besides the cracks at the tips which can be clearly observed, cracks looking like annual rings also can be observed, at the inner area of the indentation. The result shows that the addition of HA deteriorated the fracture toughness of TZP-3Y20A composite.

Figure 9. Influence of sintering temperature on fracture toughness of TZP-3Y20A/HA composites containing different volume fraction of HA.

102

· Processing and Properties of Advanced Ceramics and Composites

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

The effect of sintering temperature on fracture toughness of TZP-3Y20A/HA composites was shown in Figure 9. The fracture toughness increased from 2.12 MPa.m1/2 to 5.27 MPa.m172 with increasing temperature from 1000-1400°C for TZP-3Y20A composite without HA addition. The fracture toughness improvement is partially attributed to the increase in density of TZP3Y20A composite as shown in Figure 5. It turns out that the higher the density, the lower the probability of microcracks forming during cooling, due to lack of stress concentrators. This in turn reduces the contribution of crack deflection toughening mechanism. When 10 vol.% of HA was added to the TZP-3Y20A powder and sintered under the same conditions, the fracture toughness showed an obvious decrease compared with TZP-3Y20A composite at the same sintering temperature. It was observed that the fracture toughness increased with increasing temperature, from 1.62 to 3.74 MPa.m12 in the 1000-1300°C temperature range, and then slightly decreased to 2.58 MPa.m172 when the temperature increased to 1400°C. As HA content was increased from 20 to 40 vol. %, the fracture toughness showed similar trend as the samples of TZP-3Y20A/10% HA, and the fracture toughness decreased steadily with increasing HA content at the same temperature. In other words, the fracture toughness increased steadily with decreasing of HA content. Interestingly, as HA content was increased to 50%, the fracture toughness increased instead of decreasing. It was found that the fracture toughness increased with increasing temperature, from 2.24 to 2.83 MPa.m172 in the 1000-1200°C temperature range, reaching the maximum fracture toughness of 2.83 MPa.m172, and then decreased with increasing temperature. The improvement in fracture toughness in TZP-3Y20A/50% HA samples could be attributed to the decrease in porosity. When the HA content increased from 40 to 50%, more HA filled in the pores of the continuously porous skeleton fabricated by fine grains of ZrC>2 and AI2O3, as shown in Figure 5. For dense HA ceramics, the fracture toughness fell within the range of 0.79-1.40 MPa.m172 as reported in the literature [4]. For TZP-3Y20A composite studied in the present paper, the fracture toughness fell within the range of 2.12-5.27 MPa.m172. Comparing to dense HA, the improvement in fracture toughness of TZP-3Y20A/HA composites is beneficial. This is due to the excellent fracture toughness of TZP-3 Y20A.

Figure 10. Influence of sintering temperature on Vickers hardness of TZP-3 Y20A/HA composites containing different volume fraction of HA.

Processing and Properties of Advanced Ceramics and Composites

·

103

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

Results of Vickers microhardness studies are given in Figure 10. It can be seen that the Vickers hardness of TZP-3Y20A increased with increasing temperature, from 4.96 to 13.26 GPa in the temperature range of 1000-1300°C. The different amount of porosity of in samples can be an important factor leading to different Vickers microhardness value. As sintering temperature increasing, pores in the sintered compact are remarkably eliminated and the homogeneous microstructure can be acquired (see Figure 3 (a)). Hardness slightly decreased at 1400°. This is probably caused by the influence of grain size on the hardness of the sintered composite. In other words, coarser grain size lead to lower hardness [23]. When different volume fraction of HA from 10 to 50% were added to TZP-3Y20A, Vickers hardness decreased with increasing of HA content at the same temperature. Take samples sintered at 1300 °C as an example, hardness shows a decreasing trend from 10.57 to 6.17 GPa, when HA content was increased from 10 to 50%. Considering hardness of dense HA ceramics fell within the range of 5.7-6.6 GPa [24], for ZrO2(3Y)-20wt.% A1203 composites studied in the present work, the hardness shows a maximum value of 13.26 GPa sintered at 1300 °C. Vickers hardness of TZP-3Y20A/HA composite fell within the hardness value range of dense HA and of TZP-3Y20A composite, which means the merits of compatibility of HA and hard Zr02(3Y)-Al203 were intermediated by fabrication of the TZP-3Y20A/HA composite. CONCLUSIONS The present work shows that TZP-3Y20A/HA composites with the addition of different volume fraction of HA were fabricated successfully using spark plasma sintering. The densification behavior, microstructure and mechanical properties of composites are investigated as a function of sintering temperature and HA content. The sintering temperature has a significant effect on the final densities achieved in the TZP-3Y20A/HA compacts. The density of TZP-3Y20A composite increased steadily with temperature and a maximum value of 97.8% was obtained after sintering at 1400 °C. The addition of HA had a barrier effect on diffusion between grains of Zr0 2 and A1203 and thus limited the grain growth of Zr0 2 and AI2O3. Sintering the TZP-3Υ20Α/ΗΑ composites at 1400 °C led to the appearance of TCP in the samples. The TCP is from the decomposition of HA due to the limited thermal stability at high temperature, and not from the reaction between Zr0 2 and HA. Flexural strength, fracture toughness and Vickers hardness values increased with increasing sintering temperature and were strongly affected by the amount of HA, which compared well with densities obtained at different sintering temperatures. Flexural strength, fracture toughness and microhardness showed a decreasing trend with increasing of HA content at the same temperature. This indicates that bioactivity is achieved by addition of HA at the expense of the mechanical properties of TZP-3Y20A composite. The maximum flexural strength, fracture toughness and Vickers hardness of 967.1MPa, 5.27 MPa.m1,2 and 13.26 GPa were achieved respectively, at a relative density of 97.8%. Flexural strength, fracture toughness and Vickers hardness values of the TZP-3Y20A/HA composite fell within the value range of dense HA and of TZP-3 Y20A composite, indicates the advantages of bioactivity of HA and hard Zr02(3Y)-Al203 were obtained by fabrication of the ZP-3Y20A/HA composite.

104

· Processing and Properties of Advanced Ceramics and Composites

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

ACKNOWLEDGEMENTS The authors sincerely thank Prof. Zhengxin Lu for performance of the TEM observations and helpful discussions. REFERENCES fll R. Murugan and S. Ramakrishna, Development of nanocomposites for bone grafting, Compos. Sci. Tech., Vol 65, 2005, p 2385-2406 [21 E. S. Thian, N. H. Loh. K. A. Khor, S. B. Tor, Effects of debinding parameters on powder injection molded Ti-6A1-4V/HA composition parts, Adv. Powd. Tech., Vol 12(3), 2001, p 361370 [3] W. Suchanek, M. Yashima, M. Kakihana and M. Yoshimaru, Hydroxyapatite ceramics with selected sintering additives, Biomaterials, Vol 18, 1997, p 923-933 f4] S.F. Li, H. Izui and M. Okano, Densification, microstructure and behavior of hydroxyapatite ceramics sintered by using spark plasma sintering, J. Mater. Eng. Tech, Vol 130(031012), 2008, DOI: 10.1115/1.2931153 [5] D.E. MacDonald, F. Betts, M. Stranick, S. Doty and A.L. Boskey, Physicochemical study of plasma-sprayed hydroxyapatite-coated implants in humans, J. Biomater. Res., Vol 54, 2001, p 480-490 [61 A. Rapacz-Kmita, A. Slosarczyk and Z. Paszkiewicz, Mechanical properties of Hap-ZrCh composites, J. Euro. Ceram. Soc, Vol 26, 2006, p 1481-1488 [71 X.G. Miao, Y.M. Chen, H.B. Guo and K.A. Khor, Spark plasma sintered hydroxyapatiteyttria stabilized zirconia composites, Ceram. Inter., Vol 30, 2004, p 1793-1796 [81 Y.M. Kong, CJ. Bae, S.H. Lee, H.W. Kim and H.E. Kim, Improvement in biocompatibility of Zr02-A1203 nano-composite by addition of HA, Biomaterials, Vol 26, 2005, p 509-517 [9] W. Li and L. Gao, Fabrication of HA-Zr02 (3Y) nano-composite by SPS, Biomaterials, Vol 24, 2003, p 937-940 [101 B.T. Lee, C.W. Lee, M.H. Youn, H.Y. Song, Relationship between microstructure and mechanical properties of fibrous Hap-(t-Zr02)/Al203-(m-Zr02) composites, Mater. Sci. Eng. A, Vol 58, 2007, p i 1-16 [11] Jerome Chevalier, What future for zirconia as a biomaterial. Biomaterials, Vol 27, 2006, p 535-543

Processing and Properties of Advanced Ceramics and Composites

·

105

Spark Plasma Sintered Alumina-Zirconia Nano-Composites by Addition of Hydroxyapatite

[12] Z. Shen, E. Dolfasson, M. Nygren, L. Gao, H. Kawaoka and K. Niihara, Dense hydroxyapatite-zirconia ceramic composites with high strength for biological applications, Adv. Mater., Vol 13, 2001, p 214-216 [13] X.G. Miao, Y.M. Chen, H.B. Guo and K.A. Khor, Spark plasma sintered hydroxyapatiteyttria stabilized zirconia composites, Ceram. Inter., Vol 30, 2004, p 1793-1796 [14] Yoshimura M, Phase stability of zirconia. Am. Ceram. Soc. Bull., Vol 67, 1988, p 19501955 [15] Tosoh, Supplier's powder specifications, Japan, 1998. [16] C. Oprea, V. Ciupina and G. Prodan, Investigation of nanocrystals using TEM micrographs and electron diffraction technique, Rom. J. Phys., Vol 53, 2008, p 223-230 [17] B. T. Lee, 1. C. Kang, S. H. Cho and H. Y. Song, Fabrication of a continuously oriented porous Al 2 0 3 body and its in vitro study, J. Am. Ceram. Soc, Vol 88, 2005, p 2262-2266 [18] V.V. Silva, F.S. Lameiras and R.Z. Dominguez, Microstructural and mechanical study of zirconia-hydroxyapatite (ZH) composite ceramics for biomedical applications, Compo. Sci. Tech., Vol 61, 2001, p 301-310 [19] M. Nygren and ZJ Shen, On the preparation of bio-, nano- and structural ceramics and composites by spark plasma sintering. Solid State Sci., Vol 5, 2003, p 125-131 [20] P.E. Wang and T.K. Chaki, Sintering behavior and mechanical properties of hydroxyapatite and dicalcium phosphate. J. Mater. Sci.: Materials Medicine, Vol 4, 1993, p 150-8 [21] R. Chaim, Pressureless sintered ATZ and ZTA ceramic composites, J. Mater. Sci., Vol 27, 1992, p 5597-5602 [22] J.S. Hong, L. Gao, S.D.D.L. Torre and Hiroki Miyamoto, Key Miyamoto, Spark plasma sintering and mechanical properties of Zr0 2 (Υ 2 θ3)-Α1 2 0 3 composites, Materials Letters, Vol 43, 2000, p 27-31 [23] R.S. Mishra, C.E. Lesher. A.K. Mukherjee, High-Pressure Sintering of Nanocrystalline γΑ1 2 0 3 , J. Am. Ceram. Soc, Vol 79 (11), 1996, p 2989-2992 [24] A. Rapacz-Kmita, A. Slosarczyk and Z. Paszkiewicz, Mechanical properties of Hap-Zr0 2 composites, J. Euro. Ceram. Soc, Vol 26, 2006, pl481-1488

106

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

COMPARISON OF SLIP CAST TO HOT PRESSED BORON CARBIDE T. Sano and E. S.C. Chin U.S. Army Research Laboratory Aberdeen Proving Ground, Maryland, USA B. Paliwal Department of Mechanical Engineering, Johns Hopkins University Baltimore, Maryland, USA M. W. Chen Institute for Materials Research, Tohoku University, Sendai Japan Department of Mechanical Engineering, Johns Hopkins University Baltimore, Maryland USA ABSTRACT To meet the possible increase in future demand for armor materials, an increase in the throughput during manufacturing is necessary. One possibility is the use of the slip casting and sintering technique to form ceramic armor compacts as an alternative to current hot pressing techniques. Dynamic uniaxial compression tests with the Kolsky bar were conducted on two types of slip cast boron carbide, and compared with results from the standard hot pressed boron carbide. One type was slip cast, sintered, and hot isostatically pressed, while the other was only slip cast and sintered. Microstructural characterization by transmission electron microscopy showed graphite inclusions and more annealing twins than in the hot pressed boron carbide material. Examination of fragments recovered from the compression tests determined that the fracture mode of both slip cast materials was brittle transgranular cleavage. The compression test results show comparable compressive strengths between the sip cast and hot pressed boron carbide despite higher density of graphite in the slip cast material. INTRODUCTION Hot pressing boron carbide (B4C) powder is the commercial technique used to form personnel armor plates and components for various applications. B4C is used widely in abrasive, wear resistant components, and armor applications due to its high hardness and low density properties. It is possible for B4C components formed by the hot pressing technique to reach nearly full theoretical density and achieve high mechanical performance. Compared to sintering processes, hot pressing requires less additives for better densification and strength. These additives could however also form precipitates or secondary phases at the grain boundaries and be detrimental to the mechanical performance. The limitation of the hot pressing technique is the high operation cost per batch and only plates or cylindrical shapes of a limited size can be produced. Also, in addition to a larger die, to achieve the same pressure applied to a smaller specimen, a much larger hot press machine size is required for larger specimens. Recently, an alternative technique of forming B4C compacts was described by Matsumoto et al.1. In this technique, B4C powder was slip cast, sintered, and hot isostatically pressed (HIPed). The mechanical properties of these slip cast and HIPed B4C materials were reported2 to be as good as the hot pressed B4C materials. The aim here is to evaluate the dynamic

107

Comparison of Slip Cast to Hot Pressed Boron Carbide

mechanical performance of slip cast-sintered-HIPed, and slip cast-sintered B4C and compare them with hot pressed B4C. In addition to the mechanical testing, microstructural characterization of the HIPed and SCS samples was conducted to better characterize the material pre and post compression tests. EXPERIMENTAL Two types of B4C samples were obtained. One was slip cast and sintered, and the other was slip cast, sintered, and HIPed. The slip cast and sintered samples will be referred to as the "SCS" B4C samples, and when identifying the samples that were also HIPed, described as "HIPed" to distinguishing between the two types. The third sample compared in this study is hot pressed B4C. This B4C sample is the armor grade reference benchmark material and will be referred to as the hot pressed sample. This hot pressed B4C was analyzed in a previous work3 and the data is taken from the paper on the prior analysis. The surfaces of the as received slip cast B4C (SCS and HIPed) samples were imaged with a scanning electron microscope (SEM) equipped with a field emission gun. The samples were polished on the Struers automatic polisher with diamond slurries of decreasing diamond abrasive size at each polishing step, until reaching 0.25 μιη. The samples were then polished on a vibramet polisher with 0.02 μπι colloidal silica. The post-polished slip cast samples were imaged on the SEM. The Knoop hardness of the two samples was measured with a microhardness tester, at loads from 3 N to 98 N (0.3 Kg to 10 Kg). To determine the existence of any impurities and different B-C phases, x-ray diffraction was conducted. The impurity phases were verified, and grain boundaries examined with transmission electron microscopy. The slip cast samples were also machined into the 4.0 mm x 5.2 mm x 3.0 mm geometry for dynamic mechanical testing with the Kolsky bar. Each sample was loaded onto the Kolsky bar ends with lithium grease, which also acts to minimize the friction between the sample and the titanium alloy platens. The full Kolsky bar setup is described in the work by Ramesh and Narasimhan4. Each sample was subjected to dynamic compression at the rate between 150 and 160 MPa^sec. The stress and strain rate throughout the compression test cycle was captured by a high speed camera at 2 or 3 μ8 intervals with 300 to 700 ns exposure time. The sample area was enclosed in a clean polycarbonate box with a clean sheet of paper lining the bottom of the box to collect the fragments after the compression experiment. After each experiment, the polycarbonate box was cleaned and a new sheet of paper was installed to minimize sample contamination and to collect the fragments from the next experiment. The collected fragments were labeled with the B4C processing type and experiment number then examined with the SEM and energy dispersive spectrometry, or EDS. The fracture surfaces of the HIPed B4C samples were compared with the SCS B4C as well as with the hot pressed B4C samples. RESULTS The surfaces of the as machined SCS and HIPed B4C samples were examined in the SEM and EDS. Samples were also prepared and examined with TEM. The various microscopy techniques revealed dark and pore-like areas to be graphite inclusions. Compared to the hot pressed B4C, the slip cast B4C appeared to have more, though smaller sized, graphite inclusions. The as received slip cast B4C samples were x-rayed to determine the phases and identify any impurities. The diffraction peaks for both samples were consistent with each other and can be inferred that both samples have the same phase and impurities. When the diffraction peaks in

108

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Slip Cast to Hot Pressed Boron Carbide

both samples were identified, B4C, BnC2, and carbon (graphite) were determined to be the phases present. Figure 1 shows the peak identification of the HIPed sample. The same phases were identified in the SCS sample. The volume percents of the graphite flakes for both slip cast B4C were measured from SEM images and calculated to be 10%. Even with these graphite inclusions, the density of the HIPed B4C, calculated by the Archimedes method, was 2.50 g/cm3, or 99.2 % of the theoretical density, and 2.45 g/cm3, or 97.2 % of the theoretical density for the SCS B4C. Assuming from the lack of porosity, determined by visual examination of SEM micrographs, that the HIPed sample is fully dense (i.e. no porosity), the remaining 0.8 % or 0.02 g of the density was comprised of graphite. With this assumption, the vol % of graphite was calculated to be 9 vol %. Assuming the same amount of graphite was in the SCS sample, the remaining 2 % of the theoretical density is therefore due to pores. The total volume of the machined sample was measured to be 0.06 cm3. From the difference in the density between the HIPed and SCS samples, the approximate volume percent of pores in the SCS sample was calculated to be 2.0 vol%. Neither volume percent of pores nor graphite was calculated for the hot pressed B4C, but the density was determined previously by Chen et al/ to be 2.49 g/cm3, or 98.8 % of the theoretical density.

Fig. 1. The HIP diffraction pattern with B4C, graphite, and Bi3C2 peaks labeled.

Processing and Properties of Advanced Ceramics and Composites

· 109

Comparison of Slip Cast to Hot Pressed Boron Carbide

Fig. 2. Amorphous interface boundary between graphite and a B4C grain. Compared to the hot pressed B4C, which had possible precipitates and distinct chemical compound at the triple junctions3, the slip cast B4C had fewer impurities. SEM and TEM analysis of both slip cast samples show only graphite inclusions, some trapped within the grain and others at triple junctions. The interface between the graphite inclusion and B4C grains were determined by TEM to be amorphous as shown in Fig. 2. Another observation by SEM and TEM was the numerous twins; more than in the hot pressed B4C samples, as shown in Fig. 3. This is in agreement with previous work by Schwetz et al5. Examination of the 100 nm to 1 μιη grain size B4C powder used in the slip cast samples did not show any twins. However the polished surfaces of slip cast samples that did not undergo compression tests, show many grains with twins. Hence the twins observed in the slip cast samples are not deformation twins and are growth twins formed during the processing steps. After polishing, hardness measurements were conducted on the slip cast B4C samples. Vickers indentation was initially attempted. However due to the high hardness and low toughness properties of B4C, the indentation marks were not measurable. Hence Knoop indentations were performed with varying load. Table I shows the comparison of the Knoop hardness measurements at a load of 19.6 N (HK(2)), in accordance to ASTM C1326. Each sample that underwent dynamic uniaxial compression test with the Kolsky bar was imaged with a high speed camera and the stress and strain values recorded. Figure 4 (a) shows camera frame shots of the SCS sample #1 during the compression test and (b) is the plot of the stress and strain rate over time. In Fig. 4 (a), the sample exhibited cracking from the edges inward. The other SCS samples, as well as the HIPed samples, regardless of the compressive strength, showed similar trends in the stress and strain profiles. All samples displayed throughsample cracking at the time interval just past the maximum stress peak, and destruction of the sample shortly thereafter. The comparison among the SCS, HIPed, and the hot pressed B4C compressive strengths are shown in Fig. 5.

110

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Slip Cast to Hot Pressed Boron Carbide

Fig. 3. Polished surfaces of B4C that was a) SCS, b) HIPed, and c) Hot pressed. Numerous growth twins are visible in a) and b).

Processing and Properties of Advanced Ceramics and Composites

· 111

Comparison of Slip Cast to Hot Pressed Boron Carbide

Table I. Average Knoop Hardness Values at 19.6 N Load Std. Dev. (GPa) Material Ave. (GPa)

(a)

HIP

21.1

0.6

SCS Hot Pressed7

18.6 19.6

0.9 1.8

Inter-frame time 3μ5, exposure time 500ns

Fig. 4, a) High speed camera images of SCS sample #1 during the Kolsky bar compression test. The image numbers correspond to the stress and time plotted in b).

112

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Slip Cast to Hot Pressed Boron Carbide

The fragments from the Kolsky bar experiments were carefully collected and examined in the SEM. All samples displayed fracture by brittle, transgranular fracture mode, evidenced by the cleavage fracture surfaces. Figure 6 (a) is an example of a fractured surface of a fragment from a SCS sample (Sample #1). Twins were observed on all the fractured surfaces. Graphite inclusions were also frequently observed on fractured surfaces, whether due to them actively influencing the fracture path or due to the high density of the inclusions in the material. Similar to Fig. 6 (a), Fig. 6 (b) is a SEM image of a fragment from a HIPed sample. Compared to the sintered B4C samples, the hot pressed samples revealed the same cleavage fracture, however with very few fractured surfaces with twins.

Fig. 5. Compressive strengths of SCS, HIPed, and hot pressed B4C samples.

Fig. 6. (a) An SCS sample fragment surface, (b) A HIPed sample fragment surface.

Processing and Properties of Advanced Ceramics and Composites

· 113

Comparison of Slip Cast to Hot Pressed Boron Carbide

DISCUSSION All B4C samples had graphite inclusions, which for the slip cast samples, was found at the grain boundaries, triple junctions, and trapped within the grains. The graphite was already in the initial powder, and was not formed during the processing. The Vickers hardness values provided by the manufacturer for the HIPed B4C was the highest at 39 GPa (load unknown) followed by the SCS B4C with 32 GPa. The Vickers hardness HV(0.3), provided by the manufacturer for the hot pressed B4C was 26.5 GPa. When Vickers hardness tests were conducted by the authors, severe spalling occurred and the data was deemed unreliable. However Knoop hardness values were successfully obtained. The HK(2) values for the HIPed B4C was 21.1 GPa with a standard deviation of 0.6 GPa and the SCS B4C was 18.6 GPa with a standard deviation of 0.9 GPa. These values as well as the tested range of hardness values from HK(0.3) to HK(10) were in agreement with the Knoop hardness results of hot pressed B4C by Swab7. Except for at HK(10), the HIPed B4C consistently had the highest hardness values, followed by hot pressed, then SCS. This hardness trend follows the density, in which the densest B4C was the HIPed sample, followed by hot pressed, then the SCS sample. For the Kolsky bar compression test, having rough surfaces is known to lower the compressive strength, even with lubricant applied along the contact sides of the test specimen8. If too much lubricant is used, the lubricant can be detrimental by filling in the pores and other depressions on the rough surfaces and force crevices to open. Since the as machined slip cast B4C had many scratches on the surface, it can be expected that with a smoother surface finish, the compressive strengths would be higher. The way to minimize the surface finish problem would be to machine cylindrical dog-bone shaped specimens. However the machining cost for such a geometry would be expensive. Nevertheless assuming that the hot pressed B4C has similar machining difficulties affecting the surface finish as the slip cast material, the compressive stress results should still be comparable. Even with limited number of data and despite all the processing and microstrucutral differences, all three B4C samples performed within a standard deviation from each other in the dynamic uniaxial compression test. Provided that the surface finish was similar, this indicates the importance of the material properties on the compressive strength rather than the processing method to achieve that strength. This also shows that impurities and growth twins play limited roles on the compression performance. It can be deduced that as long as the initial B4C powder is at least 96 wt% pure, the grain size ranges from 5 to 15 μπι, and the sample density is above 98.8 % of the theoretical density, the compressive strength can be expected to fall within the range of 3000 to 4100 MPa. It is shown here that slip casting or slip casting with the HIP step can be used to form unconventional shapes of B4C samples with similar mechanical performance as B4C samples that are hot pressed. Also, though not realized in this study, the addition of the HIP process increases the density which, based on general trends, should increase the mechanical performance. Hence the HIP process could not only complement B4C production, but could possibly produce better B4C compacts. CONCLUSION Microstructural characterization and dynamic uniaxial compression tests with the Kolsky bar were conducted on B4C samples that were SCS and slip cast, sintered, and HIPed. Growth twins were more frequently observed on the surfaces of SCS and HIPed samples than those of hot pressed, benchmark B4C samples. The compressive strengths results were compared with those of previously tested commercially hot pressed B4C samples. The compressive strength of

114

· Processing and Properties of Advanced Ceramics and Composites

Comparison of Slip Cast to Hot Pressed Boron Carbide

the HIPed and non HIPed samples were in the range of compressive strength of the hot pressed B4C samples. Hence slip casting and sintering B4C is a feasible technique to obtain unconventionally shaped samples with comparable compressive strength as hot pressed B4C samples. ACKNOWLEDGEMENTS The authors would like to thank the discussions and characterization assistance provided by the following individuals: Prof. K. T. Ramesh of Johns Hopkins University, Dr. James McCauley, Dr. Ryan McCuiston, and Mr. Herbert Miller of the U. S. Army Research Laboratory. REFERENCES 1. A. Matsumoto, A. Kawakami, and T. Goto, "Slip Casting and Pressureless Sintering of Boron Carbide," Ceram. Trans., 133 M. Matsui, S. Jahanmir, H. Mostaghaci, M. Naito, K. Uematsu, R. Wasche, R. Morrell editors; pp. 223 The American Ceramics Society, Westerville, OH (2002). 2. A. Matsumoto, T. Goto, and A. Kawakami, "Slip Casting and Pressureless Sinter of Boron Carbide and Its Application to the Nuclear Field," J. Ceram. Soc. Japan, Suppl. 112-1, 112[5] S399-S402 (2004). 3. M. W. Chen, J.W. McCauley, J. C. LaSalvia, and K. J. Hemker, "Microstructural Characterization of Commercial Hot-Pressed Boron Carbide Ceramics," J. Am. Ceram. Soc.,88[7] 1935-42(2005). 4. K. T. Ramesh and S. Narasimhan, "Finite Deformation and the Dynamic Measurement of Radial Strains in Compression Kolsky Bar Experiments," Int. J. Solids structures, 33[25] 3723-38(1996). 5. K. A. Schwetz, W. Grellner, and A. Lipp, "Mechanical Properties of HIP Treated Sintered Boron Carbide," Inst. Phys. Conf. Ser. [75] 413-424 (1986). 6. ASTM C1326-96 "Standard Test Method for Knoop Indentation Hardness of Advanced Ceramics" 2003 Annual Book of ASTM Standards, Vol. 15.01. 7. J. J. Swab, "Recommendations for Determining the Hardness of Armor Ceramics," Int. J. Appl. Ceram. TechnoL, 1 [3] 219-25 (2004). 8. J. C. LaSalvia, verbal communication, 2006.

Processing and Properties of Advanced Ceramics and Composites

· 115

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Amorphous Ceramics

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

MECHANICALLY DRIVEN AMORPHIZATION AND BULK NANOCRYSTALLINE SYNTHESIS OF ULTRA-HIGH TEMPERATURE CERAMICS H. Kimura Department of Mechanical Engineering, School of Systems Engineering National Defense Academy Yokosuka, Kanagawa, Japan ABSTRACT This article reports the process control methodology for the mechanically driven amorphization and the bulk nanocrystalline synthesis of ultra-high temperature ceramics (UHTCs) such as B4C, SiC, ZrB2 and B4C· SiC in non-equilibrium solid state powder processing. The amorphization by mechanical grinding UHTCs particle is formulated by a linear chemical reaction with the repeated impaction number N, d(Vr-V)/dN=-k(V[-V) where Vf and V is the amorphous volume at the final and intermediate stage respectively. Planetary ball milling is used as a model experiment by which to characterize the solid state amorphization of UHTCs; the k is found to relate to a manipulatable milling parameter; it increases with decreasing powder weight and temperature and in order of B4C
119

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

Reaction Milling NanoMechanochemistry

Model Nanomech antes

Figure 1 Integrated nanosystem and strategic nanodesign for the synthesis of ultra-high temperature ceramics using non-equilibrium solid state P/M processing. MECHANICALLY DRIVEN AMORPHIZATION Nanomechanochemical Reaction The 'nanomechanochemical reaction', as this term is used, is a novel mechanically driven phenomenon of ceramics in which to achieve the non-equilibrium synthesis via an accumulation of internal strain energy inside particles by repeated impaction, although the mechanochemical reaction occurs mainly in the aid of surface activation and is formulated with the variables of reaction time and surface area. The mechanically driven amorphization is set up on the basis of linear chemical reaction with the process variable of repeated impaction number, N as follows: d(VrV)/(W=-k(VrV) (1) where V and Vf are the amorphous volume at the intermediate and final stage respectively. For the process analysis of ball milling, the equation (1) becomes (Vr-V)/(VrV0 = exp(-kiV) (2) where Vj is the amorphous volume at the initial stage. Planetary ball milling can be used as the model experiment of mechanically driven amorphization as illustrated in Figure 2. The repeated

Figure 2 A scheme of planetary ball milling for mechanically driven amorphization of UHTCs.

120

· Processing and Properties of Advanced Ceramics and Composites

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

impaction number as is the macroscopic variable is given by the relationship of the form, N=aRt, where a is the ratio of the rotation of vessel on its axis to an orbital rotation of disk, R and t is milling time. The impaction of the falling ball at an inner wall of vessel is considered as main driving force for nanomechanochemical reaction inside the milled powder. PROCESS CONTROL METHODOLOGY Powder weight and temperature The mechanical grinding (MG) of the B4C powder with the mean crystallite size of 160 nm can produce the amorphization without the formation of an intermediate phase at a revolving velocity of 10 s"1 using the mono-vial typed planetary ball mill (Fritsch P-6) with a=T.82 as shown in Figure 3. When taking an integrated X-ray peak intensity (/P, /) of crystalline phase at the initial and intermediate stage respectively, equation (2) becomes, avoiding the need of Vb -In (7//P) = kN (3) During mechanically driven amorphization of B4C particle, the III? continuously decreases; its reduction rate decreases with increasing sample mass, and concomitantly the crystallite size (d) as deduced from X-ray line broadening decreases down to about 80 nm as shown in Figure 4.

Figure 3 The amorphization sequence by mechanical grinding commercial B4C particles by XRD.

Figure 4 The mean crystallite size and IIh versus milling time during MG amorphization of B4C

Processing and Properties of Advanced Ceramics and Composites

· 121

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

Figure 5 A process analysis for mechanically driven amorphization of the B4C particle on the basis of linear chemical reaction with the variable of repeated impaction number. Figure 5 shows the relationship between the -In (///P) and the repeated impaction number with two different powder weights of 20 and 30 gr at both room temperatures of 20 and 30 °C in the case of B4C. The In (/p//) is found to be proportional to JY, showing that the mechanically driven amorphization follows the equation (1) without a preparatory stage for the B4C particle having the d of 160 nm. Then, the k decreases with increases in powder weight and temperature inside the vessel, and is expected to be coupled with manipulative parameters of planetary ball milling such as size, number and density of the ball, diameter of the vessel, R and a. Structure MG amorphization of stoichiometric β-SiC particle is obtained by (a) 'high energy' milling of the nanocrystalline (nc) powder and (b) 'low energy milling' of the commercial powder with submicron grain size f51 as shown in Figure 6. Two amorphization processes are differentiated by the relationship between an amorphous volume fraction (X) and an average crystallite size during amorphization as shown in Figure 7. Further milling of the nanocrystalline

Figure 6 Mechanically driven amorphization of (a) nanocrystalline and (b) sub-μm sized SiC particles using the mono-vial and two-pot typed planetary ball mil respectively.

122

· Processing and Properties of Advanced Ceramics and Composites

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

Figure 7 The amorphous volume versus crystallite size in the cases of (a) nanocrystalline SiC and (b) submicron sized SiC and B4C particles. (nc) stoichiometric β-SiC, as synthesized according to c-Si+h-C->u-(c-Si+am-C)-»ml-(c-Si/amC)-»>nc-(P-SiC), shows a decrease from 4.2 to 2.6 nm in average crystallite size during the amorphous (am) synthesis; this dependence roughly follows a relation of the form, X=l{dl(d+A)Y, where A is the amorphous boundary thickness, in the nanocrystal. While, during MG amorphization of the commercial β-SiC and B4C particle, the average crystallite size, which is described by a master curve, decreases from sub-μιη to approximately 28 nm with increasing V. Figure 8 shows the relationship between In (Vf-Vi)/(VrV) and total repeated impaction number N{ for mechanically driven amorphization of the nanocrystalline and commercial SiC powders. In this plotting, one see a good linearity, but its extrapolation into zero level at the axis of amorphous volume gives a certain impaction number Np indicating the presence of a preparatory stage necessary for the onset of the amorphization. So, the repeated impaction number should be taken using N=NX-NP for the process analysis on the basis of eqn. (1). Unexpectedly, the k for nanocrystalline SiC that is at a thermodynamically higher energy state is smaller than that of sub-micron sized SiC, although MG of nc-particle was conducted at a higher

Figure 8 Relationship between In (VrVi)/(VfV) and N{ for mechanically driven amorphization of the nanocrystalline and commercial SiC powders

Processing and Properties of Advanced Ceramics and Composites

·

123

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

(b)

(a)

Figure 9 High-resolution electron microscopy of the solid state amorphization by mechanical grinding the nanocrystalline powder. ball acceleration of 18 G under /?=8.3 s"1 in the mono-vial typed planetary ball mill (Fritsch P-6) relative to 11 G at R=\0 s"1 using the two-pot type mill (Fritsch P-7) with a smaller sized ball for the sub-μπι particle. This discrepancy is accounted for by a higher temperature inside the vessel in the case of nanocrystalline SiC analogous to the result as shown in Figure5. High-resolution electron microscopy confirms that the lattice spacing of 0.26 nm inside the crystallite is equal to that of (111) plane of β-SiC as shown in Figure 9, suggesting that MG amorphization of nanocrystalline SiC occurs at an intercrystal network. On the other hand, a lower temperature is needed to suppress the recovery of heavily introduced defects inside an excited crystallite in the case of the mechanically driven amorphization of sub-micron sized β-SiC. Composition The mechanically driven amorphization of metallic boride, ZrE*2 occurs by mono-vial typed planetary ball milling at R=\0 s"1 as shown in Figure 10, while the formation of a unidentified metastable crystalline phase and partial crystallization occurs at a given high temperature.

30

40

50

60

2<9/(7i/180)rad

Figure 10 X-ray diffraction pattern of mechanically driven amorphization ofZrBi at an orbital rotation of the disk of 10 s'1 using the mono-vial typed planetary ball mill.

124

· Processing and Properties of Advanced Ceramics and Composites

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

Figure 11 Relationship between In (VrVi)/(VfV) and N for mechanically driven amorphization of the ZrBi particle. This figure includes the results of SiC and B4C for comparison. Figure 11 shows the relationship between In (\Wi)/(Vf-V) and N for mechanically driven amorphization of ZrB 2 , SiC and B 4 C. The k shows a decrease in order of B 4 C am-(SiC-B 4 C). Figure 13 shows a compositional image of the amorphous B4C-SiC particles, as synthesized by mechanical grinding at 504 h, in a back scattering mode by field emission scanning electron microscopy (FESEM). A finely divided nanoscale particle (bright particle), which may be considered as amorphous SiC, is

Figure 12 XRD of ball-milled B4C-SiC and B4C at the milling time of 504 h using the mono-vial typed planetary ball mill.

Processing and Properties of Advanced Ceramics and Composites

· 125

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

Figure 13 FESEM of the binary amorphous B4C-SiC particle synthesized by mechanical milling at 504 h in a back scattering mode. distributed throughout the amorphous B4C-S1C powder agglomerate with a diameter in a range from 0.1 to 0.6 μπι. The distribution profile of the particle diameter (dp) is fitted with a curved line as calculated by a power law relationship, Np=Adp]'*, where Np is the particle number, not by a customarily used function of hanging bell typed size distribution as shown in Figure 14. This result will be coupled to the mechanically driven amorphization via the nanomechanochemical reaction as formulated on the basis of the linear chemical reaction.

Figure 14 The size distribution of finely divided crystalline particle in the amorphous B4C-S1C particle as ball- milled at 504 h. BULK NANOCRYSTALLINE SYNTHESIS Process Modeling and Control Figure 15 shows the densification of the amorphous SiC compact under an applied stress of 100 MPa without additive in instrumented pulse-electric current sintering (PECS). MG amorphous SiC powder exhibits the rapid densification in heating above 1743 K mainly during

126

· Processing and Properties of Advanced Ceramics and Composites

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

Temperature, Ts/K 1800 1900 2000

10 20 30 Time, tl min

Figure 15 Electricfieldassisted consolidation of the amorphous SiC powder compact under the applied pressure of 100 MPa in pulse electric current system. Newtonian viscous flow in the nanocrystallized phase and then is consolidated at a full-density during superplastic flow at a relatively low temperature of 2033 K |S| . The densification during Newtonian viscous flow (τγ=σ^/3ε\ whereη is the process viscosity and
Figure 16 Arrhenius-type relationship of Newtonian viscous flow for the rapid densification during heating using pulse electric current consolidation of the amorphous SiC powder.

Processing and Properties of Advanced Ceramics and Composites

·

127

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

In Process Nanocrystalline Control Densification Figure 17 shows the relative density versus the surface temperature (7s) for the nanocrystalline SiC powder compact as prepared by PECS. The temperature necessary to obtain full densification of nanocrystalline SiC tends to decrease with increasing applied pressure as expected from equation (4), and is a considerably low level of 2033 K at the applied stress of 100 MPa, relative to that in customarily used powder metallurgy. Consolidated nanocrystalline β-SiC 1 Q 0.9|

SiC φ 13.2 mm 115 MPa

T3

♦

100 MPa

■a 0.7

— "

0.6|

A

72 MPa

0.5 1800

1900 2000 Surface temperature, Tsl K Figure 17 The relative density of the nanocrystalline SiC product synthesized using pulse electric current consolidation of the amorphous powder under the applied pressure of 100 MPa. has a compositionally uniform structure without the incorporation of other elements, contrary to the powder compact as prepared using the precursor-derived amorphous SiBCN synthesized via thermolysis of element-organic polymer. The d greatly decreases with decreasing temperature for consolidated SiC as shown in Figure 18. Then, a full-density B4C product, as consolidated at approximately 1600 K under a pressure of 100 MPa using PECS, has the d of approximately 5 nm being nearly equal to a threshold. In process structure control densification under electric field t7,81 is now provided to synthesize a non-equilibrium phase 19 ' in covalent bond type ceramic. 70 U

SiC 100 MPa

30 U 1550 1600 1650 1700 1750

_L

T/C

Temperature, Figure 18 The average crystallite size versus temperature for the pulse electric current synthesized nanocrystalline SiC product using the mechanically synthesized amorphous powder.

128

· Processing and Properties of Advanced Ceramics and Composites

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

MECHANICAL CHARACTERISTICS Figure 19 shows the applied load (P) as a function of length (c0) of the crack in Vickers indent for full-density nanocrystalline SiC as prepared by PECS of the amorphous powder. The fracture toughness (K{C), as deduced from P/c03/2=K]C/A(E0/H,)n with A=0.016 and n=0.5 (E0 is Young's modulus and Hv is Vickers hardness number) for a median/radial crack, is approximately 13 MPa-m° 5 , this value is higher than those of μπι sized SiC denoted as Al-2233

Figure 19 Fracture toughness as evaluated by the indentation micro fracture method for nanocrystalline SiC as prepared by the pulse current consolidation of the mechanically synthesized amorphous powder. This figure includes the cases of conventionally processed μπι sized SiC for the comparison. and 6H-SiC, and corresponds to the flexural strength (σ?) of more than 1 GPa when considering a proportionality between <jf and K\c |10 ' in nanocrystalline ceramics. Nanocrystalline UHTCs powder compact has the advantage of high-strain rate superplastic flow [11] and/or Newtonian viscous flow to develop a variety of the net shape forming techniques at a relatively low temperature for ultra-high-temperature ceramics and its composites. CONCLUSIONS The methodology of mechanically driven amorphization and pore free consolidation have been constructed in non-equilibrium solid state powder processing, in order to provide a basis for the material design and synthesis of nanocrystalline UHTCs in widespread applications. Firstly, the formulation based on the linear chemical reaction for nanomechanochemical reaction is generalized for the amorphous powder synthesis and control of SiC, B4C, ZrE$2 and a binary system of S1C-B4C by using planetary ball milling. Then, the pulse electric current consolidation is used to set up a way of in process nanocrystalline control densification of the amorphous covalent ceramic powder under an electric Field, and is used to obtain pore free nanocrystalline products of UHTCs at considerably low temperatures. Thus-obtained nanocrystalline SiC compact has unique mechanical property inherent to the nanocrystal such as high fracture

Processing and Properties of Advanced Ceramics and Composites

· 129

Mechanically Driven Amorphization and Bulk Nanocrystalline Synthesis

toughness and high-speed superplastic flow. The nanomaterial system and design will be a fruitful strategic method to synthesize nanocrystalline UHTCs having widespread applications in the threefold approach consisted of nanoprocessing, nanostructure and nanofunction [l2,13] . REFERENCES [I] H. Kimura, Advanced Powder Processing of Three-Dimensionally Nanostructured Ceramics, Materials Integration, 12 (1999), p 19-26. [2] H. Kimura, Integrated Material System for Bulk Nanocrystalline Ceramics, 4th Pacific Rim Int. Conf. on Advanced Materials and Processing (PRICM4), (2001), p 163-166. [3] H. Kimura, A Breakthrough via Nanocrystalline Synthesis in Structural Ceramics, J. Metstable and Noncrystalline Materials, 15-16 (2003), p 591-598. [4] H. Kimura, Amorphous Alloy by Reaction Ball Milling, Materials Processing and Design, Rapid Solidification Technology (Lancaster Technomic, 1993), p 71-123. [5] H. Kimura, K. Hanada and K. Ishigane, Mechanical Alloying and Powder Consolidation in SiC and Hydroxyapatite, Materials Science Forum, 502 (2005), p 211-216. [6] H. Kimura, Non-Equilibrium Powder Processing of Full Density Nanoceramics, Advances in Powder Metallurgy & Paniculate Mateials-1999 (MPIF), 12(1999), p 55-61. [7] H. Kimura, Pore Free Consolidation with Nanocrystalline Control in Ceramics, Ceramic Transactions, 194 (2005), p 251-262. [8] H. Kimura and K. Hongo, In Process Nanocrystalline Control Consolidation of the Amorphous ZrO2-20mol%Al2O3 Powder, Materials Transactions, 47 (2006), p 1374-1379. [9] H. Kimura and T. Uchino, Materials Processing for Properties and Performance (MP3), Vol. II (Institute of Materials East Asia, 2004), p 347-358. [10] H. Kimura, Mechanical Characterization of Nanocrystalline Ceramics Synthesized via NonEquilibrium Solid State Processing, J. Jpn. Soc. Powder Powder Metallurgy, 55 (2008), p 502-508. [II] H. Kimura and Y. Fujimoto, High-Speed Superplastic Forging of Three Dimensionally Nanostructured Ceramics via Pulse Electric Discharge Typed Hot Pressing, J. Japan Society Powder Powder Metallurgy, 46 (1999), p 1274-1278. [12] H. Kimura, P/M Science and Technology for Bulk Nanocrystalline Materials, Materia Japan, 42 (2003), p 37-44. [13] H. Kimura, Integrated Materials Design of Bulk Nanocrystalline Ceramics, Proc. of Third Int. Conf. on Processing Materials for Properties (PMPIIf) (TMS, Bangkok), in the press.

130

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

PREPARATION AND CHARACTERIZATION OF FUSED SILICA BASED CERAMIC CORES USED IN SUPERALLOY CASTING M. Arin, S. Sevik, and A. B. Kayihan Tubitak Materials Institute Gebze, Kocaeli, Turkey ABSTRACT Fused silica based ceramic core samples with different compositions were formed by injection molding and sintered at 1550°C for 1 h. Compositions were altered by addition of ZKD2 in either calcined form or stabilized form with 3mol % yttria. The effects of crystalline phase content, sintering conditions, and grain size on flexural strength, porosity, and liquid metal interactions of ceramic samples have been investigated. The experimental work revealed that the optimal composition for ceramic cores is fused silica with 17 wt% stabilized Zr02. Increase in fracture strength, low shrinkage and low thermal expansion coefficient were the reasons for this selection. When tested for metal interaction, the selected ceramic showed no sign of a reaction with metal. When penetrated into metal, ceramic cracked due to differential densification during sintering, differental thermal expansion coefficients, loss of adherence between the grains, and phase transformation of Zr02 upon cooling. INTRODUCTION Leachable ceramic cores, used for obtaining complex internal structures such as cooling passages in Nickel-base superalloy gas turbine blades, are often silica-based ceramics. These complex ceramic cores are produced by forming in injection, transfer molding, or rapid prototype printing, followed by sintering, and turbine blade is cast around them [1-3]. During casting the core is heated at 15-300°C/min, and after casting it is removed by leaching in NaOH or KOH which do not attack Ni-base superalloy. In order to cast turbine blades around applicable ceramic cores, there are some restrictions on the ceramic [3,41 : • The ceramic should not react with the superalloy during casting so it must be chemically compatible with the alloy • The ceramic should be dimensionally stable • The ceramic should have high thermal shock resistance properties and enough strength • The ceramic should be weaker than the superalloy so that it can crush before hot cracking occurs in the metal during cooling because of high stresses • The ceramic should be easily removable and should not deteriorate mechanical properties of turbine blade. Silica seems to be used as cores exclusively in turbine blade industry, because of the ability to remove it without affecting the casting's properties. For massive and complicated cores, complex silicate compositions have been used successfully. The cores can be prepared by vitreous silica particles by adding some amounts of zircon (ZrSi04) or zirconia (Zr02) and an organic binder to bond the particles together. Ingredients are mixed together as a slurry to be molded into a core by injection moulding machine. Zircon which is known for its high refractoriness, low thermal expansion, and low thermal conductivity, is the only compound in the Zr0 2 - Si0 2 system that

131

Preparation and Characterization of Fused Silica Based Ceramic Cores

decomposes into basic components of zirconia and glassy silica. Zircon is generally highly resistant to most chemicals with the exception of hydrofluoric acid [5]. Zirconia on the other hand creates more stable and durable interfaces. Improvement in mechanical properties of the cores by adding zircon or zirconia is attributed to the tetragonal to monoclinic phase transformation of particles at the tip of a propagating crack [6]. Pre firing is applied at 1000°C after injection moulding in order to enhance strength of the ceramic core. %vol 25 porosity is needed for crushing the cores conveniently after the casting process. [3]. Wang and Hon [6] investigated phase transformation kinetics of fused silica and concluded that cristobalite seed assists the crystallization but induce the compressive stress from volume expansion of alpha to beta transformation. Chao and Lu [7] investigated the composition and sintering temperature effects on strength of the zircon-fused silica core samples. They found that strength being inversely related to sintering temperature was due to beta to alpha transformation and loss of coherency between the silica grains. The purpose of this study was to prepare four different ceramic core samples containing fused silica, zircon and zirconia. Microstructure, flexural strength, phase transformation kinetics, and interaction with the superalloy was investigated. EXPERIMENTAL PROCEDURE Four different ceramic core sample slurries containing industrial grade fused silica with an addition of 17 wt% calcined ZrU2 or 3 mol%Y203 stabilized ZrU2 or ZrSi04, and also about 38% paraffin, oleoic acid, and polyethylene as binders were mixed using the blender in lowpressure injection molding machine. The slurries were then pressed into rectangular plates with dimensions of 60mm><25mmx3mm (length x width x thickness) by using 5atm pressure. After binder removal at 1150°C for 1 h in Protherm furnace, the samples were sintered at 1550°C for lh. Crystalline phases of the as-sintered samples were determined by XRD (Shimadzu XRD 6000) whereas the microstructure of the fracture surfaces, microcharacterization and grain size distribution were investigated by SEM (Jeol JSM 6335F). Porosity of the samples was measured by means of quantachrome poremaster. The samples were tested with three-point bending test using Zwick Z250 universal test equipment in order to calculate flexural strength of samples. The flexural strengths of the samples were taken as the average of five bending tests. In order to determine chemical stability of the ceramic samples, interaction test was performed in Balzers vacuum furnace 15 kW 10kHz, by preparing immersion test apparatus. 60mmxl2.5mm><3mm plates of ceramic were cut using diamond wheel cutting machine, the ceramic plate was suspended from a graphite rod using a wire above an alumina crucible under vacuum. Metal chunks of Ni-base superalloy (Inconel 617) were placed inside the alumina crucible as a metal charge which could be brought to its melting point that is 1330-1360°C for inconel 617. After loading, the furnace was evacuated to 10"4 Torr, and after the pump is closed the furnace was subsequently flushed with a constant stream of argon about 400 Torr to minimize contamination. Once the metal was molten, the energy input to the system was held constant and the lower part of the ceramic was pushed into the molten metal and was left still for 1 min and 2 min intervals by turning off the generator. After cooling, metal-ceramic combinations were cut horizontally so that the cross section of the ceramic as coated with metal can be investigated by SEM.

132

· Processing and Properties of Advanced Ceramics and Composites

Preparation and Characterization of Fused Silica Based Ceramic Cores

Table 1 Composition (%wt) of the selected superalloy Inconel 617 Cr Ni Co Mo Al C

22.0 balance 12.5 9.0 1.0 0.07

RESULTS AND DISCUSSION A microstructural investigation of fracture surfaces of Fused Silica- based ceramic samples with the addition of Zr0 2 (coded as Z), 3 mol%Y203 stabilized Zr0 2 (as 3YZ), or ZrSi04 (as ZS), sintered at 1550°C for lh, revealed that two phases could be observed in Figure 1, i.e., Zr0 2 (white), Si0 2 (gray). Because of the grain size difference between the Zr0 2 , and ZrSi04, and SiO? powders, the pore distribution of the mixed samples is high according to fused silica sample. The high porosity percentage might have been due to micro cracking and loss of coherency between the grains during debinding or incomplete infilling during injection moulding since air holes replaces binder during debinding. It is observed that the grain size increases with the addition of calcined zirconia (Z), however decreases with the addition of yttria-zirconia(3YZ) and zirconia-silica(ZS). It is known that grain size decreases with increasing yttria content [9]. Also the crystal structure of zircon is tetragonal like 3 YZ with each silicon atom surrounded by four oxygen atoms [5], although calcined zirconia is monoclinic. The crystalline phases can be observed in XRD analysis graph shown in Figure 3.

FS

FS

Processing and Properties of Advanced Ceramics and Composites

·

133

Preparation and Characterization of Fused Silica Based Ceramic Cores

Figure 1 SEM Micrographs of Fracture Surfaces of Ceramic Core Samples

134

· Processing and Properties of Advanced Ceramics and Composites

Preparation and Characterization of Fused Silica Based Ceramic Cores

Figure 2 Porosity Graph of Ceramic Core Samples As seen from the crystalline phases in XRD graph, all sintered samples consists of a mixture of predominantly α-cristobalite and a minor amount of α-quartz. The only tetragonal Zr02 peak was observed in FS-3YZ sample. FS-Z sample contains only monoclinic Zr02 phase. It can be concluded that, some of tetragonal ZrSiU4 and ZrU2 phase has been transformed into monoclinic phase upon cooling in FS-3YZ and FS-ZS samples. Because of this transformation, fracture strength has increased for FS-ZS and FS-3YZ samples as seen in Table 2. It can be concluded that enhancement of fracture toughness occurs by transformation toughening mechanism which prevents crack propagation due to monoclinic phase [9]. Table 2 Mechanical Properties of the Ceramic Core Samples E-Modulus (MPa) 553±131 1225±40 2196±990 2985±1167

Sample FS FS-Z FS-ZS FS-3YZ

Fracture Strength (MPa) 3±2 3±2 13±6 14±5

Table 3 Inconel 617 composition Element Weight% Atomic% OK SiK CrK MnK ZrL

18.65 59.17 5.85 2.24 14.10

Totals

100.00

32.56 58.85 3.14 1.14 4.32

Processing and Properties of Advanced Ceramics and Composites

·

135

Preparation and Characterization of Fused Silica Based Ceramic Cores

Figure 3 XRD Analysis Graph of Ceramic Core Samples

Figure 4 shows the SEM micrographs of FS-ZS and FS-3YZ ceramic samples as coated with Inconel 617, together with an EDX spectrum from an area starting from the surface of the sample through middle which is darker in color. The rest (whiter) is ceramic completely. It can be observed from the Figure 5 that there is no significant reaction zone between the metal and ceramic. The level of metal attack increases with time. Metal propagates through the ceramic grains due to high porosity (Figure 6) and possible microcracking during sintering due to incomplete debinding process. The metal zone is larger in FS-ZS sample than in FS-3YZ sample. This can be a consequence of higher porosity level of FS-ZS sample and/or the high grain coherency by the addition of 3 mol% yttria to stabilize ZrC>2.

136

· Processing and Properties of Advanced Ceramics and Composites

Preparation and Characterization of Fused Silica Based Ceramic Cores

FS-ZS - Imin

FS-ZS - 2min

FS-3YZ- Imin

FS-3YZ- 2min

Figure 4 SEM Micrographs Showing the Ceramic-Metal Interaction

Processing and Properties of Advanced Ceramics and Composites

·

137

Preparation and Characterization of Fused Silica Based Ceramic Cores

Figure 5 SEM Micrograph Showing the Ceramic-Metal Interaction Zone ofFS-ZS

Figure 6 SEM Micrograph Showing the Metal Attacking in Ceramic-Metal Interaction Zone of FS-ZS

138

· Processing and Properties of Advanced Ceramics and Composites

Preparation and Characterization of Fused Silica Based Ceramic Cores

CONCLUSIONS An optimal composition for fused silica based ceramic cores has been found in the sample prepared with fused silica powders with the addition of 17 wt% Zr0 2 stabilized with 3 mol% Y2O3. The optimum composition has higher strength, as flexural strength increasing from -5 MPa to -15 MPa and the shrinkage after sintering is - 1 % due to high chemical stability and low coefficient of thermal expansion of fused silica. It was concluded that Ni based superalloys can not react with FS-ZS ceramic under described experimental conditions. As mentioned above, ceramic cores should have a certain amount of porosity for a practical production. However when metal penetrates to ceramic plates, pores are filled with liquid metal and ceramic samples form microcracks due to different thermal expansion coefficients of metal and ceramic. Those microcracks propagate even more while cooling. Zr02 phase in the ceramic tends to go on a phase transformation and forms a tension across ceramic-metal interface. As amount of stress on the metal/ceramic interface increases, cracks propagates fast and leads to fractures through ceramic samples. REFERENCES [1] S. Uram, Application of Ceramic Cores for Investment Casting, Proceedings of the Third World Conference on Investment Casting, European Investment Casters' Federation, 1972, p 1-9 [2] S. Uram, Ceramic Cores for High-Melting-Point Alloys, Foundary Mag., 99 [7], 1971, p 4853 [3] C. Huseby, M. P. Borom, and C. Greskovich, High Temperature Characterization of SilicaBase Cores for Superalloys, Am. Ceram. Soc. Bull., 58 [4], 1979, p 448-452 [4] A. A. Wereszczak et al., Dimensional Changes and Creep of Silica Core Ceramics Used in Investment Casting of Superalloys, Journal of Mater. Sci. 37, 2002, p 4235-4245 [5] P. Charaska, K. Neufuss, H. Herman, Plasma Spraying of Zircon, JTTEE5 6 1997, p 445-448 [6] C. Aksel, Key Eng. Mat. Vols. 264-268 (2004) pp 1791-1794 [7] L. Y. Wang, M. H. Hon, The Effect of Cristobalite Seed on the Crystallization of Fused Silica Based Ceramic Core - A Kinetic Study, Ceramics International 21, 1995, p 187-193 [8] C. H. Chao, H. Y. Lu, Optimal Composition of Zircon-Fused Silica Ceramic Cores for Casting Superalloys, J. Am. Ceram. Soc. 85 [4], 2002, p 773-79 [9] S. Lawson, J. ofEur. Cer. Soc, Vol 15, 1995, p 485-502

Processing and Properties of Advanced Ceramics and Composites

·

139

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Coatings and Films

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

PHOTON EFFECTS IN ULTRA-THIN OXIDE FILMS: SYNTHESIS AND FUNCTIONAL PROPERTIES S. Ramanathan, M. Tsuchiya, C. L. Chang and C. Ko School of Engineering and Applied Sciences, Harvard University, MA 02138, USA ABSTRACT We present a brief summary of our recent results on synthesis and mechanistic studies on effects of photon irradiation on structure and functional properties of ultra-thin oxide films, with representative examples from fluorite, rutile structured oxides and amorphous alumina. It is shown that ultra-violet irradiation is a unique approach to tailor oxygen concentration, film microstructure in thin film oxides as well as functional properties in a range of temperatures. The technique is versatile; it can be used both during synthesis as well as a post-processing approach. I. INTRODUCTION Ultra-thin metal-oxide films are of great interest from scientific and technological perspectives. They can serve as model systems to investigate surface and interface effects on ion transport.1 Microstructures in such thin films can be tuned by choice of synthesis and substrate conditions in turn providing paths to explore structure-property relations including space charge effects on electrochemical conductivity and grain boundary mobility2. Further, ultra-thin oxide films are imperative for a variety of technologies in electronics and energy. Examples include the extensive on-going research in high-k dielectric thin films for CMOS devices3,4 and in the case of high temperature applications, in areas such as solid oxide fuel cells5,6. Within this context, reducing the thickness of the electrolyte from few microns down to tens of nanometers is deemed to be one of the approaches to enabling low to intermediate temperature operation of fuel cells . In turn, such ultra-thin electrolytes require novel processing that makes them dense, pin-hole free, low leakage and mechanically robust to withstand severe mechanical strains that may arise from temperature cycling and intrinsic stress differences4. Examples of ultra-thin electrolytes that are being investigated include well known ion conductors in the fluorite family, namely yttria-doped zirconia and Gd-doped ceria7'8. It is important to note if the solid oxide fuel cells are operated at lower temperatures (say < 600 °C), several other candidates that were previously not considered for high temperature operation could become attractive: provided their ionic transference number is sufficiently high to prevent fuel cell losses. Within the context of energy sciences, several other applications of oxide films exist in the areas of corrosion protection9, efficient materials processing10, coatings11 and so forth. We have been researching the effect of photon irradiation (primarily in the ultra-violet range) on both synthesis as well as functional properties of ultra-thin fluorite-structured oxides and related systems. We refer here to ultra-thin regime as sub-lOOnm thickness. The rationale for exploring this approach is the following: it has been conclusively shown that oxidation under UV can enhance the quality of oxide thin films in a number of ways ranging from improvement in film density to reducing oxygen-related defects to improving crystallinity, dielectric constant and so forth. These studies have been performed in a range of oxides, including, but not limited to, Si02, Zr02, (Ba.Sr)Ti03 indicating the broad relevance of the processing approach. Mechanisms leading to such effects include the enhanced incorporation of oxygen through additional driving force arising from the electric field12 as well as creation of atomic oxygen and ozone in the near-vicinity of the oxide surface13,14. In the case of oxidation, presence

143

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

of atomic oxygen leads to reduced activation barrier for oxygen incorporation significantly.14 Oxidation of semiconductors such as silicon and III-V compounds have been investigated as well15 as well as in-situ exposure during complex oxide growth16 and in general high-quality oxides have resulted from this synthesis technique. II. EXPERIMENTAL Thin film yttria-doped zirconia (referred to as YDZ) was synthesized by a number of ways: Rf-sputtered from a composite target; oxidation of Y-Zr alloys that were co-sputtered; electron-beam evaporation from a composite target. In some cases, the oxide films were exposed to photon irradiation at fixed temperature and oxygen partial pressure. In the case of alloys, we performed experiments that included both oxidation under photon irradiation as well as thermal oxidation for control. Alumina films were grown starting from thin film aluminum followed by oxidation either under photon irradiation or in natural (native oxide) conditions. Vanadium oxide films were grown by reactive sputtering followed by UV treatments in an environmental probe station.17 The film thickness typically ranged from few nm to about lOOnm, measured by cross-sectional transmission electron microscopy (TEM) and X-ray reflectivity (XRR). Ion scattering experiments on select samples were performed at the Pacific Northwest National Laboratory to extract information on composition and interfacial layers. This was particularly important for zirconia films grown on semiconductor substrates such as silicon. Electrochemical conductivity studies were performed in-plane on these films in a custom high temperature probe station. Multiple experiments were performed on phase stabilized samples to avoid transient effects. The conductivity studies were performed in various ambient as well. In select cases, conductivity measurements were performed directly under photo-illumination in a specially designed probe station with transparent windows. The UV sources were Hg vapor lamps of varying intensity, which emit primarily in 254nm and 185nm range and further details on the photon sources can be found elsewhere.18 III. RESULTS AND DISCUSSION Figure 1 shows a cross-section high-resolution transmission electron micrograph of yttria-stabilized zirconia grown on germanium substrate.

Figure 1: X-TEM image taken from a YSZ film grown by e-beam evaporation on Ge substrate.

144

· Processing and Properties of Advanced Ceramics and Composites

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

The film is crystalline and grown conformally on the substrate with negligible interfacial layer. As-deposited grain size is ~ 5-10 nm and they are randomly oriented. The substrate was chosen since it doesn't form an interfacial layer with YSZ and can be used for some model studies. The absence of interfacial layers was also confirmed by Rutherford backscattering (RBS) performed at PNNL. Figure 2 shows a representative RBS spectrum from the sample, SIMNRA fitting of the data agreed well with absence of reaction layers. Such films were later used for annealing experiments to investigate phase stability. YSZ films grown on Si substrates invariable led to an interfacial Si0 2 formation that has also been reported by other researchers. In some experiments, we have actually exploited these phenomena to measure kinetics of interfacial layer growth as an indirect probe of oxygen diffusion through the YSZ films as a function of doping and synthesis and will be discussed later. Figure 3 shows plan view image of ultra-thin YSZ film grown by oxidation of a Y-Zr precursor alloy under UV irradiation at room temperature.

Figure 2: Rutherford backscattering data taken from the YSZ film on Ge subsrate at PNNL (ack. V. Shutthandan). Fitting the data with SIMNRA indicates absence of interfacial layers in agreement with the TEM results.

Processing and Properties of Advanced Ceramics and Composites

· 145

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

Figure 3: Plan view TEM micrograph of UV-oxidized YSZ film on silicon substrate. The asgrown film is polycrystalline with nanometer sized grains (< 5 nm) As-grown film is crystalline and cubic. We have previously reported in detail the phase transformation routes for thin film YSZ as a function of size and doping effects19. Upon high temperature treatment, there is grain growth; typical grain sizes are of the order of film thickness or smaller after annealing at T > 700 °C or so. Representative bright field TEM image of YSZ film following annealing at 900 °C is shown in figure 4.

Figure 4 : Plan view TEM image of YSZ film grown by Y-Zr alloy oxidation under UV irradiation along with inset diffraction pattern. Grain size is 15-20 nm and grains are randomly oriented.

146

· Processing and Properties of Advanced Ceramics and Composites

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

Cross-sectional TEM image of UV-grown 7.5 mol% yttria doped zirconia film is shown in figure 5 below along with thermally grown YSZ film for comparison. An interfacial layer between the YSZ film and Si were formed during the annealing at 900°C for 1 hour in air. It is interesting to note that the thickness of the interfacial layer formed in both cases is quite different, even though top zirconia film thickness, dopant concentration were identical. Molecular oxygen decomposes at the zirconia surface and diffuses into the film as atomic oxygen, i.e., the interfacial layer oxidation proceeds without large chemisorption barrier.20 Therefore, the interfacial growth is limited by diffusion of atomic oxygen species in zirconia. The difference in interface layer thickness indicates oxygen transport in UV-grown 7.5 mol% YSZ films is slower than that in thermally grown films with identical Y concentration.This may be due to a combination of reasons including differences in film density as well as oxygen defect concentration.

Figure 5: Cross-sectional TEM images of YSZ film grown by (1) thermal oxidation and (2) UV oxidation of precursor Y-Zr alloys. The Y composition is identical in both cases as it is fixed during co-sputtering. In a separate set of experiments, we have investigated the interfacial layer thickness evolution as a function of yttrium doping concentration to monitor defect association phenomena. Further discussions on this aspect can be found in a reference by Tsuchiya and Ramanathan.21 We have measured the electrochemical conductivity of such films in detail as well using Pt electrodes. Figure 6 shows a set of Nyquist plots taken from a UV-oxidized YSZ film as a function of temperature. A single arc is seen, that is typical of in-plane thin film measurements. The absolute values of the conductivity are in good agreement with those reported in literature for YSZ films. The results overall indicate that the UV-grown ultra-thin YSZ films are of high quality, dense and pin-hole free and may be suitable for applications involving ultra-thin barriers or ion conductors.

Processing and Properties of Advanced Ceramics and Composites

· 147

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

■ • Δ

200

400 600 Z' (kQ)

925 °C 885 °C 845 °C

800

Figure 6: Nyquist plots taken from YSZ film grown by oxidation of Y-Zr precursor alloy under UV irradiation We briefly discuss the effects of UV-assisted synthesis on properties of alumina thin films next. We have initiated a systematic study of effect of UV irradiation on properties of ultrathin alumina films grown on Al surface. The motivation for this study includes correlating point defects to impedance across an ultra-thin oxide film. Al thin films were grown by sputtering followed by oxidation to form a thin (~ 3 nm) AI2O3 film on top. The oxides were grown by one of the following approaches: (1) natural oxidation, ie oxidation in oxygen gas in a load lock, (2) UV oxidation, ie oxidation under UV photons in a custom built load lock chamber and (3) combination, ie native oxide exposed to UV photons for varying periods of time. These samples were further studied by X-ray photoelectron spectroscopy, electrochemical impedance spectroscopy (EIS) in a 0.5M NaCl aqueous medium and atomic force microscopy. Figure 7 shows the impedance data plotted in Nyquist format from alumina films synthesized under varying UV exposures. It can be clearly seen that the resistance (corresponds to the x-axis value of the radius of the semicircle) of the UV-synthesized alumina film (120 minutes) has the highest impedance value. An interesting point to note is that the impedance of native oxides can be improved by UV exposure. This is primarily due to the reduction in oxygen vacancies that is enabled by the UV-enhanced oxygen incorporation. From cross-sectional TEM, we have estimated there is not a significant difference in thickness between the various oxides, hence a primary reason for improvement in impedance could be the reduction in leakage currents enabled by combination of microstructural and defect properties. X-ray photoelectron spectroscopy studies indicate that UV-synthesized alumina films can effectively alter the chloride ions uptake behavior compared to native oxide films. A detailed analysis of the compositional depth profiling across these oxide films can be found elsewhere.

148

· Processing and Properties of Advanced Ceramics and Composites

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

Figure 7: Nyquist plots taken from alumina films exposed to varying periods of UV illumination, the impedance of the films changes and reaches a self-limiting value that is kinetically controlled. We then present some representative data on thin film vanadium oxide (VO2). This is a fascinating materials system owing to the presence of multiple oxidation states of vanadium. Obtaining phase pure vanadium di-oxide is a non-trivial problem especially on semiconductor substrates such as silicon.22 VO2 undergoes a metal-insulator transition (MIT) in the vicinity of 67 °C that can be tuned by strain, choice of substrate and doping.22 The strength of the MIT strongly depends on various factors including the phase purity and crystallinity of the sample. It is important to be able to synthesize high-quality V0 2 films for several reasons: it is predicted to be an ideal Mott transition material23, and experimental investigation into these phenomena requires high-quality materials. Further, V0 2 is of interest for a variety of catalysis as well as solid state sensing applications such as bolometers24 that take advantage of the transition. Hence, we have been studying the synthesis of VO2 films under a range of experimental conditions including UV-assisted synthesis. Figure 8 shows the MIT induced by temperature for two representative VO2 films grown with different oxygen content on sapphire followed by UV exposure for varying times at room temperature. It can be clearly seen that the resistance of the films can be modulated one either sides of the phase transition by UV irradiation. It is important to note that these measurements were performed after illumination (ie not during the electrical measurements) and the changes in resistance are likely due to changes in oxygen nonstoichiometry. The sample exposed to UV photons while being kept at higher temperature showed more significant alteration in the temperature dependence of resistance Further studies were performed as a function of temperature and time to investigate the kinetics of stoichiometry changes due to photon irradiation.17 The trends are in agreement with enhanced kinetics predicted by field-assisted ion migration typically found for oxidation.

Processing and Properties of Advanced Ceramics and Composites

· 149

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

Figure 8: Hysteresis curves of two sets of vanadium oxide Films (exhibiting varying magnitude of MIT ratio, four and two orders respectively in (a) and (b)) without UV exposure as well as 60min and 240-min UV exposure at room temperature (25 °C). Two insets for each figure show the resistance changes in metal and insulator phases in more detail respectively. IV. ACKNOWLEDGEMENTS The authors acknowledge Dr. V. Shutthanandan (PNNL) for assistance with the ion scattering experiments. Financial support from ARO, ONR, IQSE and SEAS, Harvard that have enabled various aspects of this research is gratefully acknowledged. V. REFERENCES 1 2

3

4 5

150

J. Maier, Nature Materials 4 (11), 805 (2005). I. W. Chen, Interface Science 8 (2-3), 147 (2000); I. W. Chen, Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 166 (1-2), 51 (1993). P. T. Chen, Y. Sun, E. Kim et al., Journal of Applied Physics 103 (3) (2008); G. D. Wilk, R. M. Wallace, and J. M. Anthony, Journal of Applied Physics 89 (10), 5243 (2001); J. L. Gavartin, A. L. Shluger, A. S. Foster et al., Journal of Applied Physics 97 (5) (2005). A. A. Demkov, Physical Review B 74 (8) (2006). J. H. Shim, C. C. Chao, H. Huang et al., Chemistry of Materials 19 (15), 3850 (2007).

· Processing and Properties of Advanced Ceramics and Composites

Photon Effects in Ultra-Thin Oxide Films: Synthesis and Functional Properties

H. Huang, M. Nakamura, P. C. Su et al., Journal of the Electrochemical Society 154 (1), B20 (2007). A. Karthikeyan, M. Tsuchiya, C. L. Chang et al., Applied Physics Letters 90 (26) (2007); T. Suzuki, I. Kosacki, and H. U. Anderson, Solid State Ionics 151 (1-4), 111 (2002); X. D. Zhou, W. Huebner, I. Kosacki et al., Journal of the American Ceramic Society 85 (7), 1757 (2002); I. Kosacki, T. Suzuki, V. Petrovsky et al., Solid State Ionics 136, 1225 (2000). A. Karthikeyan, C. L. Chang, and S. Ramanathan, Applied Physics Letters 89 (18) (2006). L. M. Serna, K. R. Zavadil, C. M. Johnson et al., Journal of the Electrochemical Society 153 (8), B289 (2006); K. R. Zavadil, J. A. Ohlhausen, and P. G. Kotula, Journal of the Electrochemical Society 153 (8), B296 (2006). A. Krishnan, X. G. Lu, and U. B. Pal, Scandinavian Journal of Metallurgy 34 (5), 293 (2005). R. G. Wellman and J. R. Nicholls, Journal of Physics D-Applied Physics 40 (16), R293 (2007). C. L. Chang and S. Ramanathanz, Journal of the Electrochemical Society 154 (7), G160 (2007). S. Ramanathan, D. Chi, P. C. Mclntyre et al., Journal of the Electrochemical Society 150 (5), Fl 10 (2003). S. Ramanathan, P. C. Mclntyre, J. Luning et al., Philosophical Magazine Letters 82 (9), 519(2002). C. F. Yu, M. T. Schmidt, D. V. Podlesnik et al., Journal of Vacuum Science & Technology B 5 (4), 1087 (1987). V. Craciun, J. M. Howard, N. D. Bassim et al., Applied Surface Science 168 (1-4), 123 (2000); K. Ramani, C. R. Essary, V. Craciun et al., Applied Surface Science 253 (15), 6493 (2007). C. Ko and S. Ramanathan, Journal of Applied Physics 103 (10) (2008). M. Tsuchiya and S. Ramanathan, Applied Physics Letters 91 (25) (2007). M. Tsuchiya, A. M. Minor, and S. Ramanathan, Philosophical Magazine 87 (36), 5673 (2007). S. Ferrari and G. Scarel, Journal of Applied Physics 96 (1), 144 (2004); B. W. Busch, W. H. Schulte, E. Garfunkel et al., Physical Review B 62 (20), R13290 (2000). M. Tsuchiya and S. Ramanathan, Applied Physics Letters 92 (3) (2008). J. Nag and R. F. Haglund, Journal of Physics-Condensed Matter 20 (26) (2008). M. M. Qazilbash, M. Brehm, B. G. Chae et al., Science 318 (5857), 1750 (2007); H. T. Kim, B. J. Kim, Y. W. Lee et al., Physica C-Superconductivity and Its Applications 460, 1076(2007). B. J. Kim, Y. W. Lee, B. G. Chae et al., Applied Physics Letters 90 (2) (2007); C. H. Chen, X. J. Yi, X. R. Zhao et al., Sensors and Actuators a-Physical 90 (3), 212 (2001).

Processing and Properties of Advanced Ceramics and Composites

· 151

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

FARADAYIC PROCESS FOR ELECTROPHORETIC DEPOSITION OF THERMAL BARRIER COATINGS FOR USE IN GAS TURBINE ENGINES Joseph Kell, Heather McCrabb Faraday Technology, Inc. Clayton, Ohio, USA ABSTRACT Faraday is developing an electrically mediated electrophoretic deposition process (Faradayic EPD) for thermal barrier coatings. This process is capable of creating a uniform coating of yttria-stabilized zirconia (YSZ), in a non-line of sight fashion, for use in thermal barrier systems in natural gas and synthesis gas environments. Ni-based superalloy substrates, coated with an MCrAlY bond coat, were deposited on and underwent a binder burnout and sintering process. The samples were then subjected to surface roughness measurements, microstructural examination, thermal conductivity measurements, thermal cycling tests, and other appropriate tests of merit to determine their feasibility for use in gas turbine engines. Experimental studies showed that the process could uniformly deposit YSZ with good surfaces, decreased edge effects, and with properties consistent with similar systems processed through other methods. INTRODUCTION Operating environments in gas turbine engines present a significant challenge from both the design and manufacturing viewpoints. This environment is especially aggressive in the hot sections of the turbine and leads to failure of the turbine blades and vanes. This environment, which consists of hot, corrosive gasses that are often contaminated with chlorides and sulfates, and an enriched oxygen atmosphere, is accompanied with severe mechanical wear, and erosion1 as well as harsh thermo-mechanical conditions such as thermal cycling and thermal shock. Thermal barrier systems, generally consisting of a combination of a thermal barrier coating (TBCs) and a bond coat, have been developed as an effective way of protecting the materials within a gas turbine engine from the damage that results from these harsh environments. The bond coat is generally placed directly onto the substrate material and serves two functions; it protects the substrate from oxidation and alleviates many of the thermal expansion mismatch related stresses that develop between the substrate (usually a Ni-based superalloy) and the TBC (often yttria-stabilized zirconia)2,3. The TBCs are then placed on top of the bond-coated substrate. Their function is to mitigate the damage that results from mechanical, oxidative, and thermal environments by generating a steep thermal gradient through the coating system, which results in the substrate experiencing lower temperatures and therefore having improved properties. In order to provide this protection, the TBC material must have certain properties to ensure its robustness in its role. These properties include being chemically inert, having a high melting point, a low thermal conductivity, a low sintering rate, and a thermal expansion coefficient similar to the base substrate . Yttria-stabilized zirconia (YSZ) is the principal material used for TBCs as it is largely chemically inert and will not oxidize and, when applied onto the bond coat, it provides a degree of corrosion protection. In addition, it performs well when subjected to thermal cycling and thermal shock, largely due to its relatively high thermal expansion coefficient and melting point. It also has a low thermal conductivity, which further helps protect the bond coat and substrate

153

Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings

against the high temperatures encountered within gas turbine engines5. Generally, YSZ TBCs are applied by either air plasma spray (APS)6,7 or electron beam assisted physical vapor deposition (EB-PVD)6'8 9. Unfortunately, both of these deposition methods have limitations, which are often contradictory. APS films can be produced quickly and economically and have a low thermal conductivity, but are noted for having thickness variation and spallation problems10,11. EB-PVD produces films that have good wear and spallation resistance but also have higher thermal conductivities than their APS counterparts and are generally slower and more expensive to produce7"8'11. As a result, neither process shows significant improvement over the other without losing out on the main advantages of the competing process. However, electrophoretic deposition (EPD) may provide a solution to this predicament. In conventional EPD, 2 electrodes are immersed into a stable suspension of charged deposition particles. A constant voltage is then applied across the electrodes, creating an electric field that migrates the particles to the oppositely charged electrode (substrate)12. Due to the field's ability to readily wrap itself around electrodes, EPD is a largely non-line of sight process. It also has several other advantages, including fast deposition rates, the ability to coat complex shapes uniformly, low levels of coating contamination, a reduction of material waste encountered in spraying or non-directional coating methods, and relatively simple deposition equipment. In addition, many of the problems associated with conventional EPD or direct current EPD (DC EPD), such as the thickness non-uniformities, hydrolysis, and other electrochemically related problems, can be resolved using electrically mediated waveforms, also known as the Faradayic EPD process13. Previous work at Faraday has demonstrated that electrically mediated waveforms (including pulse, pulse-reverse, and more complex waveforms) can substantially modify the deposition related characteristics of electrodeposited coatings14*15*16 and anodically finished surfaces17 through enhanced control of the electric field. EXPERIMENTAL Suspensions consisting of YSZ particles (8 wt% YiCh-stabilized Zr02, submicron powder, Sigma Aldrich) in ethanol (EtOH) were made for EPD. Polyvinyl alcohol (86-89% hydrolyzed, dissolved in water) (PVA) was used as a binding agent and poly(diallyldimethylammonium chloride) (PDDA) dissolved in water as the cationic dispersant. YSZ, EtOH, and PDDA were mixed together and ultrasonicated to create a stable coating suspension. PVA was then added subsequently to prevent competitive adsorption between the binding and charging agents. The solution was again mixed and ultrasonicated and allowed to age overnight. A glass electrophoresis cell was used for the EPD experiments. Iridium oxide coated titanium was used for the anode and bond coated nickel-based superalloy substrates for the cathode. The electrode distance was 1 cm. The substrates were placed into an in-house manufactured housing made of glass reinforced PTFE to avoid contamination issues and to prevent warping of the holder during the drying step. Depositions were performed with the cathode and anode in the vertical plane and the deposition of particles onto the face of the substrate. Solutions were agitated just prior to each EPD experiment to ensure uniform particle distribution prior to deposition. NiCoCrAlY bond coated IN939 buttons were used as substrates. The nominal diameter of the substrates was -2.54 cm. General EPD conditions, bath properties, and substrate properties are outlined in Table I. For this paper, two studies were done to examine various parameters and their effects on the Faradayic EPD process. Basic waveforms for both deposition experiments are shown in

154

· Processing and Properties of Advanced Ceramics and Composites

Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings

Figure 1. The first study consisted of forward pulsed waveforms using a low duty cycle and a iv (a: (b)

n vf

Vf

^ .3 0 11^ ™ ™ £ \r

«of

«or

ΓΊ

time

Fm^mmm

^ w

vr

Figure 1. Basic waveforms for (a) Faradayic pulsed EPD experiments and (b) Faradayic pulse reversed EPD experiments. high frequency while varying the deposition voltage. The duty cycle (dc) is defined as the percentage of time the electric field is turned on and the frequency is the inverse of the sum of the electric field on-time and off-time for a single pulse. The frequency and duty cycle were chosen based off of previous optimization work carried out at 50 V1 ,16. To allow for comparison with prior EPD experiments, the total deposition time for the Faradayic EPD samples was calculated by using the following equation: Ea*td*dc = Ea*t0= 900 V*s*cm"1

(1)

where Ea is applied electric field in V/cm, td is the total deposition time in seconds, dc is the duty cycle (unitless) and to is the sum of the total time that the electric field is on during the experiment in seconds (to = 180 s was used in previous work). Voltage was used as the depositions are being performed using voltage control. For this study, voltages of 25 V, 50 V, 100 V, and 200 V were investigated. After the deposition, the samples were dried for a few hours at ~95°C to drive off the solvent (i.e. EtOH) and immediately cooled in a desiccator prior to being weighed. Table I: Typical Substrate, Bath, and EPD conditions for the Faradayic EPD of TBCs Substrate Properties Bath and Deposition Properties 1 cm Base Material IN939 Electrode Distance Bond Coat PVA NiCoCrAlY Binder -2.54 cm Charging Agent PDDA Diameter 50 V cm'1 Thickness Electric Field 0.35 cm 2.7 mA cm'2 Bond Coat Ra Current Density 10.84 ±0.56 Mm Deposition Rate (50 V) 29.5 ± 5.7 mg cm'2 min"1

The second study examined the effects of using a simple pulsed reverse waveform consisting of a forward on time, reverse on time, an off time, a low forward and reverse duty cycle, a high frequency, 50 V for the forward pulse, -50 V for the reverse pulse. The values of the reversed duty cycles (dcr) are expressed in ratios of the forward duty cycle (dCf) (dcr = {0.00 dct, 0.25 dCf, 0.50 dCf, 0.75 dCf, 1.00 dCf). The frequency and forward duty cycle were chosen based

Processing and Properties of Advanced Ceramics and Composites

· 155

Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings

off of previous optimization work carried out at 50 V15*16. To maintain consistency with previous experiments, the total deposition time was chosen to maintain the total on time of the forward pulse at 3 minutes. Samples were then dried for a few hours at -95 °C to drive off the solvent (i.e. EtOH) and then immediately cooled in a desiccator prior to being weighed. X-ray diffraction (XRD) was also performed on sintered samples to determine if the EPD processing and sintering were affecting the phase of the YSZ. A sample was deposited using a low duty cycle and a high frequency at 50 V. The sample then underwent a combined binder burnout and 4-hour sintering processes at 1092 °C. A section of the film was removed after sintering for XRD. The sample was scanned from 10° to 80° along 2Θ and compared to XRD PDF cards for standard zirconia phases (tetragonal, cubic, and monoclinic ΤΛΟΊ). RESULTS AND DISCUSSION Faradayic Pulsed EPD - Voltage Modulation Experiments Pulsed current depositions were performed in a similar manner as in previous work15,16 while only changing the applied peak voltage. This was done to establish if the voltage optimized in DC EPD work18 was still optimal for pulsed EPD. An experimental array consisting of four voltages (25 V, 50 V, 100 V, 200 V) was established to study how the peak voltage affected the coating properties. Deposition times were adjusted to normalize the deposition masses to expected values. During the tests, no visible hydrolysis was observed, however upon removal from the bath, the edges of the 100 V and 200 V samples showed evidence that some hydrolysis had occurred, indicating that large peak voltages yield significant hydrolysis. Figure 2a shows the effects that changing the peak voltage has on the deposition mass. The measured values of deposition mass showed a distinct, non-linear response to the peak voltage. This is possibly the result of the hydrolysis interfering with the deposition of material onto the substrate at large peak voltages; however this is in contrast to the results from DC EPD studies that showed large amounts of hydrolysis have a minimal effect on the deposition mass18. Therefore, there may be a mass transport issue similar to the case of a limiting current being observed during electrolysis. The analysis of deposition mass also showed a large variance in the amount of material deposited at 25 V. In addition, the 25 V samples were not always uniformly covered, with three of the five samples having several regions of very thin film thickness dispersed with a matrix of thicker film. The 25 V samples had a coefficient of variation (CoV) of

Figure 2. Effects of (a) voltage and (b) the duty cycle of the reverse pulse on the deposition mass.

156

·

Processing and Properties of Advanced Ceramics and Composites

Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings

25%, indicating a relatively large spread as typically depositions have a CoV of < 10%. These results tend to indicate that, for this bath composition, the deposition peak voltage should be maintained at 50 V. Faradayic Pulsed Reverse EPD - Duty Cycle Modulation Experiments Pulsed reverse current depositions maintained similar waveforms to standardized pulsed depositions (low duty cycle, high frequency) but introduced a reverse pulse of equivalent voltage as the forward pulse immediately after the forward pulse, followed by an off time. The reverse time was varied to determine how this would affect the deposited film. Prior to this experiment, it was unclear if the individual particles in the film would respond to the reverse pulse, as it remained unclear if the particles still maintained a charge after being deposited onto the cathode. During the tests, no hydrolysis was observed and no evidence of hydrolysis was apparent on the films after they were removed from the suspension. Figure 2b shows the effects of changing the reverse duty cycle with respect to the forward duty cycle, which is expressed as a ratio of the reverse duty cycle (dcr) to the forward duty cycle (dCf). In this case, the measured values of deposition mass showed a linear response of the deposition mass with respect to the dcr/dcr ratio. It was also noticed that when dcr = dCf, the film on the surface of the sample was not visible and the coating around the sides of the sample appeared similar to that achieved while dip coating the sample with the current EPD suspension. These results indicate that the particles not only maintain a charge after deposition, but they also show that the adhesion force of the individual particles is small enough that they can be pulled off of the surface of the deposit individually by the reverse pulse, as opposed to the entire mass being removed collectively. X-Ray Diffraction X-ray diffraction (XRD) was performed on a film deposited under low duty cycle, high frequency deposition conditions and then sintered at 1092 °C for 4 hours. The scan is shown in Figure 3. The peaks of the scan show that the YSZ film is in the tetragonal phase and that no

Figure 3: XRD scan of pulsed EPD YSZ TBC compared to the characteristic peaks of tetragonal zirconia.

Processing and Properties of Advanced Ceramics and Composites

· 157

Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings

monoclinic phase is present in the sample. This is particularly important because the tetragonal to monoclinic phase transformation results in a volume expansion that is detrimental to the mechanical stability of the coating. While it is believed that the film does not contain any cubic phase, this cannot be determined directly from this scan, as the cubic phases peaks are very close in both 2Θ and intensity to several of the tetragonal peaks. Further work will try to determine if any cubic phases are present within the Faradayic EPD film. CONCLUSION Depositions were performed to determine the effects of peak voltage on pulsed EPD and duty cycle on pulse reversed EPD. Several important characteristics were found. First, hydrolysis related defects occur for peak deposition voltages of 100 V or higher. In addition, pulsed depositions performed at 25 V showed coating uniformity problems and difficulty controlling the deposition rate. Overall deposition rate did not scale linearly with voltage; likely due to competing processes (hydrolysis, etc.) and mass transport limitations becoming more pronounced at higher voltages. For pulse reverse EPD experiments, the deposition rate scaled linearly with the duty cycle ratio dcr/dCf. Also, when dcr = dCf, the net deposition rate was nearly zero. This confirmed that the particles maintained some charge after deposition while the bulk of the film was not fully charged, less it be removed completely during the reverse pulse. Also, XRD confirmed that the EPD process and thermal post-processing do not affect the phase of the YSZ and that the YSZ is tetragonal. ACKNOWLEDGEMENTS This material is based upon work supported by the Department of Energy under Grant No DE-FG02-05ER84202. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the Department of Energy. Faraday gratefully acknowledges Siemens Power Generation for supplying substrates for deposition experiments. Faraday also acknowledges the University of Dayton Research Institute (UDRI) for performing the XRD work. REFERENCES ! D. Wolfe, J. Singh, Functionally gradient ceramic/metallic coatings for gas turbine components by high-energy beams for high-temperature applications, J. Mater. Sci., 33 (14), 3677-3692 (1998). 2 D.R. Mumm, A.G. Evans, Failure of a Thermal Barrier System Due to a Cyclic Displacement Instability in the Thermally Grown Oxide, Mat. Res. Soc. Symp. Proc, 645E, M2.6.1-M2.6.6 (2001). 3 C. Leyens, U. Schultz, M. Bartsch, M. Peters, R&D Status and Needs for Improved EB-PVD Thermal Barrier Coating Performance, Mat. Res. Soc. Symp. Proc, 645E, Ml0.1.1-Ml0.1.12 (2001).

X.Q. Cao, R. Vassen, and D. Stoever, Ceramic materials for thermal barrier coatings, J. Eur. Ceram. Soc, 24 (1), 1-10 (2004). T. Narita, S. Hayashi, L. Fengqun, K.Z. Thosin, The Role of Bond Coat in Advanced Thermal Barrier Coating, Mat. Sc Forum, 502, 99-104 (2005).

158

· Processing and Properties of Advanced Ceramics and Composites

Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings

J.D. Vyas, K.L. Choy, Structural characterisation of thermal barrier coatings deposited using electrostatic spray assisted vapour deposition method, Mat. Sc. Eng., Mil (1-2), 206-212 (2000). H. Wang, R.B. Dinwiddie, Characterization of Thermal Barrier Coatings Using Thermal Methods, Advanced Engineering Materials, 3 (7), 465-468 (2001). 8 D.D. Hass, P.A. Parrish, H.N.G. Wadley, Electron Beam Directed Vapor Deposition of Thermal Barrier Coatings, J. Vac. Sci. Technol., 16 (6), 3396-3401 (1998). P. Hancock, M. Malik, Materials for Advanced Power Engineer, Part I, edited by D. Coutsouradis etal., Kluwer Academic, Dordrecht, The Netherlands, 685-704 (1994). I0 W. Beele, G. Marijnissen, A. van Lieshout, The Evolution of Thermal Barrier Coatings-Status and Upcoming Solutions for Today's Key Issues, Surf. & Coatings Tech., 120/121, 61-67 (1999). n E. Tzimas, H. Mullejans, S.D. Peteves, J. Bressers, W. Stamm, Failure of Thermal Barrier Coating Systems Under Cyclic Thermomechanical Loading, Acta Materialia, 48 (18-19), 46994707 (2000). 12 T. Ishihara, K. Sato, Y. Takita, Electrophoretic Deposition of Y203-Stabilized ZrCh Electrolyte Films in Solid Oxide Fuel Cells, J. Am. Ceram. Soc, 79 (4), 913-919 (1996). 13 H. McCrabb, M. Inman, E. Taylor, Faradayic Electrophoretic Deposition of Thermal Barrier Coatings, Poster Presented at 31st International Conference & Exposition on Advanced Ceramics and Composites, Daytona Beach, Fl, USA, The American Ceramics Society, ICACC-S2-0542007 (2007). 14 E.J. Taylor, etal., Electrically Mediated Plating of Semiconductor Substrates, Chip Scale Packages & High-density Interconnect PWBs, Plating and Surf. Tin., 89, 88, (May 2002). 15 H.A. McCrabb, J.W. Kell, B. Kumar, Faradayic Process For Electrophoretic Deposition Of Thermal Barrier Coatings, presented at 32nd International Conference & Exposition on Advanced Ceramics and Composites (ICACC 2008), Dayton Beach, Fl, USA, The American Ceramics Society, ICACC-S8-013-2008 (2008). 16 J.W. Kell, H.A. McCrabb, B. Kumar, Electrophoretic Deposition of Thermal Barrier Coatings by the Faradayic Process, Proceedings of the Materials Science and Technology 2008 Conference, Accepted 23 Jun 2008. 17 J. J. Sun, etal., Electrically Mediated Edge & Surface Finishing for Automotive, Aerospace & Medical Applications, Plating and Surf. Tin., 89, 94, (May 2002). 18 J. Kell, H. McCrabb, B. Kumar, Faradayic Process for Electrophoretic Deposition of Thermal Barrier Coatings, in Advanced Processing and Manufacturing Technologies for Structural and Multifunctional Materials II: Ceramic Engineering and Science Proceedings, Volume 29, Issue 9, ed. T. Ohji and M. Singh, Wiley, to be published Oct. 2008

Processing and Properties of Advanced Ceramics and Composites

· 159

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

A NOVEL METHOD TO SPRAY TUNGSTEN CARBIDE USING LOW PRESSURE COLD SPRAY TECHNOLOGY J. Wang, J. Villafuerte CenterLine (Windsor) Ltd, Windsor, ON, Canada ABSTRACT High carbide content composite coatings of tungsten carbide-aluminum and tungsten carbide-copper were produced by low-pressure cold spray system using a novel method. This method consisted of coating precursor carbide powders with either aluminum or copper prior to cold spraying. The resulting microstructure and mechanical properties of these coatings were compared with coatings produced from mixtures of tungsten carbide and copper or aluminum powders. It was observed that deposits produced from metal-coated tungsten carbide were denser, displayed higher hardness, retained more carbide phase and had better dispersion of carbides than deposits produced from equivalent metal-carbide mixtures. 1. INTRODUCTION Tungsten carbide (WC) - Cobalt (Co) coatings are well known for their extensive use in wear resistance applications. The hard WC phase provides wear resistance while Co binder increases toughness. The implementation of regulations to eliminate the environmental impact related to the evolution of Cr+6 during traditional chromium plating processes has opened other possibilities for WC-Co. Over the last few years, there has been a trend to replace electrolytic hard chrome (EHC) with WC-Co hard coatings sprayed by high velocity oxyfuel (HVOF) [1"3]. However, because of the elevated process temperature, HVOF tends to promote oxidation as well as undesirable metallurgical transformations in the substrate material[ ~5\ Low-pressure, cold-gas dynamic spraying (low pressure cold spray) is a unique lowtemperature spraying process in which the spray materials are not melted in the spray gun but rather kinetically deposited on the substrate at low temperatures (less than 200C). Subsequently, there are no thermal effects such as oxidation, distortion, residual stresses and/or undesirable metallurgical transformations. The resulting deposits are dense (less than 0.5% porosity) and with good bonding strength (>4000 psi). Mechanical and/or metallurgical bonding is possible due to extensive and localized plastic deformation resulting from high velocity particle impact[6]. A schematic of a low pressure cold spray system is illustrated in Fig. 1.

161

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology

One of the characteristics of low-pressure cold spray is the use of a limited amount of ceramic particles mixed into the metal powder to further increase bond strength and density of the deposits. After spraying, a residual of the ceramic phase is left embedded in the metal matrix, producing some dispersion strengthening [7]. On the other hand, attempting to spray primarily ceramic material, such as WC, using low-pressure cold spray technology has proven particularly challenging. This is because a minimum volume fraction ductile species is necessary for kinetic bonding to occur [8]. Chemical vapor deposition (CVD), electro plating, and electroless plating are known techniques used to modify the surface chemistry of metallic and non-metallic powders [9]. Ductile metals such as copper and aluminum can be deposited on the surfaces of individual WC particles. The presence of a ductile layer around WC is expected to improve the spray-ability of WC by low-pressure cold spray. Also, a higher volumetric fraction of WC in the deposit is expected, which could yield better mechanical and physical characteristics for wear resistance. The objective of this work was to study the feasibility of cold spraying WC using metal-coated WC powders and compare the resulting properties with those from coatings made using metalcarbide powder blends. 2. EXPERIMENTAL 2.1 Feedstock Powders

Fig. 2. Raw powder morphology used in the study (a) Al, (b) Cu, and (c) WC. A number of commercially pure powdered materials were used in this study, including aluminum (-325 mesh, Atlantic Equipment Engineers, USA), copper (-325 mesh, Acupowder International LLC, USA), and WC (14-25 urn, Buffalo Tungsten Inc, USA), as illustrated in Fig. 2. Table 1 depicts the composition of Al-WC, Cu-WC, Al-coated WC, and Cu-coated WC

162

· Processing and Properties of Advanced Ceramics and Composites

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology

powders used for these experiments. Al-WC and Cu-WC blends were produced by mechanical mixing. Aluminum coating on WC particles was produced by chemical vapour deposition using a proprietary technique. Copper coating on WC particles was produced by electroplating. All powders were cold sprayed on 1018 carbon steel and 6061 aluminum coupons.

Feedstock 1 Cu-coated WC Cu-WC blend-1 Cu-WC blend-2 Cu-WC blend-3 Al-coated WC Al-WC blend-1 Al-WC blend-2 1 Al-WC blend-3

WC Composition (wt%) 80 20 40 70 70 20 40 70

1,

2.2 Cold Spraying A SST™ portable low-pressure cold spray system was employed to produce the composite coatings. A convergent-divergent (De-Laval) round-section nozzle was used with an expansion ratio of 6.4 and divergent section length of 120mm. Compressed air at 80-90 psi was used as the carrier gas. Air temperature and pressure were monitored at the spray gun. The substrate materials (steel 1018, aluminum 6061) were grit blasted with 80-grit alumina prior to spraying. A summary of the cold spraying parameters is depicted in Table 2. Table 2 Processing parameters of the cold spray Parameters Settings Compressed air 1 Gas 375-540 Gas temperature (°C) Gas pressure (psi) 80-90 Powder feed rate (%) 45 15 1 1 Standoff distance (mm) 2.3 Characterization Coated samples and powders were sectioned using conventional metallographic techniques. Optical microscopy and scanning electron microscopy (SEM) were employed for powder morphology, microstuctural analysis, and phase quantification. Energy-Dispersive XRay (EDX) was conducted on selected samples. Hardness testing was performed using a Brinell tester.

Processing and Properties of Advanced Ceramics and Composites

· 163

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology

3. RESULTS AND DISCUSSION 3.1 Al-coated and Cu-coated WC powder

Fig. 3 SEM images of WC powder (a) before coating and (b) after aluminum coating using chemical vapour deposition (CVD). CVD was used to coat individual WC particles with aluminum, as illustrated in Fig. 3. It was apparent that aluminum did not deposit homogeneously over the surface of WC particles. Rather, sub-micron and micron sized spherical aluminum particles were observed over the carbide particles as aggregated clusters. These observations were confirmed by EDX analysis on these spheroids. The amount of aluminum introduced in the CVD chamber was based on the designed thickness of aluminum coating on the WC particles with the assumption that all aluminum would present in the powder coatings. Therefore, presence of aluminum as submicron spheres may have decreased the amount of effective aluminum available on carbide surfaces. Electroplating was used to deposit copper on carbide particles. Contrary to the case of aluminum, electroplating produced a continuous layer of copper around WC particles (Fig. 4). Further microstructural observations of cross-sections of selected copper-coated particles confirmed the presence of a uniform and continuous layer of copper (Fig. 5).

Fig. 4 SEM images of WC powder (a) before coating and (b) after copper coating using an electroplating technique.

164

· Processing and Properties of Advanced Ceramics and Composites

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology

Fig. 5 Cu coated WC powder cross-sectional microstructure (a) optical microscopy image (b) back-scattered electron image. 3.2 WC-Al and WC-Cu Composite Coatings Al-coated and Cu-coated WC feedstock powders were successfully cold sprayed with the Centerline portable system, using compressed air. Fig. 6 illustrates the typical microstructures of WC-Al and WC-Cu composite coatings in the as-sprayed condition. The addition of a layer of soft metal around carbide particles appeared to provide sufficient surface with the necessary ductility to stimulate solid-state particle-to-substrate and particle-to-particle bonding. The powder appeared to behave somewhat like a regular soft metal powder upon cold spraying. After spraying, the carbide phase appeared well dispersed within a pure aluminum matrix. The volumetric percentage of dispersed carbide phase in the deposit was estimated to be 55% for both Al-coated and Cu-coated powders.

Fig. 6 Microstructure of cold sprayed (a) aluminum-coated WC and (b) copper-coated WC. As a comparison, Al-WC and Cu-WC blends with three different weight fractions of WC powder (20wt%, 40wt% and 70wt%) were cold-sprayed using the same system and parameters (Table 2). Fig. 7 illustrates the appearance of deposits made from Al-WC blend and Cu-WC

Processing and Properties of Advanced Ceramics and Composites

· 165

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology

blends. Consistent with others [7J, the amount of retained carbide phase in the aluminum matrix reached a plateau that could be well below the amount of carbide present in the original Al-WC blend. For Al-70wt% WC feedstock, the maximum amount of carbide phase retained in the deposit was about 30 vol%. For Cu-70wt% WC feedstock, the maximum amount of carbide

retained was higher at about 65 vol%. Additionally, in both cases, the retained carbide was not homogeneously dispersed in the matrix. Fig.7. Microstructure of composite coatings produced from mixtures of (a) WC-30wt % pure aluminum powder and (b) WC-30wt % pure copper powder. Hardness measurements indicated higher hardness values for cold spray deposits produced from metal-coated carbide feedstock compared to cold spray deposits produced from metal-carbide powder blends (Fig. 8 and 9). This was attributed to better dispersion of carbide phase in the aluminum or copper matrix produced from metal-coated carbide feedstock.

Fig. 8. The macro-hardness of the coatings from aluminum-coated WC powder and Al-WC blends.

166

· Processing and Properties of Advanced Ceramics and Composites

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology

Fig. 9. The macro-hardness of the coatings from copper-coated WC powder and Cu-WC blends. 4. CONCLUSIONS It was demonstrated that WC could be cold sprayed by pre-coating the surface of the carbide particles with commercially pure aluminum or copper. Cold spray deposits using pre-coated carbide particles showed a high amount of retained carbide phase with good dispersion and low porosity. These characteristics yielded hardness values 170% and 21% higher than equivalent deposits produced from Al-WC and Cu-WC blends, respectively. Electroplating of copper on carbide feedstock produced a continuous layer of copper around individual carbide particles, which appeared advantageous for cold spraying. Chemical vapour deposition (CVD) of aluminum on WC feedstock produced discontinuous clusters on WC particle surface and some sub-micron aluminum spheroids. ACKNOWLEDGEMENTS This work was funded by National Research Council of Canada under IRAP# 657343. The metal coating of WC powder were supported and conducted by Federal Technology Group and Advanced Powder Solution Inc. Metallographic inspection was conducted at the Physics Department of the University of Windsor. REFERENCES f 11 K.O. Legg et al."Investigation of Plasma Spray Coatings as an Alternative to Hard Chrome Plating on Internal Surfaces", SERDP Project WP-1151 Final Report, 2006, http://www.serdp.org/Research/upload/WP-1151 -FR.pdf

Processing and Properties of Advanced Ceramics and Composites

· 167

A Novel Method to Spray Tungsten Carbide Using Low Pressure Cold Spray Technology

[2] B. Sartwell et al. "Validation of HVOF WC/Co Thermal Spray Coatings as a Replacement for Hard Chrome Plating on Aircraft Landing Gear" Naval research Lab, NRL/MR/6170—048762, 2004, [3] B. Sartwell et al. "Validation of HVOF WC/Co, WC/CoCr and Tribiology 800 Thermal Spray Coatings as a Replacement for Hard Chrome Plating on C-2/E-2/P-3 and C-130 Propeller Hub System Components", Naval research Lab, 2003, [4] H.L.D. Lovelock, J.M. Benson, P.M. Young, J. Therm. Spray Technol. 7 (1998) 97. [5] B.H. Kear, G. Skandan, R.K. Sadanji, Scr. Mater. 44 (2001) 1703. [6] H. Assadi, F. Gartner, T. Stoltenhoff, H. Kreye , Acta Materialia 51 (2003) 4379. [7] E Irissou, JG Legoux, B Arsenault, C Moreau, J. Therm. Spray Technol. 16 (2007) 661. [8] F.J. Brodmann, "Cold Spray Process Parameters: Powders", The Cold Spray Materials Deposition Process Fundamentals and Applications, Edited by V.K. Champagne, Woodhead Publishing Ltd., 2007, PI 14. [9]. Yih, US Patent # 5911865 and #6010610.

168

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

FOREIGN OBJECT DAMAGE VS. STATIC INDENTATION OXIDE/OXIDE CERAMIC MATRIX COMPOSITE

DAMAGE

IN AN

Sung R. Choi,* Donald J. Alexander, David C. Faucett Naval Air Systems Command, Patuxent River, MD 20670, USA ABSTRACT Static indentation was conducted using an oxide/oxide ceramic matrix composite with 1.59 mm-diameter steel balls. Surface morphology and cross-sections of indents were characterized and compared with those of dynamic impact sites. Many similar features were in common, such as deformation, densification, and the mode of damage generation, despite some difference in severity of those variables. Impact force subjected to foreign-object-damage testing was estimated by using the static indentation data to determine at least a first order of approximation. INTRODUCTION Monolithic ceramics or ceramic matrix composites, because of their inherent nature of brittleness, are susceptible to localized surface damage and/or cracking when subjected to impact by foreign objects. It is also true that ceramic components may fail structurally by soft particles when the kinetic energy of impacting objects exceeds certain limits. The latter case has been often encountered in aeroengines in which combustion products, metallic particles or small foreign objects ingested can cause severe damage to airfoil related components, resulting in serious structural problems. Therefore, foreign object damage (FOD) associated with particle impact needs to be considered when ceramic materials are designed for aeroengine structural applications. In the previous studies [1,2], FOD behavior of two gas-turbine grade monolithic silicon nitrides, AS800 and SN282, was determined. Ceramic target specimens were impacted at their centers by 1.59-mm (diameter) steel ball projectiles in a velocity range from 220 to 440 m/s and their FOD behavior was characterized in terms of impact morphology and post-impact strength. The key material parameter affecting the most FOD was found to be the fracture toughness of a target material. Effect of projectile materials on FOD was also determined in AS800 using four different projectiles of hardened steel, annealed steel, silicon nitride, and brass balls [3]: The hardness of projectile materials was found to be the most important parameter to control the severity of impact damage. The work was extended to gas-turbine grade, melt-infiltrated (MI) Sylramic™ SiC/SiC [4] and oxide/oxide (N720™/Aluminosilicate) [5] ceramic matrix composites (CMCs). Unlike the silicon nitride ceramics, both the MI SiC/SiC and the oxide/oxide CMCs exhibited no catastrophic failure up to 400 m/s and showed greater resistance to FOD than monolithic silicon nitride counterparts [4,5]. The current work is to characterize the static indentation behavior of the oxide/oxide CMC that was used for the FOD work previously [5]. The oxide/oxide specimens were indented using the same 1.59 mm-diameter steel balls. Their responses to static indentation with respect to the deformation and the degree of damage were characterized and compared with the impactassociated damage and deformation. The static indentation data thus obtained were utilized to * Corresponding author; Email:sung.choi [email protected]

171

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

assess the approximated impact force as a function of impact velocity conducted in a previous FOD testing [5]. EXPERIMENTAL PROCEDURES Material The composite material employed in this work was described in detail previously [5]. Briefly, N720™ oxide fibers, produced in tow form by 3M Corp. (Minneapolis, MN), were woven into 2-D 8 harness-satin cloth. The cloth was cut into a proper size, slurry-infiltrated with the matrix (aluminosilicate, AS), and 12 ply-stacked followed by consolidation and sintering. No interface fiber coating was utilized. The fiber volume fraction of the composite panels was about 0.45. Typical microstructure of the composite is shown in Fig. 1. Significant porosity and microcracks in the matrix were typical of the composite. Porosity was about 20-25 %, bulk density was 2.55 g/cm3, flexure strength was 141±4 MPa, and elastic modulus was 67 GPa by the impulse excitation of vibration technique. Flexure bars of about 10 mm in width, 50 mm in length, and about 3 mm in as-furnished thickness were cut from the composite panels. The indent sides of each bar were polished with 600 SiC paper for indentation. Indentation Testing Static indentation testing was performed with an electromechanical test frame (Type 8562, Instron, Canton, MA) using 1.59 mm-diameter, hardened (HRC>60) chrome-steel ball indenters, the same as the ball projectiles that were utilized in the previous FOD work on the oxide/oxide [5]. Indentation loads ranging from 294 N to 2940 N were used, with typically a total of five indents for a given indentation load. Indentation load was applied onto the polished sides of the specimens for about 20 s in load control. In-situ indentation deformation (depth) was determined using an LVDT (Type GT5000, RDP) to obtain load-versus-deformation curves. Indentation morphology of indent sites was examined with SEM. For some selected specimens, their cross-sectional views containing the indent sites were also scrutinized to characterize their respective subsurface damage/deformation. RESULTS AND DISCUSSION Load-versus-Deformation Curves Typical loading/unloading curves obtained at the maximum indentation loads of P=490, 980, and 1960 N are shown in Fig. 2. A completely non-linear behavior of the composite was noted in response to its ball indentation. The fact that the loading curves differed significantly from the unloading curves indicates that the composite material underneath the ball indenter was subjected to considerable densification or compaction. This can be seen more clearly from the unloading/reloading curves obtained from the multiple, continuous runs, as shown in Fig. 3, where several loading/unloading sequences were made with increasing maximum indent loads from 490 to 1960 N. The densified phenomenon of the composite was also observed previously from the specimens subjected to projectile impact [5]. The densification was attributed to the nature of open (porous) and soft structure of the composite. The aspect of densification will be addressed in later sections. Indentation Site Morphology Front impact damages generated in oxide/oxide target specimens were observed to be typically in the form of craters [5]. The similar was true for static indentation. Comparison in

172

· Processing and Properties of Advanced Ceramics and Composites

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

Figure 1. Microstructure of oxide/oxide (N720/AS) ceramic matrix composite used in this work.

Figure 2. Static indentation load vs. indentation displacement curves of an oxide/oxide CMC by 1.59 mm-diameter steel balls (individual runs).

Figure 3. Static indentation load vs. indentation displacement curves of an oxide/oxide CMC by anl .59 mm-diameter steel ball (continuous run with one indenter).

Processing and Properties of Advanced Ceramics and Composites

· 173

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

surface morphology between impact and static indentation is shown in Fig. 4. In case of impact, impact sites at a low impact velocity of 100 m/s were not easily observable; however, when impact velocity was > 200 m/s, impact sites were well developed with their size being dependent on impact velocity. For the case of static indentation, the indent and its damage were well developed at indentation loads of > 980 N. Also, as seen from the figure, an equivalent severity of full surface-damage was likely to be around at 300 m/s and 1960 N, respectively, for impact and static indent. Figure 5 presents the detailed indent morphology around an indent and along the contact boundary, made at an indentation load of 1470 N. It is readily discernable that the central indent region was smooth because of significant compaction under the ball indenter; whereas, the outside indentation boundaries were characterized with fibers/matrices breakage and crushing, due to tension or shearing motion caused by indentation process. More severity of fiber/matrix cracking/damage was typified in impact than in static indentation. Subsurface Morphology Figure 6 shows typical examples of the cross-sectional views of both impact and indent sites. Of a special importance was the aspect of the material beneath the impact or the indenter which showed material's compaction or densification. Note that the area just beneath the impact or the indent does show the disappearing of existing horizontal pores. This compaction can be understandable by considering that the material is soft and open in its microstructure, as noted before. It was observed that in case of impact the compaction increased with increasing impact velocity [5]. It is also noted that neither the projectiles impacted up to 400 m/s nor the ball indenters indented up to 2940 N were flattened or noticeably plastically-deformed. This was again attributed to the composite's 'soft' and open structure, as mentioned before. The steel balls were subjected to severe damage or catastrophic failure when impacted at 400 m/s unto hard and dense target materials such as MI SiC/SiC [4] and monolithic silicon nitrides [1,2]. Examples of microstructural response to static indentation at 1470 N are shown in Fig. 7. The regions A and B represent the areas close to the contact boundary while the region C indicates the central indent area. Note that at Region ' C the fiber tows (the second, horizontal) were subjected to significant fiber breakage. This was due to tension by continuous stretching of the tow by the ball indenter when the indenter continued deforming the composite downward. At the contact boundaries A and B, the tow did also experience considerable tension by accompanying its angled fracture, which is the initiation region of cone cracking. Of particular importance from the cross-sectional views of impact or indent sites was the formation of cone cracks initiating from the impact or the indent site, as also seen from Figs. 6 and 7. Generation of cone cracks by spherical projectiles or indenters has been observed in monolithic silicon nitrides [1-3,6-8] and MI SiC/SiC composite [4] and is typified of many brittle solids including glass under impact [9-11] or static [12] loading. Examples of MI SiC/SiC [4] and AS800 silicon nitride [2] showing cone cracking are presented in Fig. 8. Note that AS800 silicon nitride revealed the contour of a well-developed cone crack from its fracture surface, together with a volcanic-island like cone completely separated from the specimen [2]. AS800 and MI SiC/SiC composite specimens were in 2 mm-thick and subjected to 1.59 mm steel-ball impact at 400 m/s in full (for AS800) or partial (SiC/SiC) support. Regardless of material system, cone cracking is one of the common features of damage associated with spherical impact or static indentation. It has been shown that the plane of the cone cracks is typically a locus of tensile stresses in response to impact or static loading by spherical projectiles or indenters [7].

174

· Processing and Properties of Advanced Ceramics and Composites

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

Figure 4. Comparison in surface damage between impact [5] and static indent in oxide/oxide composite by 1.59-mm steel balls. Arrows indicate impact or indent sites.

Processing and Properties of Advanced Ceramics and Composites

· 175

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

Figure 5. Overall static indent morphology and features along the contact boundaries (A to D) made at 1470 N in an oxide/oxide composite by 1.59-mm steel balls. The markers in B to D represent 100 μιη in size.

176

· Processing and Properties of Advanced Ceramics and Composites

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

Figure. 6. Comparison in cross-section between static indent and impact [5] in an oxide/oxide composite by 1.59 mm-diameter steel balls The dotted lines indicate the contour of cone cracking. The arrows represent indent or impact site.

Figure 7. Cross-sectional view of an indent in oxide/oxide composite made at 1470 N by 1.59 mm-diameter steel ball, (a) Overall; (b) central region C; (c) and (d) indent contact boundaries A and B.

Processing and Properties of Advanced Ceramics and Composites

· 177

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

Figure 8. Cross-sectional views of impact sites showing the occurrence of cone cracking at an impact velocity of 400 m/s for: (a) MI SiC/SiC [4], (b) AS800 silicon nitride (full support) [2], and (c) a cone separated from (b). Arrows indicate the impact sites.

Estimation of Impact Forces As aforementioned, the surface damage generated in specimens either by impact or by static indentation was in the form of indents, or craters, and/or spallation with their size depending on impact velocity or indent load. Figure 9 shows the frontal impact damage size as function of impact velocity, obtained previously [5]. The data on surface indent-damage size as a function of indent load are also presented in the figure. The size of impact damage for the oxide/oxide increased monotonically with increasing impact velocity. The same was true for the static indent particularly at higher indent loads. By contrast, the SiC/SiC composite [4] has exhibited significant rate of increase in damage size with impact velocity, attributed to its hard and dense micro structure. Assuming that the impact event is quasi-static, a prediction of impact force was made using both the impact and the static indent data. The way to predict impact force is illustrated in Fig. 9, starting from the impact associated data and correlating them to the static indent data, thus finding the corresponding indent load, which is the quasi-static impact force of interest. Analytically, the data in Fig. 9 can be formulated through a functional-fit regression analysis to obtain

178

· Processing and Properties of Advanced Ceramics and Composites

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

Figure 9. Data of surface damage on static indentation (a) and on impact (b) [4]. An illustration of the prediction of impact force based on static indent data is shown. The example shows an approximated impact force of 1800 N at an impact velocity of 180 m/s. P^aV

+β

(l)

where P is impact force (N) and Fis impact velocity (m/s). The constants in Eq. (1) were found from the data in Fig. 9 to be a=53S and /?=790. An example of the impact-force prediction at V=200 m/s was shown in the figure, too. It should be noted that the classical elastic Hertzian contact analysis can be hardly applicable to this oxide/oxide material system due to its densified nature. Therefore, the above simplified approach might give at least a first-order approximation of impact force. However, the approach must be validated with actual impact measurements. The experimental work requires a series of information on force, deformation, and dynamics in response to a specific impact event. Also, a model to predict the impact force, although challenging greatly, needs to be developed. In case that the plastic deformation of metallic projectiles upon impacting on hard ceramic targets was significant, the use of contact yield flow stress of the projectiles has produced reasonable agreement on impact force, as shown in soft metallic projectiles such as annealed steel and brass [13]. CONCLUSIONS 1) The mode and aspect of deformation of the oxide/oxide composite by 1.59 mm-diameter steel balls were similar to those in both impact event and static indentation. However, the overall severity of damage/deformation was found to be greater in impact than in static indentation. 2) Due to the composite's soft and open structure, the compaction of the material beneath impact site or indent site occurred with its degree being dependent on impact velocity or on indent load. Also, the formation of cone cracks was typified in either impact or static indentation. 3) Prediction of impact force based on the static indent data yielded at least a first order of approximation although more rigorous estimation of the impact force awaits elaborate experimental techniques.

Processing and Properties of Advanced Ceramics and Composites

· 179

Foreign Object Damage vs. Static Indentation Damage in an Oxide-Oxide Composite

A cknowledgements This work was supported by the Office of Naval Research. REFERENCES 1. (a) S. R. Choi, J. M. Pereira, L. A. Janosik, and R. T. Bhatt, Foreign Object Damage of Two Gas-Turbine Grade Silicon Nitrides at Ambient Temperature, Ceram. Eng. Sci. Proc, 23[3], 193-202 (2002); (b) S. R. Choi et al., Foreign Object Damage in Flexure Bars of Two Gas-Turbine Grade Silicon Nitrides, Mater. Sci. Eng., A 379, 411-419 (2004). 2. (a) S. R. Choi, J. M. Pereira, L. A. Janosik, and R. T. Bhatt, Foreign Object Damage of Two Gas-Turbine Grade Silicon Nitrides in a Thin Disk Configuration, ASME Paper No. GT2003-38544 (2003); (b) S. R. Choi et a l , Foreign Object Damage in Disks of GasTurbine-Grade Silicon Nitrides by Steel Ball Projectiles at Ambient Temperature, J. Mater. Sci., 39, 6173-6182 (2004). 3. (a) S. R. Choi et al., Effect of Projectile Materials on Foreign Object Damage of a GasTurbine Grade Silicon Nitride, ASME Paper No. GT2005-68866 (2005); (b) S. R. Choi et al., Foreign Object Damage in a Gas-Turbine Grade Silicon Nitride by Spherical Projectiles of Various Materials, NASA TM-2006-214330, NASA Glenn Research Center, Cleveland, OH (2006). 4. (a) S. R. Choi, R. T. Bhatt, J. M. Perrira, and J. P. Gyekenyesi, Foreign Object Damage Behavior of a SiC/SiC Composite at Ambient and Elevated Temperatures, ASME Paper No. GT2004-53910 (2004); (b) S. R. Choi, Foreign Object Damage Phenomenon by Steel Ball Projectiles in a SiC/SiC Ceramic Matrix Composite at Ambient and Elevated Temperatures, J. Am. Ceram. Soc, 91 [9] 2963-2968 (2008). 5. S. R. Choi, D. J. Alexander, and R. W. Kowalik, Foreign Object Damage in an Oxide/Oxide Composite at Ambient Temperature, ASME Paper No. GT2008-50505 (2008); also in J. Eng. Gas Turbines & Power, 130 (2008). 6. Y. Akimune, Y. Katano, and K. Matoba, Spherical-Impact Damage and Strength Degradation in Silicon Nitrides for Automobile Turbocharger Rotors, J. Am. Ceram. Soc, 72[8], 1422-1428(1989). 7. A. G. Evans, and T. R. Wilshaw, Dynamic Solid Particle Damage in Brittle Materials: An Appraisal, J. Mater. ScL, 12, 97-116 (1977). 8. A. D. Peralta and H. Yoshida, Ceramic Gas Turbine Component Development and Characterization, van Roode, M, Ferber, M. K., and Richerson, D. W., eds., Vol. 2, pp. 665-692, ASME, New York, NY (2003). 9. S. M. Wiederhorn and B. R. Lawn, Strength Degradation of Glass Resulting from Impact with Spheres,/ Am. Ceram. Soc, 60[9-10], 451-458 (1977). 10. S. M. Wiederhorn and B. R. Lawn, Strength Degradation of Glass Impact with Sharp Particles: I, Annealed Surfaces, J. Am. Ceram. Soc, 62[l-2], 66-70 (1979). 11. C. G. Knight, M. V. Swain, and M. M. Chaudhri, Impact of Small Steel Spheres on Glass Surfaces,/ Mater. Sci., 12, 1573-1586(1977). 12. R. Mouginot and D. Maugis, Fracture Indentation beneath Flat and Spherical Punches, J. Mater. Sci., 20, 4354-4376 (1985). 13. S. R. Choi, Foreign Object Damage Behavior in a Silicon Nitride Ceramic by Spherical Projectiles of Steels and Brass, Mat. Sci. Eng. A497, 160-167 (2008).

180

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

DISTINGUISHED FUNCTIONS MAKING THE BEST USE OF THE UNIQUE COMPOSITE STRUCTURES Toshihiro Ishikawa Inorganic Specialty Products Research Laboratory, Ube Industries, Ltd., 1978-5, Kogushi, Ube, Yamaguchi, 755-8633, Japan ABSTRACT Three types of unique functional ceramics, which were developed on the basis of the techniques preparing thermo-structural ceramics, are going to be introduced. The first one is a tough, thermo-structural ceramic (SA-Tyrannohex), which consists of a highly ordered, close-packed structure of hexagonal columnar SiC-polycrystalline-fibers with a very thin interfacial carbon layer between fibers. SA-Tyrannohex shows excellent high temperature properties and durability up to 1600 °C in air along with excellent toughness. Moreover, this ceramic shows very high thermal conductivity even at temperatures over 1000 °C. Presently, wide applications for this material are developing in many fields (Rocket Nozzle, Nuclear fuel, and so on). The second one is a strong photocatalytic fiber produced from the same raw material for preparing the aforementioned SA-Tyrannohex. This is flourishing in the field of environment as a water-purification system installing the photocatalytic fiber. The third one is a durable, ultraluminous structure for incandescent, high-power white-LED, which consists of binary single crystals entangled with each other continuously. This ceramic was initially developed as an excellent thermo-structural ceramic which can withstand up to 1700 °C in air. By the use of this ceramic, we can provide a new particle- and resin-free system for a light source for illumination. Distinguished functions of these three types of ceramics are caused from each controlled unique structure. INTRODUCTION Many types of polymer-derived ceramic fibers have been developed using a polycarbosilane (-SiH(CH3)-CH2-)n as the starting material [1-4]. Through research, many modifications have been made to obtain much higher heat-resistant ceramic fibers [2-4]. Of these, we developed the highest heat-resistant SiC polycrystalline fiber (Tyranno SA fiber) possessed of the excellent heat-resistance up to 2000 °C in 1998 [3]. In the same year, using this base technology and the same raw material, we developed a sintered silicon-carbide-fiber-bonded ceramic (SA-Tyrannohex), which consists of a highly ordered, close-packed structure of very fine hexagonal columnar fibers with a thin interfacial carbon layer between fibers. This ceramic has both the same heat-resistance as the aforementioned Tyranno SA fiber and excellent tough properties by the existence of the interfacial carbon layer [5]. Other type of functional ceramic fiber with gradient surface structure was also synthesized from the modified polycarbosilane containing an excess amount of selected low-molecular- mass additives.

181

Distinguished Functions Making the Best Use of the Unique Composite Structures

which can be converted into functional ceramics by firing at high temperatures. Thermal treatment of the precursor fiber leads to controlled phase separation ("bleed out") of the low-molecular-mass additives from inside to outside of the precursor fiber. After that, subsequent calcination generates a functional surface layer during the production of bulk ceramic components. By the use of Ti(OC4H9)4 as the aforementioned low-molecular-mass additive, a strong photocatalytic fiber composed of surface gradient structure (Anatase-Ti02) and core structure (S1O2) was developed [6]. This fiber shows high strength (2.5 GPa) and strong photocatalytic activity by irradiation of an ultraviolet light with wavelength shorter than 387 nm (= 3.2 eV). Using this photocatalytic fiber, lots of environmental applications have been expanded. Furthermore, we also have other type of thermo-structural ceramic with excellent high-temperature properties up to 1700 °C in air. This consists of binary single crystals entangled with each other continuously. Making the best use of this unique structure, we newly developed a durable, ultraluminous structure for incandescent, high-power white-LED [7]. In this case, one of these single crystals exhibits effective photoluminescence and the other plays an important role as a light guide of excitation and emitting lights. All of these three types of functional ceramics were developed making the best use of the production process and unique microstructure of our thermo-structural ceramics. In this paper, will appear. THE FIRST TYPE OF FUNCTIONAL CERAMIC: TOUGH, THERMALLY CONDUCTIVE CERAMIC (SA-TYRANNOHEX) An amorphous Si-Al-C-O fiber, which is the starting material for fabricating the SA-Tyrannohex, was synthesized from polyaluminocarbosilane, which was prepared by the reaction of polycarbosilane

(-SiH(CH3)-CH2-)„

with

aluminum(III)-

acetylacetonate.

The

reaction

of

polycarbosilane with aluminum(III)acetylacetonate proceeded at 300 °C in a nitrogen atmosphere through the condensation reaction of Si-H bonds in polycarbosilane and the ligands of aluminum(III)acetylacetonate accompanied by the evolution of acetylacetone. The molecular weight then

increased

due

to

the

cross-linking

reaction

in

the

formation

of

Si-Al-Si

bond.

Polyaluminocarbosilane was melt-spun at 220 °C, and then the spun fiber was cured in air at 153 °C. The cured fiber was continuously fired in inert gas up to 1350 °C to obtain an amorphous Si-Al-C-0 fiber with diameters of about 10 micron meter (87% between 8 and 12 micro meter). This fiber contained a nonstoichiometric amount of excess carbon and oxygen (about llwt%). 8H satin weave fabrics with thicknesses of about 150 micron meter were prepared with the Si-Al-C-0 fiber. Laminated materials, prepared with the fabrics, were hot-pressed at 1900 °C and 50 MPa to obtain the SA-Tyrannohex mainly composed of beta-SiC crystals [5]. During hot-pressing, the amorphous Si-Al-C-0 fiber was converted into a sintered SiC polycrystalline fiber by way of a decomposition, which released CO gas, and a sintering process accompanied by a morphological change from a round columnar shape to a hexagonal columnar shape. In this sintering process, the concentration of

182

· Processing and Properties of Advanced Ceramics and Composites

Distinguished Functions Making the Best Use of the Unique Composite Structures

aluminum in the fiber has to be controlled less than 1 wt%. Such a sintered SiC polycrystalline fiber element (with < 1 wt% Al) showed a densified structure and transcrystalline fracture behavior. The SA-Tyrannohex showed a perfectly close-packed structure of the hexagonal- columnar fibers with a very thin interfacial carbon layer, as can be seen in Fig. 1A. The interior of the fiber element was composed of sintered beta-SiC crystal without an obvious second phase at the grain boundary and triple points. Energy-dispersive x-ray (EDX) spectra taken at these places did not indicate the presence of aluminum within the detectible limit (-0.5 wt%) for the EDX system used. Because of the existence of the very thin interfacial carbon layer, the SA-Tyrannohex exhibited a fibrous fracture behavior and a large amount of fiber pull-out could be observed, as shown in Fig. 1B.

Fig. 1 (A) Cross-section of the SA-Tyrannohex, (B) Fracture surface of the SA- Tyrannohex Accordingly, the SA-Tyrannohex showed nonlinear fracture behavior and relatively high fracture energy (1200 J/m2) compared with monolithic ceramic (for example, 80 J/m2 of silicon nitride) (Fig. 2). This is closely related to the high fiber volume fraction and the existence of a strictly controlled interface. The interfacial carbon layer has a turbostratic layered structure oriented parallel to the fiber surface. The SA-Tyrannohex showed excellent high-temperature properties compared to ordinary SiC-CMCs. The SA-Tyrannohex retained its initial strength up to 1600 °C.

Fig.2 Load-displacement curve of the SA-Tyrannohex at room temperature

Processing and Properties of Advanced Ceramics and Composites

· 183

Distinguished Functions Making the Best Use of the Unique Composite Structures

In general, the high-temperature properties of conventional SiC-CMCs are closely related to the high-temperature strength of the reinforcing fibre. The strength of Hi-Nicalon (the commercial name of SiC fiber produced by Nippon Carbon) gradually decreases with an increase of measuring temperature, even in inert atmosphere; at 1500 °C the strength is about 43% of its low-temperature strength [7]. Accordingly, it has been concluded that the above reduction in the strength of the Hi-Nicalon SiC/SiC is due to the change in the fiber property at high temperatures. However, the sintered SiC-polycrystalline fiber, which composes the aforementioned SA-Tyrannohex, is very stable up to 2000 °C [2], Moreover, the sintered SiC fiber shows negligible stress relaxation up to higher temperatures compared with other representative SiC fibers. Based on these findings, the high-temperature strength of the SA-Tyrannohex is attributed to the high-temperature properties of the fiber element. Researchers have been developing SiC-CMCs in order to obtain an oxidation-resistant, tough thermostructural material. In general, a SiC-based material easily forms a protective oxide layer on its surface at high temperatures in air, leading to the well-known excellent oxidation resistance. The formation of the protective oxide layer proceeds in air according to the following reaction: 2SiC + 30 2 = 2Si0 2 + 2CO In this reaction, the oxygen diffusion through the oxide layer is the rate-determining step. However, at temperatures above 1600 °C in air, considerable vaporization of SiO or S1O2 from the formed oxide layer begins to occur, so that a weight loss of the SiC-based material becomes conspicuous under the above conditions. Accordingly, as long as non-coated SiC-based material is used in air, the upper limit of temperature is at around 1600 °C. The SA-Tyrannohex, which showed no change even at 1900 °C in argon, also showed no marked weight loss up to 1600 °C in air, a temperature at which this material still retains its initial strength. From these findings, the SA-Tyrannohex is found to be furnished with sufficient heat-resistance even in air. The SA-Tyrannohex has potential for use in heat exchangers due to its relatively high thermal conductivity at temperatures above 1000 °C. Figure 4 shows the thermal conductivity of the SA-Tyrannohex in the direction through the thickness and the fiber direction along with other materials including representative SiC/SiC composite (CVI).

184

· Processing and Properties of Advanced Ceramics and Composites

Distinguished Functions Making the Best Use of the Unique Composite Structures

Fig.4 Thermal conductivity temperatures

of SA-Tyrannohex

along with comparative data up to high

Furthermore, SA-Tyrannohex shows the excellent stress-rupture property and

fatigue

property even at 150OC in air compared with previous CMCs. As can be seen from Fig.5, the stress-rupture property of SA-Tyrannohex at 1500°C is almost similar to- t h a t of SiC-CMC using Hi-Nicalon fibre at

1300°C. Moreover, the fatigue resistance of SA-Tyrannohex at 1500°C is

remarkably better t h a n t h a t of the aforementioned SiC-CMC at 1300°C. Because of the high thermal conductivity (about 50 W/mK), SA-Tyrannohex can withstand even if it was directly exposed by a high temperature gas flame over 2000°C. This type of tough ceramic (SA-Tyrannohex) is candidate for such demanding applications as gas turbine engine hot section components, where their high-temperature capability, high thermal conductivity, and low density make it very attractive for replacement of heavy metal super-alloy components. For land-base turbines used for power generation, the possibility of running at higher gas temperatures offers the potential for improved efficiency and reduced nitrous emission. Finally, some near-net shape products and machined products are shown in figure 6.

Fig. 5 Stress-Rupture relationships of SA-Tyrannohex along with the comparative results at high temperatures in air.

Processing and Properties of Advanced Ceramics and Composites

· 185

Distinguished Functions Making the Best Use of the Unique Composite Structures

Fig. 6 Several kinds of production made of SA-Tyrannohex

THE SECOND TYPE OF FUNCTIONAL CERAMIC: STRONG PHOTOCATALYTIC FIBER Polytitanocarbosilane containing an excess amount of titanium alkoxide was synthesized by the mild reaction of polycarbosilane (-SiH(CH3)-CH2-)n (20kg) with titanium (IV) tetra-n-butoxide (20kg) at 220 °C in nitrogen atmosphere. The obtained precursor polymer was melt-spun at 150 °C continuously using melt-spinning equipment. The spun fiber, which contained excess amount of unreacted titanium alkoxide, was pre-heat-treated at 100 °C and subsequently fired up to 1200 °C in air to obtain continuous, transparent fiber (diameter: 4-7 micron). In the initial stage of the pre-heat-treatment, effective bleeding of the excess amount of unreacted titanium (IV) tetra-n-butoxide from the spun fiber occurred to form the surface gradient layer containing large amount of titanium (IV) tetra-n-butoxide. During the next firing process, the pre-heat-treated precursor fiber was converted into a titania-dispersed, silica-based fiber covered with gradient titania. The fundamental concept of the new production process for our photocatalytic fiber is shown in Fig.7.

Fig.7 Fundamental concept of our photocatalytic fiber Figure 8 shows the surface appearance and the cross section of our photocatalytic fiber. As can be seen from this figure, the surface of the fiber is densely covered with nanoscale anatase-Ti02 particles (8 nm), which are strongly sintered with each other directly or through with amorphous silica phase. The thickness of the surface T1O2 layer is approximate 100-200 nm. The tensile strength of this

186

· Processing and Properties of Advanced Ceramics and Composites

Distinguished Functions Making the Best Use of the Unique Composite Structures

fiber as measured by a single filament method was 2.5GPa on the average using an Orientec UTM-20 with a gauge length of 25 mm and cross-head speed of 2 mm/min.

Fig.8 The surface and cross-section of our photocatalytic fiber We developed a water-purifier for pollutants using a felt material made of the aforementioned photocatalytic fiber. The average intensity of UV light on the photocatalytic fibers should be over 10mW/cm2 for obtaining good decomposition activity. By the way, the photocatalytic activity is attributed to the strong oxidant (hydroxyl radical), which generates at the surface of the T1O2 crystals by the irradiation of UV light. Accordingly, any kind of organic pollutant, which contacts the surface of the photocatalytic fiber irradiated by UV light, can be decomposed into CO2 and water finally. The aforementioned water-purifier is widely applicable in the field of environment, and already commercialized. Presently, in the case of our largest water-purifier, the treatment capacity is 80t/hour. THE THIRD TYPE OF FUNCTIONAL CERAMIC: DURABLE, ULTRALUMINOUS STRUCTURE FOR WHITE-LED Recently, much interest has been shown in the use of white light-emitting diodes (white-LEDs) as a light source for illumination. Among these, promising white-LEDs are composed of blue- or near-ultraviolet-LED chips and phosphor particles dispersed in a resinous matrix. However, insofar as phosphor particles are used, there is no way to avoid a detrimental effect on luminous efficacy caused by a back scattering on the surface of the phosphor particles and a deterioration of the resin. To create white light using blue- or near-ultraviolet-LED (light emitting diode) chips, mixing of three primary colors (red, green and blue) or complementary colors, which are emitted from phosphor particles dispersed in the transparent resin by irradiation of an excitation light from the aforementioned LED, is generally performed. These combinations are roughly divided into two types of system. The first type is the combination of a blue-LED chip and a phosphor particle, which can

Processing and Properties of Advanced Ceramics and Composites

· 187

Distinguished Functions Making the Best Use of the Unique Composite Structures

radiate yellow light by irradiation of blue light. Most types of this system adopt a LED chip coated with a resin containing the aforementioned phosphor particle. In this system, the blue light passing through the particle-dispersed resin should be effectively mixed with yellow light at a constant ratio for obtaining relevant white light, so that it would be difficult to increase the content of the phosphorous particle at random. That is to say, in order to avoid the decrement of objective white light and secure both lights in a constant ratio, the particle content and the thickness of the resin have to be strictly controlled. The second type is the combination of a nearultraviolet-LED chip and three types of phosphor particle, which can radiate each of the three primary colors. In this case, as all of the near-ultraviolet light has to be effectively used for the excitation, a relatively high content of phosphor particles should be adopted compared with the former system using the blue-LED chip. Accordingly, the aforementioned back-scattering on the surface of particles also has to be seriously taken into consideration. Of course, many high power white-LEDs using phosphor particles dispersed in resins have been developed thus far. However, in such cases there are serious problems regarding the aforementioned backscattering of light and a deterioration of resins. Recently, to avoid the scattering of light, a nano-sized very fine fluorescent powder has been developed. However, generally in the case of fine particle smaller than sub-micrometer scale, a distinct decrease in fluorescence caused by the surface defects of the particle was observed. We have addressed the issue of establishing a widely applicable process for creating a scattering-restrained, durable and resin-free system composed of transparent, binary single crystals entangled with each other continuously and without any secondary phases between both single crystals. A schematic representative of our new construction, in which the continuous fluorescent single crystal (single crystal B) exists in another continuous transparent single crystal (single crystal A) in a semicoherent state as if a block copolymer, is shown in Fig.9 along with the conventional particle-dispersed system using a resin.

188

· Processing and Properties of Advanced Ceramics and Composites

Distinguished Functions Making the Best Use of the Unique Composite Structures

Fig.9 Schematic diagram of our new binary crystalline system for realizing incandescent, high-power white-LED along with comparative information regarding conventional particle-dispersed system The important feature of our concept is that the fluorescent crystalline phase does not independently exist as a mere dispersed phase, but directly connects with another transparent crystalline phase, which plays an important role as a light guide, to create a continuous phase as if a block copolymer. Both crystalline phases interpenetrate without grain boundaries, so that most of the lattices of those crystals are continuous. Accordingly, each crystal ought to be markedly affected by the crystalline structure of the other. This means that both crystals are fundamentally different from the single-phase crystals existing independently. Our type of crystal should be newly classified as a single material as well as a block copolymer. In our new concept, both crystalline phases, whose refractive indexes are very close, are complexly entangled with each other and propagate continuously from bottom to top to provide a good transmission and effective mixing of the objective lights. Even if total reflection and reflection occur, all of the objective lights can be taken out without a marked decrement in inside of the crystal. On the other hand, in the case of a conventional particle-dispersed system, generally the refractive indexes of the particle and matrix are markedly different from each other. Accordingly, the backscattering of the objective lights on the surface of particles and a total reflection of the fluorescence emitted from inside of the phosphor particles remarkably occur. These phenomena cause the relatively low emission level. As an actual instance for our new concept, we prepared a unidirectionally solidified Al2O3ZCeo.09Y2.91 AI5O12 binary crystalline system. For preparing this binary

Processing and Properties of Advanced Ceramics and Composites

· 189

Distinguished Functions Making the Best Use of the Unique Composite Structures

crystal, we referred to a synthesis method of a eutectic composite consisting of single-crystal AI2O3 and single-crystal YAG [8]. The combination of purple LED chips and our Al203/Ceo.o9Y2.9iAl5Oi2 plate generates an incandescent white light caused by a good mixing of purple light passing through the AI2O3 phase and yellow light emitted from the AkCVCeo.^Y^iAlsO^ phase as shown in Fig. 10. In this case, four purple LED chips (0.3 mm x 0.3 mm) are attached on the A^CVCeo^Yz 91AI5O12 plate (2 mm x 8 mm). And this plate is covered with the same compositional plate. By the use of our binary single crystal, very simple and incandescent illumination systems can be created, which may take the place of the particle-dispersed system. This type of binary single crystal shows an excellent environmental resistance up to 1700 °C in air for a long time, and then can play an important role as a steady illumination component for future high power white LEDs. Furthermore, our new concept can be applied to wide compositions by changing the combination of the binary system.

Fig. 10 Demonstration of actual illumination using our Al2O3/Ce0.09Y2.9iAl5012 binary crystal plate (2 mm x 8 mm) attaching four purple-LED chips (0.3 mm x 0.3 mm, 407 nm). CONCLUSION Three types of unique functional ceramic were introduced. All of these were developed making the best use of the production process and microstructure of excellent thermostructural ceramics. The first functional ceramic was a tough, thermally conductive ceramic (SA-Tyrannohex), which consists of a highly ordered, close-packed structure of very fine hexagonal columnar fibers with a thin interfacial carbon layer between fibers. The fiber element showed excellent heat-resistance up to 2000°C, so that the aforementioned SA-Tyrannohex showed excellent high-temperature properties and very large fracture energy coursed by the existence of the interfacial carbon layer. The second one was the strong photocatalytic fiber with the surface gradient titania layer. This fiber was synthesized

190

· Processing and Properties of Advanced Ceramics and Composites

Distinguished Functions Making the Best Use of the Unique Composite Structures

making the best use of both a precursor method for preparing a thermostructural ceramic fiber and a breeding phenomenon. The third one was a new particle- and resin-free system composed of transparent, binary single crystals entangled with each other continuously. This ceramic can play an important role as a light source for illumination. Of course, this ceramic was also a thermostructural ceramic, which can withstands up to 1700°C in air for long time. REFERENCES ]

S. Yajima, Y. Hasegawa, K. Okamura, T. Matsuzawa, "Development of High Tensile Strength Silicon

Carbide Fiber Using an Organosilicon Polymer Precursor", Nature, 273, 525-527 (1978). 2

M. Takeda, J. Sakamoto, A. Saeki, Y. Imai, and H. Ichikawa, "High Performance Silicon Carbide

Fiber Hi-Nicalon for Ceramic Matrix Composites", Ceram.Eng.Sci.Proc, 16[4], 37-44 (1995). 3

T. Ishikawa, Y. Kohtoku, K. Kumagawa, T. Yamamura & T. Nagasawa, "High-Strength

Alkali-Resistant Sintered SiC Fibre Stable to 2200 oC", Nature, 391, 773-775 (1998). 4

J. Lipowitz, J. A. Rabe, G. A. Zank, A. Zangvil, Y. Xu, "Structure and Prooperties of Sylramic TM

Silicon Carbide Fiber - A Polycrystalline, Stoichiometric SiC Composition", Ceram.Eng.Sci.Proc, 18[3], 147-157(1997). 5

T. Ishikawa, S. Kajii S, K. Matsunaga, T. Hogami, Y. Kohtoku, T. Nagasawa, "A Tough, Thermally

Conductive Silicon Carbide Composite with High Strength up to I6OO0C in Air", Science, 282, 1295-1297 (1998). 6

T. Ishikawa, H. Yamaoka, Y. Harada, T. Fujii, T. Nagasawa, "A general process for in situ formation

of functional surface layers on ceramics", Nature, 416, 64-67 (2002). 7

T. Ishikawa, S. Sakata, A. Mitani, "Durable, Ultraluminous structure for Incadescent, High-Power

White-LED", Int. J. Appl. Ceram. Technoi, 3[2], 144-149 (2006). 8

Y. Waku, H. Otsubo, N. Nakagawa, and Y. Kohtoku, "Sapphire matrix composites reinforced with

single crystal YAG phase," JMater.Sci., 31, 4663-4670 (1996).

Processing and Properties of Advanced Ceramics and Composites

· 191

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

EFFECTS OF ENVIRONMENT ON CREEP BEHAVIOR OF NEXTEL™720/ALUMINAMULLITE CERAMIC COMPOSITE AT 1200°C C. L. Genelin and M. B. Ruggles-Wrenn* Air Force Institute of Technology Wright-Patterson Air Force Base, Ohio 45433-7765 ABSTRACT The tensile creep behavior of an oxide-oxide continuous fiber ceramic composite was investigated at 1200 °C in laboratory air, in steam and in argon. The composite consists of a porous alumina-mullite matrix reinforced with laminated, woven mullite/alumina (Nextel™720) fibers, has no interface between the fiber and matrix, and relies on the porous matrix for flaw tolerance. The tensile stress-strain behavior was investigated and the tensile properties measured at 1200 °C. The elastic modulus was 74.5 GPa and the ultimate tensile strength was 153 MPa. Tensile creep behavior was examined for creep stresses in the 70-140 MPa range. Primary and secondary creep regimes were observed in all tests. Creep run-out (set to 100 h) was achieved in laboratory air for creep stress levels < 90 MPa. The presence of either steam or argon accelerated creep rates and reduced creep lifetimes. Composite micro structure, as well as damage and failure mechanisms were investigated. INTRODUCTION Advances in power generation systems for aircraft engines, land-based turbines, rockets, and, most recently, hypersonic missiles and flight vehicles have raised the demand for structural materials that have superior long-term mechanical properties and retained properties under high temperature, high pressure, and varying environmental factors, such as moisture1. Typical components include combustors, nozzles and thermal insulation. Ceramic-matrix composites (CMCs), capable of maintaining excellent strength and fracture toughness at high temperatures are prime candidate materials for such applications. Additionally, lower densities of CMCs and their higher use temperatures, together with a reduced need for cooling air, allow for improved high-temperature performance when compared to conventional nickel-based superalloys2. Advanced reusable space launch vehicles will likely incorporate fiber-reinforced CMCs in critical propulsion components3. Because these applications require exposure to oxidizing environments, the thermodynamic stability and oxidation resistance of CMCs are vital issues. The main advantage of CMCs over monolithic ceramics is their superior toughness, tolerance to the presence of cracks and defects, and non-catastrophic mode of failure. It is widely accepted that in order to avoid brittle fracture behavior in CMCs and improve the damage tolerance, a weak fiber/matrix interface is needed, which serves to deflect matrix cracks and to allow subsequent fiber pullout4"7. Historically, following the development of SiC fibers, fiber coatings such as C or BN have been employed to promote the desired composite behavior. However, the non-oxide fiber/non-oxide matrix composites generally show poor oxidation resistance8,9, particularly at intermediate temperatures (~800 °C). These systems are susceptible to embrittlement due to oxygen entering through the matrix cracks and then reacting with the interphase and the fibers10"1 . The degradation, which involves oxidation of fibers and fiber " Corresponding author The views expressed are those of the authors and do not reflect the official policy or position of the United States Air Force, Department of Defense or the U. S. Government.

193

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

coatings, is typically accelerated by the presence of moisture14"20. Using oxide fiber/ non-oxide matrix or non-oxide fiber/oxide matrix composites generally does not substantially improve the high-temperature oxidation resistance21. The need for environmentally stable composites motivated the development of CMCs based on environmentally stable oxide constituents22"30. More recently it has been demonstrated that similar crack-deflecting behavior can also be achieved by means of a finely distributed porosity in the matrix instead of a separate interface between matrix and fibers31. This microstructural design philosophy implicitly accepts the strong fiber/matrix interface. It builds on the experience with porous interlayers as crack deflection paths32,33 and extends the concept to utilize a porous matrix as a surrogate. The 9 9 "}f\ ^Π *\ά. ^Χ

concept has been successfully demonstrated for oxide-oxide composites ' ' " . Resulting oxide/oxide CMCs exhibit damage tolerance combined with inherent oxidation resistance. However, due to the strong bonding between the fiber and matrix, a minimum matrix porosity is needed for this concept to work39. An extensive review of the mechanisms and mechanical properties of porous-matrix CMCs is given elsewhere40. Porous-matrix oxide/oxide CMCs exhibit several behavior trends that are distinctly different from those exhibited by traditional non-oxide CMCs with a fiber-matrix interface. For the non-oxide CMCs, fatigue is significantly more damaging than creep. Zawada et al41 examined the high-temperature mechanical behavior of a porous matrix Nextel610/ Aluminosilicate composite. Results revealed excellent fatigue performance at 1000 °C. Conversely, creep lives were short, indicating low creep resistance and limiting the use of that CMC to temperatures below 1000 °C. Ruggles-Wrenn et al42 showed that Nextel™720/Alumina (N720/A) composite exhibits excellent fatigue resistance in laboratory air at 1200 °C. The fatigue limit (based on a run-out condition of 105 cycles) was 170 MPa (88% UTS at 1200 °C). Furthermore, the composite retained 100% of its tensile strength. However, creep loading was found to be considerably more damaging. Creep run-out (defined as 100 h at creep stress) was achieved only at stress levels below 50% UTS. Creep performance at 1200 °C was further degraded by the presence of steam. Mehrman et al4' demonstrated that introduction of a short hold period at the maximum stress into the fatigue cycle significantly degraded the fatigue performance of N720/A composite at 1200 °C in air. In steam, superposition of a hold time onto a fatigue cycle resulted in an even more dramatic deterioration of fatigue life, reducing it to the much shorter creep life at a given applied stress. Because creep was shown to be considerably more damaging than cyclic loading to oxideoxide CMCs with porous matrix41"43, high-temperature creep resistance remains among the key issues that must be addressed before using these materials in advanced aerospace applications. In addition, prior studies42'44 revealed that creep performance of the oxide-oxide CMCs deteriorates drastically in the presence of steam. This study aims to evaluate the creep behavior of Nextel™720/Alumina-Mullite (N720/AM), an oxide-oxide CMC with a porous matrix, at 1200 °C in air and in steam. Creep tests were conducted in air and in steam at stress levels ranging from 73 to 136 MPa. Results reveal that test environment has a noticeable effect on creep life. The composite microstructure, as well as damage and failure mechanisms are discussed. MATERIAL AND EXPERIMENTAL ARRANGEMENTS The material studied was Nextel™720/Alumina-Mullite (N720/AM), an oxide-oxide ceramic composite consisting of a porous alumina-mullite matrix reinforced with Nextel™720 fibers. The composite, manufactured by COI Ceramics (San Diego, CA), was supplied in a form of a 3.2 mm thick plate, comprised of 12 0°/90° woven layers, with a density of ~2.63 g/cm3 and

194

· Processing and Properties of Advanced Ceramics and Composites

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

a fiber volume of approximately 40.4%. Composite porosity was -26.8%. The laminate was fabricated following the procedure similar to that described elsewhere45. No coating was applied to the fibers. The damage tolerance of the N720/AM composite is enabled by a porous matrix. A servocontrolled MTS mechanical testing machine equipped with hydraulic water-cooled wedge grips, a compact two-zone resistance-heated furnace, and two temperature controllers was used in all tests. An MTS TestStar II digital controller was employed for input signal generation and data acquisition. Strain measurement was accomplished with an MTS high-temperature aircooled uniaxial extensometer of 12.5-mm gage length. Tests in steam environment employed an alumina susceptor (tube with end caps), which fits inside the furnace. The specimen gage section is located inside the susceptor, with the ends of the specimen passing through slots in the susceptor. Steam is introduced into the susceptor (through a feeding tube) in a continuous stream with a slightly positive pressure, expelling the dry air and creating a near 100% steam environment inside the susceptor. For elevated temperature testing, thermocouples were bonded to test specimens to calibrate the furnace on a periodic basis. The furnace controllers (using noncontacting thermocouples exposed to the ambient environment near the test specimen) were adjusted to determine the power settings needed to achieve the desired temperature of the test specimen. The determined power settings were then used in actual tests. The power setting for testing in steam environment was determined by placing the specimen instrumented with thermocouples in steam and repeating the furnace calibration procedure. Thermocouples were not bonded to the test specimens after the furnace was calibrated. All tests were performed at 1200 °C using dog bone shaped specimens of 152 mm total length with a 10-mm-wide gage section. All specimens used in this study were cut from a single plate. In all tests, a specimen was heated to test temperature in 25 min, and held at temperature for additional 15 min prior to testing. Tensile tests were performed in displacement control with a constant rate of 0.05 mm/s in laboratory air. Creep-rupture tests were conducted in load control in accordance with the procedure in ASTM standard C 1337 in laboratory air and in steam. In all creep tests the specimens were loaded to the creep stress level at the stress rate of 15 MPa/s. Creep run-out was defined as 100 h at a given creep stress. In each test, stress-strain data were recorded during the loading to the creep stress level and the actual creep period. Thus both total strain and creep strain could be calculated and examined. All specimens that achieved a run-out were subjected to tensile test to failure at 1200°C to determine the retained strength and stiffness. Fracture surfaces of failed specimens were examined using SEM (FEI Quanta 200 HV) as well as an optical microscope (Zeiss Discovery VI2). The SEM specimens were carbon coated. RESULTS AND DISCUSSION Monotonic Tension Tensile stress-strain behavior at 1200 °C is nearly linear to failure. The average ultimate tensile strength (UTS) was 153 MPa, elastic modulus, 74.5 GPa, and failure strain, 0.34 %. The tensile properties and stress-strain behavior are similar to those exhibited by the N720/Alumina composite at 1200 °C42. It is worthy of note that in all tests reported herein, the failure occurred within the gage section of the extensometer. Creep-Rupture Results of the creep-rupture tests are summarized in Table I, where creep strain accumulation and rupture time are shown for each creep stress level and test environment. Creep

Processing and Properties of Advanced Ceramics and Composites

· 195

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

curves obtained in air and in steam are shown in Fig. 1. Creep curves produced in all tests conducted in air exhibit primary and secondary creep regimes, but no tertiary creep. Transition from primary to secondary creep occurs early in creep life, primary creep persists during the first -10 h of the creep test. Creep strain decreases as the applied stress increases from 73 to 136 MPa. Creep strains accumulated at stresses < 114 MPa considerably exceed the failure strain obtained in the tension test. Table I. Summary of creep-rupture results for the N720/AM ceramic composite at 1200 C in laboratory air and in steam Test Environment Creep Stress (MPa) Creep Strain (%) Time to Rupture (h) 73 Air 0.60 100a Air 100a 91 0.59 114 Air 0.47 22.3 Air 136 0.28 0.59 Steam 73 2.49 37.0 Steam 1.57 91 4.18 Steam 114 0.48 0.38 Steam 136 0.01 0.11 Run-out.

Figure 1. Creep strain vs time curves for N720/Alumina-Mullite composite at 1200 °C in air and in steam: (a) at 73 and 91 MPa and (b) at 114 and 136 MPa. The test environment appears to have little influence on the appearance of the creep curves obtained at stresses > 91 MPa, where only primary and secondary creep regimes are observed. In contrast, the creep curve obtained at 73 MPa in steam shows primary, secondary and tertiary creep. The secondary creep transitions to tertiary creep after only ~ 10 h. Moreover, the test environment has a significant effect on creep strains accumulated at stresses < 91 MPa. At 73 MPa, the creep strain accumulated in steam is 4 times that accumulated in air. At 91 MPa, the presence of steam increases the creep strain almost by a factor of 3. Minimum creep rate was reached in all tests. Creep rate as a function of applied stress is presented in Fig. 2, where results for N720/Alumina composite from prior work are included for comparison. In order to facilitate comparison between results obtained for N720/AM specimens with fiber volume fraction of 0.40 and those obtained for N720/A specimens with fiber volume fraction of 0.44, all creep stress levels in Fig. 2 were adjusted for Vf = 0.40. In air

196

· Processing and Properties of Advanced Ceramics and Composites

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

the minimum creep rate of N720/AM increases by approximately two orders of magnitude as the creep stress increases from 73 MPa to 136 MPa. It is seen that in air, the secondary creep rate of the N720/AM composite is approximately an order of magnitude lower than that of the N720/A CMC for a given creep stress. The N720/AM creep rates increase dramatically in steam. For a given creep stress, the creep rate in steam is nearly two orders of magnitude higher than that in air. In steam, the N720/AM creep rates are slightly lower than those observed for N720/A composite, especially for applied stress level < 114 MPa. 1.0E-04 ■ N720/AM,Air • N720/AM, Steam

1.0E-05

DN720/A,Air

~

ON720/A, Steam

1.0E-06

cc .E 1.0E-07

E

(Λ

1.0E-08 1.0E-09 40

50

60

70

80 90100

200

Max Stress (MPa)

Figure 2. Minimum creep rate as a function of applied stress for N720/AM ceramic composite at 1200°C in air and in steam. Data for N720/A CMC from Ruggles-Wrenn et al.44 are also shown. All data are adjusted for Vf = 0.40.

200

Q. 150 5

O

□ N720/AM, Air

■ N720/A, Air

Δ Ν720/ΑΜ, Argon

A N720/A, Argon

O N720/AM, Steam

· N720/A, Steam

·

Δ

c/>

■ •

(A

DA Δ>

E A

•

· O

Δ ■

9 »

I

ω 100 50

T = 1200 °C I

1.E-03

lllll»!

I llllim

1.E-02

1.E-01

Ml

I

I I llll

1.E+00

1.E+01

I

I

1.E+02

lllllJ

1.E+03

Time (h)

Figure 3. Creep stress vs rupture time for N720/AM ceramic composite at 1200°C in air and in steam. Data for N720/A CMC from Ruggles-Wrenn et al.44 All data are adjusted for Vf = 0.40. Stress-rupture behavior is summarized in Fig. 3, where results for N720/A composite from prior work44 are also included. Ail creep stress levels in Fig. 3 were adjusted for Vf = 0.40. As expected, creep life decreases with increasing applied stress. In air, the creep run-out stress for

Processing and Properties of Advanced Ceramics and Composites

· 197

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

N720/AM was 91 MPa. The presence of steam drastically reduced the creep lifetimes of N720/AM composite. The reduction in creep life due to steam was > 95% for applied stress levels > 91 MPa, and 63% for the applied stress of 73 MPa. In steam, creep run-out was not achieved. It is notable that in air and in steam the creep lifetimes of N720/AM were similar to those of the N720/A composite. Table II. Retained properties of the N720/AM CMC subjected to prior creep at 1200 °C in air Creep Stress I Retained Strength Retained Modulus Strain at (MPa) Strength (MPa) Retention (%) Modulus (GPa) Retention (%) Failure (%) 73 169 110 70.1 94 0.35 9~\ Ϊ67 Ϊ09 673 9\ 034 Retained strength and modulus of the specimens that achieved creep run-out in air are summarized in Table II. Tensile stress-strain curves obtained for the specimens subjected to prior creep are presented in Fig. 4 together with the tensile stress-strain curve for the as-processed material. Prior creep appears to have a beneficial effect on tensile strength. The strength of the N720/AM specimens subjected to 100 h of prior creep in air was -10% higher than the UTS of the untested material. Nevertheless, a reduction in modulus was observed. Modulus loss due to prior creep at 73 MPa was 6%, and modulus loss due to creep at 91 MPa was 9%. 250 200

100 hat 91 MPa, Air

\

S. 150 Σ

o

100 hat 73 MPa, Air I

100

CO

50 0 0.0

0.1

0.2

0.3

0.4

0.5

Strain (%)

Figure 4. Effects of prior creep at 1200°C in air on tensile stress-strain behavior of N720/AM. Composite Microstructure Optical micrographs of the fracture surfaces of specimens tested in creep at 73 and 136 MPa are shown in Fig. 5. The N720/AM specimens tested at 73 MPa (Figs. 5a and 5b) show a somewhat greater degree of uncorrelated fiber fracture than those tested at 136 MPa (Figs. 5c and 5d). Furthermore, specimens tested at 73 MPa have noticeably longer damage zones. Notably the test environment appears to have little effect on the fracture surface topography. For a given creep stress, fracture surfaces obtained in steam are similar to those obtained in air. A previous study46 revealed that for N720/A composite tested at 1200 °C in air and in steam, the fracture surface appearance could be correlated with the failure time. Predominantly planar fracture surface corresponds to a short life, while fibrous fracture indicates longer life. In

198

· Processing and Properties of Advanced Ceramics and Composites

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

the case of N720/A composite, the near planar fracture surfaces were attributed to matrix densification and subsequent loss of matrix porosity, which resulted in decreased damaged tolerance. In contrast, the SEM micrographs of the fracture surfaces of the N720/AM specimens tested in creep at 1200 °C in air and in steam (Fig. 6) show that for N720/AM the fracture surface appearance cannot be directly correlated with the creep lifetime. All fracture surfaces in Fig. 6 are dominated by planar regions of coordinated fiber failure. Compare the fracture surface of the specimen tested at 73 MPa in air (Fig. 6a) and that of the specimen tested at 136 MPa in steam (Fig. 6d). The fracture surfaces in Figs. 6a and 6d have essentially the same topography. Yet the specimen in Fig. 6a achieved a 100-h creep run-out and failed in a subsequent tensile test, while the specimen in Fig. 6d failed after mere 36 s of creep.

Figure 5. Optical micrographs of the fracture surfaces of specimens tested in creep at 1200 °C: (a) at 73 MPa in air, (b) at 73 MPa in steam, (c) at 136 MPa in air, and (d) at 136 MPa in steam. CONCLUDING REMARKS The tensile stress-strain behavior of the N720/AM composite was investigated and the tensile properties measured at 1200 °C. The elastic modulus was 74.5 GPa and the UTS was 153 MPa. The creep-rupture behavior of the N720/AM composite was characterized at 1200°C in air and in steam for creep stress levels ranging from 73 to 136 MPa. In air the N720/AM composite exhibits primary and secondary creep regimes. Creep strains accumulated at stresses < 114 MPa

Processing and Properties of Advanced Ceramics and Composites

· 199

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

considerably exceed the failure strain obtained in the tension test. In steam, primary, secondary and tertiary creep regimes are observed. The creep strains accumulated at stresses < 91 MPa in steam are significantly larger than those produced in air.

Figure 6. Fracture surfaces obtained in creep tests at 136 MPa conducted at 1200 °C: (a-b) in air, and (c-d) in steam. Minimum creep rate was reached in all tests. Creep strain rates range from 9.2 x 10" to 8.7 x 10"7 s"1 in air. The presence of steam accelerates creep rates of N720/A by nearly two orders of magnitude. Creep run-out of 100 h was achieved at applied stress levels < 91 MPa in air. The run-out specimens exhibited an increase in strength, but stiffness loss of up to 9% was observed. The presence of steam dramatically reduced creep lifetimes. The reduction in creep lifetime due to steam was ~ 63% at 73 MPa and ~ 98% at 136 MPa. The N720/AM fracture surfaces obtained at 1200 °C are dominated by regions of planar fracture. The near-planar fracture surfaces suggest the loss of matrix porosity and subsequent matrix densification due to additional sintering. The fracture surface appearance cannot be directly correlated with the creep lifetime.

200

· Processing and Properties of Advanced Ceramics and Composites

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

REFERENCES Έ . Zok, "Developments in Oxide Fiber Composites," J. Am. Ceram. Soc, 89(11), 33093324 (2006). L. P. Zawada, J. Staehler, S. Steel, "Consequence of Intermittent Exposure to Moisture and Salt Fog on the High-Temperature Fatigue Durability of Several Ceramic-Matrix Composites," J. Am. Ceram. Soc, 86(8), 1282-1291 (2003). 3 S. Schmidt, S. Beyer, H. Knabe, H. Immich, R. Meistring, A. Gessler, "Advanced Ceramic Matrix Composite Materials for Current and Future Propulsion Technology Applications," Acta Astronautica, 55, 409-420 (2004). 4 R. J. Kerans, R. S. Hay, N. J. Pagano, T. A. Parthasarathy, "The Role of the Fiber-Matrix Interface in Ceramic Composites," Am. Ceram. Soc. Bull., 68(2), 429-442 (1989). 5 A. G. Evans, F. W. Zok, "Review: the Physics and Mechanics of Fiber-Reinforced Brittle Matrix Composites," J. Mater. Sci., 29:3857-3896 (1994). 6 R. J. Kerans, T. A. Parthasarathy, "Crack Deflection in Ceramic Composites and Fiber Coating Design Criteria," Composites: Part A, 30, 521-524 (1999). 7 R. J. Kerans, R. S. Hay, T. A. Parthasarathy, M. K. Cinibulk, "Interface Design for Oxidation-Resistant Ceramic Composites," J. Am. Ceram. Soc, 85(11), 2599- 2632 (2002). 8 K. M. Prewo, J. A. Batt, "The Oxidative Stability of Carbon Fibre Reinforced Glass-Matrix Composites," J. Mater. Sci., 23, 523-527 (1988). *T. Mah, N. L. Hecht, D. E. McCullum, J. R. Hoenigman, H. M. Kim, A. P. Katz, H. A. Lipsitt, "Thermal Stability of SiC Fibres (Nicalon)," J. Mater. Sci., 19, 1191-1201 (1984). ,0 J. J. Brennan. Fiber Reinforced Ceramic Composites; Ch. 8. K.C. Masdayazni, editor. Noyes, New York, 1990. n F . Heredia, J. McNulty, F. Zok, A. G. Evans, "Oxidation Embrittlement Probe for Ceramic Matrix Composites," J. Am. Ceram. Soc, 78(8), 2097-2100 (1995). 12 R. S. Nutt, "Environmental Effects on High-Temperature Mechanical Behavior of Ceramic Matrix Composites," High-Temperature Mechanical Behavior of Ceramic Composites. S. V. Nair, and K. Jakus, editors. Butterworth-Heineman, Boston, MA, (1995). 13 A. G. Evans, F. W. Zok, R. M. McMeeking, Z. Z. Du, "Models of High-Temperature Environmentally-Assisted Embrittlement in Ceramic Matrix Composites," J. Am. Ceram. Soc, 79,2345-52(1996). 14 K. L. More, P. F. Tortorelli, M. K. Ferber, J. R. Keiser, "Observations of Accelerated Silicon Carbide Recession by Oxidation at High Water-Vapor Pressures," J. Am. Ceram. Soc, 83(1), 11-213(2000). 15 K. L. More, P. F. Tortorelli, M. K. Ferber, L. R. Walker, J. R. Keiser, W. D. Brentnall, N. Miralya, J. B. Price, "Exposure of Ceramic and Ceramic-Matrix Composites in Simulated and Actual Combustor Environments," Proceedings of International Gas Turbine and Aerospace Congress, Paper No. 99-GT-292 (1999). 16 M. K. Ferber, H. T. Lin, J. R. Keiser, "Oxidation Behavior of Non-Oxide Ceramics in a High-Pressure, High-Temperature Steam Environment," Mechanical, Thermal, and Environmental Testing and Performance of Ceramic Composites and Components. M. G. Jenkins, E. Lara-Curzio, and S. T. Gonczy, editors. ASTM STP 1392, 210-215 (2000). 17 J. A. Haynes, M. J. Lance, K. M. Cooley, M. K. Ferber, R. A. Lowden, D. P. Stinton, "CVD Mullite Coatings in High-Temperature, High-Pressure Air-H 2 0," J. Am. Ceram. Soc, 83(3), 657-659 (2000).

Processing and Properties of Advanced Ceramics and Composites

· 201

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

1

E. J. Opila, R. E. Hann Jr., "Paralinear Oxidation of SiC in Water Vapor," J. Am. Ceram. Soc, 80(1), 197-205(1997). l9 E. J. Opila, "Oxidation Kinetics of Chemically Vapor Deposited Silicon Carbide in Wet Oxygen," J. Am. Ceram. Soc, 77(3), 730-736 (1994). 20 E. J. Opila, "Variation of the Oxidation Rate of Silicon Carbide with Water Vapor Pressure," J. Am. Ceram. Soc, 82(3), 625-636 (1999). 21 E. E. Hermes, R. J. Kerans, "Degradation of Non-Oxide Reinforcement and Oxide Matrix Composites," Mat. Res. Soc, Symposium Proceedings, 125, 73-78 (1988). 2 A. Szweda, M. L. Millard, M. G. Harrison, Fiber-Reinforced Ceramic-Matrix Composite Member and Method for Making, U. S. Pat. No. 5 601 674, (1997). 23 S. M. Sim, R. J. Kerans, "Slurry Infiltration and 3-D Woven Composites," Ceram. Eng. Sci. Proc, 13(9-10), 632-641 (1992). 24 E. H. Moore, T. Mah, and K. A. Keller, "3D Composite Fabrication Through Matrix Slurry Pressure Infiltration," Ceram. Eng. Sci. Proc, 15(4), 113-120 (1994). 25 M. H. Lewis, M. G. Cain, P. Doleman, A. G. Razzell, J. Gent, "Development of Interfaces in Oxide and Silicate Matrix Composites," High-Temperature Ceramic-Matrix Composites II: Manufacturing and Materials Development, A. G. Evans, and R. G. Naslain, editors, American Ceramic Society, 41-52 (1995). 26 F. F. Lange, W. C. Tu, A. G. Evans, "Processing of Damage-Tolerant, Oxidation-Resistant Ceramic Matrix Composites by a Precursor Infiltration and Pyrolysis Method," Mater. Sci. Eng. A,A195, 145-150(1995). 27 R. Lunderberg, L. Eckerbom, "Design and Processing of All-Oxide Composites," HighTemperature Ceramic-Matrix Composites II: Manufacturing and Materials Development, A. G. Evans, and R. G. Naslain, editors, American Ceramic Society, 95-104 (1995). 28 E. Mouchon, P. Colomban, "Oxide Ceramic Matrix/Oxide Fiber Woven Fabric Composites Exhibiting Dissipative Fracture Behavior," Composites, 26, 175-182 (1995). 29 P. E. D. Morgan and D. B. Marshall, "Ceramic Composites of Monazite and Alumina," J. Am. Ceram. Soc, 78(6), 1553-1563 (1995). 30 W. C. Tu, F. F. Lange, A. G. Evans, "Concept for a Damage-Tolerant Ceramic Composite with Strong Interfaces," J. Am. Ceram. Soc, 79(2), 417-424 (1996). 31 C. G. Levi, J. Y. Yang, B. J. Dalgleish, F. W. Zok, A. G. Evans, "Processing and Performance of an All-Oxide Ceramic Composite," J. Am. Ceram. Soc, 81, 2077-2086 (1998). 32 J. B. Davis, J. P. A. Lofvander, A. G. Evans, "Fiber Coating Concepts for Brittle Matrix Composites," J. Am. Ceram. Soc, 76(5), 1249-57 (1993). 3 T. J. Mackin, J. Y. Yang, C. G. Levi, A. G. Evans, "Environmentally Compatible Double Coating Concepts for Sapphire Fiber Reinforced γ-TiAl," Mater. Sci. Eng., A161, 285-93 (1993). A. G. Hegedus, Ceramic Bodies of Controlled Porosity and Process for Making Same, U. S. Pat. No. 5 0177 522, May 21, (1991). 35 T. J. Dunyak, D. R. Chang, M. L. Millard, "Thermal Aging Effects on Oxide/Oxide Ceramic-Matrix Composites," Proceedings of 17th Conference on Metal Matrix, Carbon, and Ceramic Matrix Composites. NASA Conference Publication 3235, Part 2, 675-90 (1993). 36 L. P. Zawada, S. S. Lee, "Mechanical Behavior of CMCs for Flaps and Seals," ARPA Ceramic Technology Insertion Program (DARPA), W. S. Coblenz WS, editor. Annapolis MD, 267-322(1994).

202

· Processing and Properties of Advanced Ceramics and Composites

Effects of Environment on Creep Behavior of Nextel™720/Alumina-Mullite Composite

L. P. Zawada, S. S. Lee, "Evaluation of the Fatigue Performance of Five CMCs for Aerospace Applications," Proceedings of the Sixth International Fatigue Congress, 1669-1674 (1996). 38 T. J. Lu, "Crack Branching in All-Oxide Ceramic Composites," J. Am. Ceram. Soc, 79(1), 266-274(1996). 39 M.A. Mattoni, J.Y. Yang, C.G. Levi, F.W. Zok, "Effects of Matrix Porosity on the Mechanical Properties of a Porous Matrix, All-Oxide Ceramic Composite," J. Am. Ceram. Soc, 84(ll),2594-2602(2003). 40 F. W. Zok, C. G. Levi, "Mechanical Properties of Porous-Matrix Ceramic Composites," Adv. Eng. Mater., 3(1-2), 15-23 (2001). 41 L. P. Zawada, R. S. Hay, S. S. Lee, J. Staehler, "Characterization and High-Temperature Mechanical Behavior of an Oxide/Oxide Composite," J. Am. Ceram. Soc, 86(6), 981-90 (2003). 42 M. B. Ruggles-Wrenn, S. Mall, C. A. Eber, L. B. Harlan, "Effects of Steam Environment on High-Temperature Mechanical Behavior of Nextel™720/Alumina (N720/A) Continuous Fiber Ceramic Composite," Composites: Part A, 37(11), 2029-40 (2006). 43 J. M. Mehrman, M. B. Ruggles-Wrenn, S. S. Baek, "Influence of Hold Times on the Elevated-Temperature Fatigue Behavior of an Oxide-Oxide Ceramic Composite in Air and in Steam Environment," Comp. Sci. Tech., 67, 1425-1438 (2007). 44 M. B. Ruggles-Wrenn, P. Koutsoukos, S. S. Baek, "Effects of Environment on Creep Behavior of Two Oxide/Oxide Ceramic-Matrix Composites at 1200 °C," J. Mater. Sci., in press. 45 R. A. Jurf, S. C. Butner, "Advances in Oxide-Oxide CMC," J. Eng. Gas. Turbines Power, Trans ASME, 122(2), 202-205 (1999). 46 M. B. Ruggles-Wrenn, J. C. Braun," Effects of Steam Environment on Creep Behavior of Nextel™720/Alumina Ceramic Composite at Elevated Temperature," Mater. Sci. Eng. A, in press.

Processing and Properties of Advanced Ceramics and Composites

• 203

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

PERFORMANCE OF COMPOSITE MATERIALS IN CORROSIVE CONDITIONS: EVALUATION OF ADHESION LOSS IN POLYMERS VIA CATHODIC DISBONDMENT AND A NEWLY DEVELOPED NDE TECHNIQUE Davion Hill1 , Colin Scott2, Ayca Ertekin\Narasi Sridhar1 1 2

DNV Research & Innovation USA, 5777 Frantz Rd Dublin OH 43017 CC Technologies, 5777 Frantz Rd. Dublin OH 43017

ABSTRACT Fiber reinforced composites and polymer coatings are currently used in many niche applications across many industries. Composite materials are being used not only as stand alone structural members, but also as a reinforcing member for metal or concrete structures. With these new applications for composites, an evaluation of environmental variables which may affect the integrity of reinforced or repaired parts is essential, especially for wet environments. The emphasized property for composite materials in these applications is overall strength, though in wet environments the responsibility of the composite as a coating or protective barrier must also be considered. The goal of this work is to evaluate the effect of impacts (simulated per ASTM G14), cathodic protection and disbondment (simulated per ASTM G95), and electrolytes corrosive to the substrate on the performance of the steel-composite system. In addition, a proof of concept non-destructive evaluation (NDE) technique is evaluated for detection of disbondment and adhesion loss in parallel with ASTM G95. INTRODUCTION Fiber reinforced composites are now used in many niche applications such as repair or permanent structural reinforcement for pipeline, bridge, and ship structures. A variety of monitoring techniques for assessing the integrity of these composite structures are available as a result of new emerging technologies and rapid growth. In this rapidly growing field, there are still questions concerning long term performance. Acoustic non-destructive evaluation (NDE) techniques are easily applicable to composites because it is an established technology with readily available evaluation tools for most materials. Monitoring of composites in the field is an essential function to assess integrity over the long term. While the emphasis on composite materials for these applications is strength, in some cases the responsibility of the composite as a coating or protective barrier must also be considered. Polymeric materials can experience degradation in some environments and these variables should be considered. Impacts, cathodic protection, and electrolytes corrosive to the substrate will inevitably develop into conditions that affect the performance of the composite itself. Degradation of the polymer matrix and adhesion loss can combine to defeat the function of the composite, and if the substrate is cathodically protected, cathodic disbondment (CD) may be a concern. Extensive work at at the Gas Research Institute (GRI) found that the strength of polymethyl methacrylate resin matrix reinforced with glass fiber (e-glass) was sufficient to reinforce a pressurized pipe such that under extreme loads, the repaired region would exceed the strength of the pipe elsewhere1. In that work it was noted that the composite had adequate strength and was expected to be a sound anti-corrosion coating, however the polymer matrix

205

Performance of Composite Materials in Corrosive Conditions

alone offered marginal corrosion protection. It was recommended that an additional coating be applied over the composite repair. Mechanical strength was the emphasis of research on that repair, however other factors could affect the integrity of the repair. Composite materials are expected to perform in an environment where at least 63% of failures are related to some kind of external force or imposed condition that is outside the expected design limits of the material". Damage to a coating precludes eventual damage to the pipe due to corrosion. Damage to a pipeline may not always result in immediate rupture and therefore may not be immediately apparent. However, such latent damage can cause deterioration of long-term performance due to corrosion or stress corrosion cracking. In particular, evaluation of composite reinforcements under cathodic disbondment conditions is essential for a prediction of long term performance. Cathodic protection (CP) is a measure taken to protect metals that are in service in corrosive conditions. Both passive (sacrificial anodes) and active (applied voltage) systems are used to bias current flow and protect the structural member. For pipelines, the impressed current anode may be a graphite- or resinjacketed, metal core cable running alongside the pipe. The voltage is usually adjusted between the anode and pipe such that a -850 mV potential (vs. a Cu-CuSC>4 reference electrode) is maintained on the pipe surface. These specifications are appropriate for most soils". The events that threaten coatings are also a threat to composite reinforcements or repairs. After a defect (such as an impact) has been incurred, moisture in the environment and a cathodic current combine to create conditions favorable for cathodic disbondment. For coatings, the disbondment mechanism begins with the formation of an alkaline environment by cathodic reduction of dissolved oxygen or water4. Generated alkalinity in CP conditions can react with organic polymers that are used in the adhesive layer or mastic in a process called saponification to disbond the coating at the interface between coating and metal at a defect5. The degree of alkalinity is highest near the defect site because the cathodic potential tends to be the most negative at this location and the concentration of oxygen is the highest. The alkaline formation is cyclic and self supporting as the disbondment around the defect grows and encourages a propagating disbondment front away from the defected area6. Since fiber reinforced composites rely on similar adhesives and polymers for matrix materials, the mechanism of cathodic disbondment applies to this system as well, whenever cathodic protection is applied to composite-repaired structures. In this work a three phase test program was developed to evaluate the cathodic disbondment performance of composite patches on metallic substrates. The first two phases are described here and the last will be published at a later date, as some of this work is still ongoing. The first phase was designed to verify the cathodic disbondment mechanism in composite materials and adhesives. In the second phase, an NDE technique was developed to monitor cathodic disbondment in a proof of concept experiment. In the third and final phase, the NDE technique will be applied to composite materials in service. EXPERIMENTAL The effect of impact and subsequent water exposure at cathodic potentials was examined. Steel plates measuring 4" square were laminated with e-glass fiber in a polyester resin. The plates were manufactured by hand lay up of multiple layers of fiberglass. The plates were sand blasted prior to application of the composite. A test matrix was developed for plates with 1-3 layers of fiberglass subjected to three levels of impact with increasing intensity.

206

· Processing and Properties of Advanced Ceramics and Composites

Performance of Composite Materials in Corrosive Conditions

Impact tests were performed via ASTM G14 . The impact testing apparatus used a weight dropped from a fixed and recorded height. The weight tip was hemispherical. The weight was dropped down a tube such that impacts with the surface were uniform and repeatable. A representative impacted specimen is shown in Figure 1. After impact the specimen was tested for conductivity per ASTM G628. The conductivity test used a wet sponge at the end of a wand, with a DC voltage applied between the end of the wand and the substrate. The substrate was connected via an electrode clamp. If the impact penetrated the composite material, water from the sponge on the wand created a conductive path into the holiday (defect) and to the substrate. If conductivity was achieved, the system emitted an audible signal. Specimens were tested with this method to assess whether the impact penetrated to the substrate, because visual observation of impact depth can be misleading. The ASTM G62 is a qualitative "pass or fail" inspection, and the purpose of this step was to determine whether the impact fully penetrated the composite layers. Cathodic disbondment was performed at room temperature via an attached cell method per ASTM G959. The ASTM G95 standard was originally intended for evaluating cathodic disbondment of coatings with drilled holidays. In this case the standard was modified to evaluate CD performance of composite materials (instead of coatings) with an impact site (instead of a drilled defect). An acrylic cell centered about the defect site and affixed to the surface of the composite with a silicone adhesive. Once secured, the cell was filled with a 3% NaCl solution. A platinum wire was inserted into the solution inside a frit, and a 3V DC potential (vs. a saturated calomel reference electrode) was applied between the platinum and the steel plate onto which the composite was adhered. ASTM G95 was performed for 90 days. A photograph of the cathodic disbondment cell is shown in Figure 2. The test cell was removed upon test completion and a dye solution was injected into the impact area in order to capture the approximate disbonded area underneath the composite. The accepted disbondment measurement method for coatings is to mechanically score the coating in a pie-section pattern centered about the defect, remove the loose pieces with a putty knife, and measure the approximate disbonded area. Since composite materials do not "flake" in this fashion, the dye solution was attempted instead with reasonable success.

Processing and Properties of Advanced Ceramics and Composites

· 207

Performance of Composite Materials in Corrosive Conditions

Figure 1 The impact to the glass fiber reinforced polyester caused noticeable deformation around the impact site, as shown in this photograph..

Figure 2 adhesive.

The cathodic disbondment cells are shown in the photo, attached to the composite plates via silicone

In conjunction with the fiberglass disbondment tests, proof of concept NDE testing was performed in parallel on a two part epoxy commercial pipeline coating. An intentional drilled

208

·

Processing and Properties of Advanced Ceramics and Composites

Performance of Composite Materials in Corrosive Conditions

holiday was introduced in the center of a pipe section and the ASTM G95 test method was performed. During the test, a "pitch-catch" acoustic signal was applied at 40 kHz. The receiving sensor was triggered to receive the pulsed signal within a limited time window. The signal was monitored approximately every ten minutes for 20 days. The NDE monitoring configuration is shown in Figure 3.

Figure 3 The photograph shows the ASTM G95 test in progress with NDE sensors on either side of the cell.

The received signal is similar to an exponentially decaying resonance pattern, like what might be expected in the amplitude behavior of a damped simple harmonic oscillator (SHO), or the decaying ring of a struck bell. The solution to the damped SHO differential equation is of the -yt

form c , where γ is the decay constant. Using a widely available spreadsheet program, an exponential decay curve was fit over each normalized signal and the decay constant was extracted from the fit. The captured signal was stored in a single file with a time and date stamp, so this algorithm was applied to each individual file to extract a decay constant with a time signature. Then, the decay constant was plotted as a function of time in a single chart to identify a qualitative pattern, if any, in the signal behavior. The decay constant (y, in this case) was used as an indicator of the signal behavior, i.e., long or short decay periods. The results of disbondment tests of the composite materials are shown below. The results of the NDE technique on the coating are also shown. RESULTS Specimens with one layer of fiberglass began to fail after 30 days of cathodic disbondment testing, it was found that all specimens would fail within a 90 day period, though most specimens that failed the ASTM G62 inspection also failed ASTM G95 in 30 days or less because the solution was in direct contact with the substrate. Specimens that passed ASTM G62 eventually failed because the cracks induced by impact eventually led to water ingress and penetration.

Processing and Properties of Advanced Ceramics and Composites

· 209

Performance of Composite Materials in Corrosive Conditions

Failure during ASTM G95 testing was first characterized by leaking of the test solution at the composite-steel interface at the edge of the plate. Subsequent evaluation revealed the dissolution of the polyester matrix in the majority of the exposed area and sometimes beyond, as shown in Figure 4. In most cases, total disbondment was achieved.

Figure 4 Photographic views of glass fiber reinforced polyester composites exposed to a cathodic disbondment cell show (a) the entire test area illustrating evidence of dissolution of the polyester matrix with the presence of localized dry fiber areas, and (b) a close-up of the polyester matrix dissolved beyond the cathodic cell.

After testing, the CD test solution was subjected to a standard environmental quality test at an EPA certified laboratory. The results indicated the presence of styrene, alcohols, and carboxylic acid, as well as trace detection of chloroform. In almost all cases the entire exposed area of the attached cell showed evidence of disbondment. In the case of the above figures, a dye was injected into the impact site to determine if a disbondment "pocket" was present around the site. The disbonded area as

210

· Processing and Properties of Advanced Ceramics and Composites

Performance of Composite Materials in Corrosive Conditions

measured by the dye is shown in Figure 5. The pictured specimen is the same specimen shown in previous figures. It is evident that the inspection dye penetrated underneath the glass fiber composite and revealed a disbondment "pocket" underneath the composite layers.

Figure 5 Total disbonded area for 2 layers of fiberglass with polyester matrix.

It should be noted that one polyester specimen was tested with ASTM G95 without an impact, and it was permitted to run beyond 90 days. The cell exhibited visual dissolution of the polyester matrix and began minor leakage from the cell seal after approximately 105 days. As part of the second phase of this work, the NDE method was tested in parallel on a two part epoxy coating (without fiber reinforcement). This test indicated that the disbondment mechanism is a discrete rather than gradual process. The qualitative change in signal behavior is shown in Figure 6, where the early signal is more damped than the later signal after disbondment. In the figure, the early signal decays to near zero values in nearly half the time it takes for the post-disbondment signal to decay to zero. It is noted that the acoustic signal decayed quickly before the disbondment while the decay period is extended after the disbondment. The decay constant vs. time trend is shown in Figure 7. Behavior was consistent for several weeks until apparent disbondment occurred. Post test inspection revealed that the disbonded area nearly encompassed the entire exposure area. The cell began to leak within 3-4 days of the decay constant change.

Processing and Properties of Advanced Ceramics and Composites

· 211

Performance of Composite Materials in Corrosive Conditions

Figure 6 The figure shows a heavily damped signal (before disbondment) overlayed on a lightly damped signal (after disbondment).

Figure 7 The figure shows the results of the NDE disbondment measurement with plots of decay constant (normalized) vs. time and the instantaneous change magnitude vs. time.

DISCUSSION The tests performed in this work are accelerated tests designed to generally describe performance of materials in controlled conditions. Composite systems to be used on pipelines are generally thicker (as much as Vi") with a heavier weave than what was tested here, but the materials of construction are typically E-glass with either a polyester or urethane polymer matrix. The soil environment will likely be less aggressive than ASTM G95, however swampy

212

· Processing and Properties of Advanced Ceramics and Composites

Performance of Composite Materials in Corrosive Conditions

conditions or river crossings are considered to be volatile soil environments. As the composite repair market develops and grows, there is variation in product design and specification. These products are applied by hand in the field, so despite recommended practice the actual application may vary. During actual service, the energy of impact is likely to be greater than what was tested here because these products are often employed to repair dented steel pipes - a result of a high energy impact usually incurred by a backhoe shovel. If an impact of similar energy is incurred to the composite itself in this environment, the post-impact exposure time will be significantly longer because most composite systems are expected to be a long term repair (20-50 years). This work shows that under cathodic protection conditions in wet or moist soil, an impact scenario may pose a threat to the long term integrity of the composite, provided that water ingresses between composite and the cathodically protected steel. These results would indicate that the dominant variables are moisture content in the soil, conductivity of the soil, and exposure time. Polymer matrix materials in composites are likely to be more brittle than coatings. Coatings are developed specifically for hydrophobic properties, while composite matrix materials are designed for optimum mechanical properties. In addition, coatings are designed for optimal adhesion to a metal substrate while matrix materials are designed for improved adhesion between fiber layers. Water absorption and uptake of the two materials are therefore different, though the eventual degradation mode during cathodic disbondment is similar. This work indicates that both polymer materials are subject to degradation in alkaline environments induced by the electrochemical conditions supported by cathodic protection. Cathodic disbondment tests were verified to be a mechanism of integrity failure on the fiber reinforced composites. It is evident that the polyester matrix in this case is susceptible to cathodic disbondment in a fashion similar to coatings. The results of the standard environmental quality test on the test fluid verified the presence of styrene. As styrene is known to be a major ingredient in the polymerization of most vinyl ester and polyester resins, this indicates evidence of the dissolution of polyester. The environmental quality test also confirmed high concentrations of alcohols and carboxylic acids. Since polyesters are usually made through condensation of alcohols and carboxylic acids, the presence of these ingredients in the solution is attributed to the decomposition of the polyester matrix. Trace amounts of chloroform likely indicates the dissolution of organics and subsequent reactions with NaCl and evolved Cl (g) on the anode. It should also be noted that after the tests the solution had a soapy consistency, though the pH was near 7. Dissolution of the polyester is considered to be the likely explanation for this observation. The NDE technique successfully captured the disbondment mechanism in the coating and demonstrates that degradation of adhesion in polymeric materials can be measured. The disbondment was indicated by a sudden change in the decay constant behavior when the decay period deviated significantly prior to total leakage from the cell. The disbondment mechanism was a sudden rather than a gradual process in this case. This is consistent with the expectation that the disbondment of the coating would lead to less constraint on the pipe surface, therefore lessening the dampening effect. The acoustic signal detected a significant change in the pattern such that the coating evidently "popped off rather than "peeled" during disbondment. This may indicate a balance between internal stresses in the coating and the forces of adhesion, such that a critical threshold event is evident when the adhesion forces become less than the internal stresses, and total disbondment occurs. The results of this test indicate that this technique may be a viable solution for early warning systems and monitoring for sensitive coated structures, or metallic structures that are coated or reinforced with polymeric materials.

Processing and Properties of Advanced Ceramics and Composites

· 213

Performance of Composite Materials in Corrosive Conditions

At the time of this publication additional composite specimens are under examination using epoxies, acrylics, and polyesters with Kevlar, carbon fiber, and E glass. A similar NDE verification test is currently underway on a Kevlar-wrapped pipe with a polyester matrix to more succinctly capture the behavior of the composite system from an NDE perspective. The scope of the ongoing work is to evaluate the NDE technique for detection of cathodic disbondment in composite materials under test conditions that are closer to a field environment, similar to ASTM G8. In addition, a year long field exposure test is underway to better understand the cathodic disbondment mechanism in actual soils under actual pipeline operating conditions. The goal of the work is to not only evaluate the cathodic disbondment mechanism in composite materials but to also illustrate two possible monitoring and sensing techniques for real time sensor applications. CONCLUSIONS - The glass fiber reinforced polyester composites tested showed susceptibility to cathodic disbondment and degradation in alkaline environments - A dye-based disbondment investigation technique was successfully demonstrated for composite materials A way to quantify adhesion loss with NDE techniques was developed, using the decay constant of the signal as a metric The NDE technique successfully detected cathodic disbondment of epoxy coatings - The NDE method shown here could be a system health assessment tool for structures that are coated or reinforced with polymeric materials ACKNOWLEDGEMENTS Rob Denzine and Gary Snyder provided much of the data collection and fabrication for this work. Intern Kyle Wilson patiently sorted data and performed analysis. This work was funded in part by the Pipeline and Hazardous Materials Safety Administration (PHMSA). REFERENCES 1. Summary and Validation of Clock Spring for Permanent Repair of Pipeline Corrosion Defects. Gas Research Institue. October, 1998. GRI-98/0227 2. Pipeline and Hazardous Materials Safety Administration. PHMSA Significant Incidents Files, August 25, 2008. 3. Control of External Corrosion on Underground or Submerged Metallic Piping Systems. National Association of Corrosion Engineers. Recommended practice. NACE RPO169-2002 4. Application of a General Reactive Transport Model to Predict Environment Under Disbonded Coatings. N. Sridhar, D.S. Dunn, M. Seth. Corrosion, July 2001 5. Principles and Prevention of Corrosion. Denny A. Jones. 2nd Ed. 1996 6. Role of Hydrogen and Hydroxyl Ion in Cathodic Disbondment. N. Kamalanand, ct al. Anti-Corrosion Methods and Materials. Volume 45, No. 4 1998 p 243-247 7. ASTM G14 - 04 Standard Test Method for Impact Resistance of Pipeline Coatings (Falling Weight Test) 8. ASTM G62 - 07 Standard Test Methods for Holiday Detection in Pipeline Coatings 9. ASTM G95 - 07 Standard Test Method for Cathodic Disbondment Test of Pipeline Coatings (Attached Cell Method)

214

Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

EFFECT OF VARIATIONS IN PROCESS SHEAR ON THE MIXEDNESS OF AN ALUMINA - TITANIA SYSTEM C. August, M. Jitianu, R. Haber Rutgers University, Department of Materials Science and Engineering Piscataway, New Jersey, United States of America ABSTRACT Alumina-titania compounds are used in industries ranging from catalyst production to wear resistance coatings. Their formulations require different mixing techniques and forming processes. The goal of this study is to understand the effect of different mixing techniques on the component distribution (mixedness) of the alumina-titania batch. Techniques employed include wet mixing, dry mixing with compounding, and dry mixing with multiple extrusions. An analysis of the effect of processing conditions on the mixedness of these alumina-titania extrudates has been performed using scanning electron microscopy followed by energy dispersive x-ray analysis. Rheological properties of the extruded pastes have also been measured. Overall mixedness of the alumina-titania batch has been found to vary with the process shear applied by mixing. INTRODUCTION Increased manufacturing interest in thin-walled extrudates has lead to a need for greater understanding of the effects of green body composition and particle alignment on the rheological and final properties of a batch. The concept of mixedness can be described as the distribution of a single component within a matrix of another component(s). Quantifying mixedness of a mineral within a green body could be a useful tool in determining processing variations on the effect of green body composition. Alumina-titania compounds are used in a wide range of applications from nuclear waste hosting to wear resistance coatings. One notable use is in NOx catalyst materials, where alumina-titania compositions have been shown to have more hydrothermal stability than zeolite catalysts and higher activity than alumina-zirconia catalysts. Forming of alumina-titania compounds can include extrusion into complex honeycomb structures as well as extrusion in pellets for catalytic use.1"4 Sodium stearate is a water soluble anionic surfactant used in many industries ranging from cosmetics and detergents to petrology and ceramic manufacturing.5 In ceramic manufacturing, sodium stearate is widely used as an extrusion lubricant. It has been used in the production of aluminum-titanate diesel filters and honeycomb cordierite catalysts, as well as catalyst supports.6"9 The adsorption mechanism of anionic surfactants has been found to vary greatly with both physical nature and chemistry of the adsorbent surface. It was found that the amount of surfactant adsorbed was inversely related to the adsorbent oxygen concentration, but was not dependent on the adsorbent pore distribution.10 There are many methods of rheologically evaluating a green ceramic. One practical method of evaluation is torque rheometry. Torque rheometry mimics the compounding process found on a larger scale in many manufacturing procedures. The torque rheometer measures the rotational force required to rotate a torque blade, of specific geometry, through the material at a constant speed over time. Low torque values have been shown to be less likely to break down

215

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

agglomerates than high torque values, though higher torque puts more strain on mixing equipment and requires more energy." Dynamic stress rheometry, DSR, can be used to measure a shear modulus, G\ and a viscous modulus, G". These parameters can be analyzed to determine a material's yield stress. The yield stress is the point at which a material transitions from elastic flow to viscous flow. DSR can be used to analyze pastes, slurries, or gels depending on the geometry used with the equipment. Dynamic stress rheometry has been used to analyze the rheology of alumina slurries with various particle size distributions. It was shown that the G' curve obtained with dynamic stress rheometry could be used to predict the extrusion properties of a paste.12 Mixedness, for powders, can be defined as the spatial and intensity fluctuations of discrete components in a bed of defined porosity and packing hierarchy.13 One of the first comprehensive profiles on mixedness defined three mechanisms of mixing; shear mixing, adjacent planes of particles increase the interface between them by sliding relative to each other, convective mixing, wherein groups of particles are transferred from one location to another, and diffusive mixing, wherein adjacent particles exchange places.14 In powder mixing, these three methods often take place simultaneously.15 There have been many attempts to predict and quantify mixedness, M. Some methods involve calculations based on the probability of particle interaction. Other methods use particle variance within a body as the key factor in quantifying mixedness. Still other methods define mixedness as a ratio of entropies. One of the difficulties in determining mixedness is that the measured quantities are often affected by the measurement method.13'15'16 Another technique for measuring mixedness involves analyzing compositional maps obtained by measuring the x-rays resulting from an SEM raster. In this method, a finite area of a sample is analyzed with an SEM. At each raster point, the incident beam interacts with the atoms it encounters, exciting inner shell electrons to be ejected. This leaves the atoms in an ionized state, which causes them to relax into their ground state. The energy difference between their ionized and ground states is emitted as an x-ray. Because the energy states of an atom are characteristic for each element, so is the difference between the ionized and ground states. The energy of the resultant x-ray will therefore be characteristic of the specific atom it came from. A Si(Li) detector can record the energies of x-rays for each raster point in a sample. In this way, the x-ray detector can determine the elemental composition of each location in the area raster, thus creating a compositional map.17 A compositional map is just a binary image in which each pixel is either on (contains element) or off (does not contain element). Statistical evaluation of compositional maps can extract quantifiable data about the elemental distribution within the sample. One study analyzed EDX maps with Point Pattern Analysis. This technique utilizes nearest neighbor distances of defined particles to create mean vs. variance charts. These distances are defined by using a tessellation analysis. Tessellation is a method of dividing a volume into sub-units, in this case, based on particle, or seed, locations. This analysis mathematically defines a cell around each seed. The cell walls are determined by the distance between the seed and its nearest neighbors. There are two main tessellation algorithms used in this kind of analysis. The Voronoi method uses the distance between seed centers to determine the tessellation tile boundaries. The Johnson-Mehl method uses the seed edges to determine the tile boundaries. By using the seed edges rather than the centers, the Johnson-Mehl method reduces the likelihood that a portion of a seed could be overlapped by an adjacent seed's tile.18

216

· Processing and Properties of Advanced Ceramics and Composites

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

A similar study utilized Image Processing Toolkit 5.0, a software package from Reindeer Graphics, Inc. (Asheville, NC), which is a plug-in for Adobe Photoshop (San Jose, CA). The IPT plug-in allows many statistical functions to be performed on a binary image. In this case, the binary image represented the pore distribution of a silicon carbide. Thresholding turned the SEM image (where each pixel had been valued on a gray scale from 0-255) into a binary image (each pixel is 0 or 1). Edge features were then excluded and the remaining image analyzed for clustering resulting in average nearest neighbor distances and distribution of distances.19 Research on the prediction and measurement of mixedness is justified by the quantity of research which shows the effects of mixedness on the final properties or a manufactured material. The effect of particle clustering on the mechanical properties of a metal matrix composite has been documented. A direct link was made between inhomogeneity and probability of composite failure.20 Another study made a link between degree of mixedness of conductive particles in a polymer matrix correlated to the resulting conductivity of the material."1 In this work, an alumina-titania green body with sodium stearate surfactant and a hydroxypropyl methyl cellulose binder will be extruded and analyzed for rheological properties, by torque and dynamic stress rheometry and mixedness by SEM/EDX mapping and tessellation analysis. Experimental variations will include the concentration of sodium stearate incorporated into the body as well as various processing conditions to examine the effect of high and low shear on the mineral distributions in the extrudate. EXPERIMENTAL PROCEDURE Sample Preparation Procedure The composition for all samples was 45.52% alumina, 35.48% titania, 3% hydroxypropyl methyl cellulose binder, and 16% distilled deionized water. Some samples also contained sodium stearate in concentrations ranging from 0.01% to 1.00%. Sodium stearate additions were always on top of 100%. The ratio of alumina to titania was chosen as a 1:1 molar ratio. The concentrations of binder and water in the batch were chosen based on the necessary consistency that would allow mixing and extrusion of the pastes with the varying sodium stearate additions. For dry mixed batches; alumina and titania were co-milled with the binder and minimal media for 5 minutes. The sodium stearate (if included) was dissolved in the water. The liquid fade was then added to the milled constituents in a low shear, low speed mixture to form "curds" which would facilitate quick addition to the compounder/rheometer. The batch was then either analyzed by torque rheometry or compounded in the Brabender Plasticorder (Duisberg, Germany) at 50 rpms for 30 minutes (shorter for specified samples). Compounded batches were then extruded into 15 mm rods with a Loomis piston extruder (Kaiserslautern, Germany). For wet mixed batches; alumina and titania were milled together with minimal media for 5 minutes. The alumina-titania powders were then dispersed in a 40% (weight percent) slurry. Manual low shear mixing and ultrasonication were used to break down large agglomerates. At this stage, sodium stearate (if included) was added to the slurry followed by more manual low shear mixing and ultrasonication. This slurry was then filter pressed into cakes. The filtrate was measured for each cake and a sample of the cake was weighed, dried, and weighed to determine remaining moisture content. The cakes were placed in a drier at 43°C and weighed every 2 minutes until the desired water content (16%) was reached. These cakes were then broken apart and combined with the binder in a low shear mixture. The batch was then either analyzed by torque rheometry or compounded in the Brabender Plasticorder. Compounded batches were then extruded into 15 mm rods with a Loomis piston extruder.

Processing and Properties of Advanced Ceramics and Composites

· 217

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

Torque Rheometry Measurements were taken on a Brabender Plasticorder with a dual sigma blade configuration at 50 rpm for 16 minutes. Six batches were analyzed through torque rheometry. Batch compositions can be found in Table I. Inorganic components of each batch were milled for 5 minutes with 5% media to promote mixing. Batches 1-5 were combined with the liquid phase (sodium stearate dissolved in water) and analyzed in the torque rheometer. Batch 6 was wet mixed after milling. The milled inorganics were dispersed in a 40% solids loading slurry. The surfactant was added to the slurry and ultrasonicated. The slurry was then filter pressed in multiple chambers to 1-inch filter cakes. These cakes were broken into pieces and dehydrated in a dryer to 16% water content. The cake pieces were then combined with the binder in a low shear mixer and analyzed with the torque rheometer. All of the batches contained an A-16 alumina, Millennium (Hunt Valley, MD) AT-1 titania, 3% hydroxypropyl methyl cellulose binder, and sodium stearate. The sodium stearate concentrations were 0.01%, 0.05%, 0.1%, 0.5%, and 1%. The ratio of alumina to titania is consistent with a 1:1 molar ratio necessary for forming AI2T1O5. Table I. Torque Rheometry A-16 Alumina Batch (%) 45.52 Tl T2 45.52 45.52 T3 T4 45.52 45.52 T5 45.52 T6

Batch Compositions Titania HPMC (%) (%) 35.48 3 35.48 3 35.48 3 35.48 3 35.48 3 35.48 3

Water (%) 16 16 16 16 16 16

Sodium Stearate (% over 100) 0.01 0.05 0.10 0.50 1.00 1.00

Mixing Method Dry mixed Dry mixed Dry mixed Dry mixed Dry mixed Wet mixed

Dynamic Stress Rheometry Dynamic stress rheometry was performed on a TA Instruments (New Castle, DE), AR1000-N Rheometer. Inorganics were combined and milled prior to the addition of liquid phase for both wet and dry mixing. Once all batch components were combined, they were compounded in a Brabender Plasticorder for 30 minutes (in most cases) and extruded as 15 mm rods using a Loomis extruder. Discs of 5 mm thickness were cut from the extrudate for analysis in the TA Rheometer. A parallel plate geometry was used on the TA rheometer. This geometry consisted of a striated base plate on which the sample was centered. The geometry, another striated plate parallel to the base, was then lowered to a set gap. This gap was chosen so that the geometry applied enough pressure to the sample to ensure consistent contact while having minimal destructive effect on the sample. For this experiment, the range of oscillations the sample underwent was between 100 and 100,000 Pa. The samples analyzed with this technique were composed of A-16 alumina, commercially available titania, HPMC, sodium stearate, and water. Batch compositions analyzed by dynamic stress rheometry can be found in Table II.

218

· Processing and Properties of Advanced Ceramics and Composites

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

Table II. Dynamic Stress Rheometry Batch Compositions A-16 Alumina Titania Water HPMC Batch (%) (%) (%) (%) Dl 45.52 35.48 16 3 D2 45.52 35.48 16 3 D3 45.52 35.48 16 3 D4 45.52 35.48 3 16 D5 45.52 35.48 3 16 D6 45.52 35.48 3 16 D7 45.52 35.48 16 3 D8 45.52 35.48 16 3

Sodium Stearate (% over 100) 0.00 0.10 0.50 1.00 0.00 0.10 0.50 1.00

Mixing Method Dry mixed Dry mixed Dry mixed Dry mixed Wet mixed Wet mixed Wet mixed Wet mixed

SEM/EDX A Zeiss Gemini 982 (Jena, Germany) field emission scanning electron microscope was used in conjunction with a Princeton Gamma-Tech (Princeton, NJ) energy dispersive x-ray system to collect elemental maps of extruded batch surfaces. Composition for each SEM/EDX analyzed batch was the same as batch D4. Inorganics in each batch were milled, dry or wet mixed, and extruded into 15 mm rods. These rods were then extruded into 1 mm strands, which were sectioned to expose surface perpendicular to the extrusion direction, mounted on SEM studs, and desiccated for 24 hrs prior to SEM analysis. Processing variations on these batches included, wet mixing, shortened compounding time, and extruding the same batch 5 times. Table III lists the processing variations for batches analyzed with SEM/EDX. For each sample, a spectrum was collected to and collection windows were set to isolate the specific window associated with each element. Maps were taken at 5000x magnification with a voltage of lOkV. Collection time for spectra and maps was 600 seconds and each map was 128x128 pixels. Table III. Batch SI S2 S3 S4

Batch Processing for SEM/EDX Mill Mix 5 minutes Dry mixed Dry mixed 5 minutes Dry mixed 5 minutes Wet mixed 5 minutes

Compound 30 minutes 30 minutes 7 minutes 30 minutes

Extrude 5X extrusion Single extrusion Single extrusion Single extrusion

Mixedness Analysis Quantitative analysis of collected elemental maps was performed on samples S1-S4 with image analysis software Image Processing Took Kit 5.0 and Adobe Photoshop. Elemental maps were thresholded then inverted, converting them to a binary image with black representing the locations of the element and white representing locations where the element was not present, as in Figure 1. White portions of the elemental map were representative of either the other mineral in the binary composition or pore space.

Processing and Properties of Advanced Ceramics and Composites

· 219

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

Figure 1. Mapping Procedure Progression: Image, Map, Binary After thresholding, the binary image was quantified. The image was processed using a tessellation method of determining clustering and randomness in the distribution of the element in the map. Analysis provided the number of features found in the image, average observed nearest neighbor distance, the average expected nearest neighbor distance, and the variances for both distances. Equations (1) and (2) show the determination of the ratio, Q, of observed, R0, to expected, RE, average nearest neighbor distance and the ratio of observed to expected variance, V. In these equations, n is the number of particles, λ is the particle density (n/area), and x is the average distance between particles. These quantities were then plotted, creating a Q/V chart where the clustering and randomness for samples could be compared.

β = .&. = - 1 = V=—

Απλη

(1) (2)

Figure 2 showcases the significance of the quantified Q/V. Coordinate (1,1) in Figure 2 occurs when a perfectly random distribution is observed. Any Q value below 1 implies that clustering of the observed particles is occurring. By that token, any Q value above 1 indicates a disinclination of the particles to cluster. A value of V below 1 indicates a more regular distribution. As V increases, the distribution becomes increasingly random.

220

· Processing and Properties of Advanced Ceramics and Composites

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

RESULTS & DISCUSSION Torque Rheometry Figure 3 shows torque rheometry results for dry mixed and wet mixed alumina-titania compositions with 1% sodium stearate. These batches contain the same composition and were subjected to the same milling time prior to the torque measurement. The dry mixed batch has a much greater peak torque value than the wet mixed batch. The time to peak torque is similar for both batches. One explanation for the difference in peak torque times is that the wet mixing provided better distribution of the surfactant prior to torque rheometry. Figure 4 shows the effect of varying the surfactant concentration in the batch on the torque rheometry curve. Batches with 1% and 0.5% (T5 and T4, respectively) exhibit similar torque curves and peak torque values. These compositions exhibit the lowest peak torque values of the concentrations. The batches tend to increase in peak torque value as the concentration of sodium stearate decreases. The exception to this occurs in the batch with 0.05% sodium stearate. Looking at the scale of Figure 4, it can be seen that the difference between the peak torques of the dry mixed batch and wet mixed batch in Figure 3 is relatively small.

Processing and Properties of Advanced Ceramics and Composites

· 221

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

» Ί

8-

\ 6 -

LII

Dry mixed - - - Wet mixed

Γ

I5

s i

Λ.

\

S 4 -1

ί

34

!i

1Ϊ

'^"-^

·""—*".

l] I

L·^ .

00

2

4

6

*-..—

8

-%,....»...,.

10

......Λ-ν—.'Vv-".·"·

12

14

Time (min)

Figure 3. Torque Rheometry of Alumina-Titania Batches, T5 and T6.

Figure 4. Torque Rheometry of Alumina-Titania Batches T1-T5

222

· Processing and Properties of Advanced Ceramics and Composites

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

Dynamic Stress Rheometry Figure 5 shows dynamic stress rheometry results for batches containing a range of sodium stearate concentrations under two kinds of processing. The concentrations examined by DSR are 0% (no sodium stearate), 0.1%, 0.5%, and 1% sodium stearate. The dry mixed batches are represented by the bolder solid and dashed lines. The graph in Figure 5 shows only the G' (shear modulus) for the compositions. Though there are many ways of analyzing a GVG", this paper focuses on the yield stress, which can be observed in the abrupt drop off in the G' curve. In this study, the yield stress has been defined as the oscillatory stress at which an order of magnitude drop in stress (Pa) occurs. Of the dry mixed batches, the batch with 1% sodium stearate exhibits the greatest yield stress with 0.1% having the lowest, followed by 0.5% and 0% with increasingly greater yield stresses. The batch with 0.1% might have the lowest yield stress as it is the batch to which the greatest amount of mixing energy was already applied during compounding (as indicated by the torque rheometry results). The wet mixed batches exhibit the same trend, regarding the concentrations of sodium stearate, as the dry mixed batches. The batch without sodium stearate exhibits the highest yield stress, followed by the 0.5%, 1%, and 0.1%, sequentially. For the batch without sodium stearate and the 0.1%o concentration, the wet mixed batches have higher yield stresses than the dry mixed. For the higher concentration batches, 0.5%, and 1%, the dry mixed batches have higher yield stresses than the wet mixed. The shape of the G' curve varies between concentrations, but remains fairly consistent between wet and dry mixed batches of the same concentration. Curve shape is likely related to agglomerate distribution and the breaking down (yielding) of different levels of agglomeration. This complicates attempts to quantify a G' curve by yield stress alone as some curves seem to exhibit multiple yields. This is something to take into consideration when comparing the dry mixed 1 % batch to the dry mixed 0.1 % batch. It appears that the 1 % has a much higher yield, but the dual yield of the 0.1%) G' curve indicates that it might have multiple levels of agglomerates, which may already have been broken down· during mixing in the 1% batch. The ultrasonication and slurry forming in the wet mixing technique breaks up large agglomerates causes the rheology of the batch to act as if it has a larger volume fraction than the dry mixed of the same composition. This can be seen in the Figure 5 wet and dry mixed curves for 0% surfactant. At large surfactant additions, this trend is reversed as the surfactant in the dry mixed batch adsorbs to surface freed by agglomerates which are broken down by the oscillatory stresses.

Processing and Properties of Advanced Ceramics and Composites

· 223

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

Figure 5. Dynamic Stress Rheometry of Batches D1-D8 Mixedness Analysis SEM/EDX and analysis of the subsequent elemental maps was performed for four batches. These batches varied by processing technique. Figure 6 shows the Q/V graph for batches S1-S4. A full spectrum was collected from the area of interest for each batch to ensure that the windows were properly gated to collect data on the specific elements of interest, aluminum and titanium. Oxygen maps were not collected. Figure 6 shows the results for the analysis of aluminum and titanium for the collected EDX maps for batches S1-S4. It is assumed that the locations on the map indicating aluminum represent locations with alumina and the locations indicating titanium represent locations with titania. Figure 6 shows the Q values for alumina and titania under all processing conditions are above 1. Referring back to Figure 2, we can see that this indicates a trend to not form clusters. This is not surprising as all the batches, despite variations between them, went through many levels of mixing and processing designed to break down agglomerates and discourage large clusters. The batch which seems to be least inclined to cluster, for both alumina and titania, is the batch which was extruded 5 times. Extrusion applies a great amount of shear force on the surface of the extrudate and pressure on the bulk of the material. This batch sustained the most shear energy of any of the batches. The next closest Q value for both alumina and titania is the wet mixed batch, followed by the short and long mix time batches, which are close in value. This might imply that after the torque value peaks, a consistent torque is reached and varying the mixing time might not have the greatest impact on agglomerate break down. The trend of the V values for alumina for all process variations remain the same for titania. The batch that was extruded 5 times has the greatest V value for both alumina and titania. This means it exhibits the least regular distribution for all the processing variations. This

224

· Processing and Properties of Advanced Ceramics and Composites

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

is followed by the wet mixed batch, then the long mix time and the short mix time. The batch with the least amount of processing (short mix time, single extrusion) shows the greatest tendency towards a regular distribution. The relationship between the alumina and the titania is consistent between batches. The alumina has similar Q values to the titania, but a consistently higher V value. This would imply that the titania is distributed more regularly than the alumina. One contributing factor is the difference in inorganic volumes. There is more alumina in the batch than titania, making it easier for the titania to form a regular distribution within a matrix of alumina.

Figure 6. Q/V Graph for Batches S1-S4 CONCLUSION SEM/EDX and subsequent image analysis has been shown to be effective at determining variations in mineral distribution in an alumina-titania composition. This technique has shown that greater processing of a batch reduces clustering within the batch. Torque rheometry has shown that decreasing the surfactant in a batch requires greater mixing energy. Dynamic stress rheometry has shown that batches which undergo greater mixing energies prior to extrusion yield earlier under oscillatory stresses. SEM/EDX analyses indicate that titania is more regularly distributed in an alumina-titania composition with 1:1 molar ratio. It also indicates that batches with greater applied shear stress during forming will have less clustering. This helps to confirm this technique as viable for monitoring processing variation on a multicomponent green body.

Processing and Properties of Advanced Ceramics and Composites

· 225

Effect of Variations in Process Shear on the Mixedness of an Alumina-Titania System

REFERENCES Yielding, P., White, T. "Crystal chemical incorporation of high level waste species in aluminotitanate-based ceramics: Valence, location, radiation damage, and hydrothermal durability" Journal of Materials Research 2 [3](1987) 387-414 2 Kawabata, K., Yoshimatsu, H„ Fujiwara, K., Yabuki, T., Osaka A., and Miura, Y. "Properties and catalytic activity of NO x reduction of alumina-titania catalysts prepared by sol-gel method" Journal of Materials Science 34 [11](1999) 2529-2534. 3 Tabata, M., Hamada, H., Kintaichi, Y., Sasaki, M , and Ito, T. Sekiyu Gakkaishi 36 (1993) 149. 4 Wilson, G.R., Kayamoto, M. Process for Forming Alumina Pellets Containing Titania and Process for Producing Hydrodesulfurization Catalyst Containing the Pellets. US Patent 4,196,101. April 1980. 5 Shanefield, D J . Organic Additives and Ceramic Processing: With Applications in Powder Metallurgy, Ink and Paint. Springer, (1996). ( 'Wu, S. Formable Ceramic Compositions and Method of Use Therefore. US Patent 5,344,799. September 1994. 7 Addiego, W.P., Brundage, K.R., Glose, C.R. Method of Producing Alumina-Silica Catalyst Supports. US Patent 7,244,689. July 2007. 8 Ogunwumi, S.B. Tepesch, P.D. Mullite-Aluminum Titanate Diesel Exhaust Filter. US Patent 6,849,181. February 2005. 9 Hickman, D.L. Batch Compositions For Cordierite Ceramics. US Patent 5,332,703. July 1994. 10 Wu, S.H. Pendleton, P. "Adsorption of Anionic Surfactant by Activated Carbon: Effect of Surface Chemistry, Ionic Strength, and Hydrophobicity." Journal of Colloid and interface Science 243 (2001) 306-315. "Pileggi, R.G., Studart, A.R., Pandolfelli, V.C., Gallo, J. "How Mixing Affects the Rheology of Refractory Castables, Part 2" American Ceramic Society Bulletin 80 [7] (2001) 38-42. I2 Mazzeo, F.A. "Extrusion and Rheology of Fine Particulate Ceramic Pastes" PhD thesis. Rutgers University. 2001. 13 Fan, L.T., Chen, S.J., Watson, C.A. "Solids Mixing" Industrial and Engineering Chemistry 62 [7](1970)53-69. 4 Lacey, P.M.C. "Developments in the Theory of Particle Mixing" Journal of Applied Chemistry 4JMay] (1954) 257-268. L Masuda, H., Higashitani, K., Yoshida, H. Powder Technology: Handling and Operations, Process Instrumentation and Working Hazards. CRC Press. 2006. l6 Ogawa, K., Ito, S. "A Definition of Quality of Mixedness" Journal of Chemical Engineering of Japan. 8 [2] (1975) 148-151. l7 Goldstein, J.I., Newbury, D.E., Echlin, P., Joy, D.C., Romig, A.D., Lyman, C.E., Fiori, C , Lifshin, E. Scanning Electron Microscopy and X-Rav Microanalysis: A Text for Biologists, Materials Scientists, and Geologists. 2 n ed. New York: Plenum Press, 1992. Gulliver, E.A. "Simulation of Mixture Microstructures via Particle Packing Models and Their Direct Comparison with Real Mixtures" PhD thesis. Rutgers University, 2003. 19 Demirbas, M.V. "Microstructure-Property Relationship in Silicon Carbide Armor Ceramics" PhD Thesis. Rutgers University, 2008 20 Segurado, J., Gonzalez, C , Llorca, J. "A Numerical Investigation of the Effect of Particle Clustering on the Mechanical Properties of Composites" Acta Materiala 51 (2003) 2355-2369. 21 Kalyon, D.M., Birinci, E., Yazici, R., Karuv, B., Walsh, S. "Electrical Properties of Composites as Affected by the Degree of Mixedness of the Conductive Filler in the Polymer Matrix" Polymer Engineering and Science 42 [7] (2002) 1609-1617.

226

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Modeling

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

MODELING OF THE PRESSURE IN 1-D GREEN CERAMIC BODIES DURING DEPRESSURIZATION FROM CONDITIONS OF SUPERCRITICAL EXTRACTION OF BINDER Kumar Krishnamurthy and Stephen J. Lombardo* Department of Chemical Engineering University of Missouri Columbia, Missouri 65211, USA *Department of Mechanical and Aerospace Engineering University of Missouri Columbia, Missouri 65211, USA ABSTRACT Supercritical extraction is an alternative to thermal decomposition for removing binder from green ceramic bodies. After depressurization from the conditions of supercritical extraction, defects are sometimes observed in the green ceramic components, and these are attributed to gradients in pressure in the green ceramic body that arise during depressurization. The pressure gradients are modeled with Darcy's Law for describing flow in porous media, and two types of depressurization modes are examined: step changes in pressure and linear changes in pressure with time. INTRODUCTION Supercritical extraction (SCE) using CO2 as a solvent has been used successfully to extract binder from green ceramic components, and thus is an alternative processing strategy for binder removal. In particular, SCE has been applied to multilayer ceramic capacitors (MLCs) substrates with BaTi03 as the dielectric that were prepared using 10 wt.% of an organic blend consisting of a 55:45 weight ratio of high molecular weight poly(vinyl butyral) and low molecular weight dioctyl phthalate1"5. These studies have shown that about half the binder can be extracted in three hours at 75-95°C and 40 MPa. Binder removal by supercritical extraction has also been effective with many injection molded and extrusion molded ceramic parts made from alumina powder containing paraffins, fatty acids, or thermoplastics as the main organic components in the binder blends6"14. It has also been reported that defect free samples can be obtained in a short period of time when organic additives are extracted from injection molded and extrusion molded ceramic parts using supercritical fluids12"14. For the BaTi03-based MLCs, SCE was performed in a semi-continuous mode1"5, which has the advantage that extraction can be performed with a smaller volume of solvent as compared to a continuous mode of extraction. The semi-continuous mode involves pressurizing the vessel to a predetermined pressure and temperature and then dwelling at these conditions for a fixed amount of time. The dwelling period is next followed by a depressurization step wherein the binder rich CO2 phase is throttled out of the vessel. During depressurization from the supercritical state, defects such as fracture and delamination were occasionally observed in BaTiO.i-based MLCs5. These defects occur especially when depressurization is performed rapidly. Although rapid depressurization has the advantage of shortening the overall process time, it may affect the yield of samples depending on the binder loading and the strength of the green body.

229

Modeling of the Pressure in 1-D Green Ceramic Bodies during Depressurization

To model the depressurization process, Darcy's Law for flow in a porous medium is a convenient starting point. In most formulations of Darcy's Law, the viscosity is taken as a constant, not dependent on pressure15. Several experimental investigations have suggested, however, that because the viscosity is known to vary with pressure, the use of a constant viscosity may not be reasonable under operating conditions when the range of pressure is large16" 20 . It was further stated that an exponential dependence of viscosity on pressure exists at high pressures or when pressure gradients are large. In this work, we model the depressurization process as flow in porous media using Darcy's Law, and we take into the account variations in the viscosity with pressure. MODEL For describing flow in porous media in a ceramic green body (see Fig. 1), Darcy's Law can be written in terms of a superficial fluid velocity, vr, in the x direction as15 KdP

μ ox

/n

where κ is the permeability of the sample, μ is the fluid viscosity, and P is the pressure. The continuity equation for the flow of a fluid in a 1-D body as given by21,22

iV>-iV,) ot

ox

(2)

where p is the molar density and εν is the porosity of the sample. Darcy's Law from Eq. (1) can then be combined with Eq. (2) to obtain the following expression for describing 1-D flow in a porous medium as

£(^)._L£&i ct

dx

(3)

Since density is the dependent variable in the model, pressure values are calculated using the Peng-Robinson cubic equation of state with values of the constants adopted from elsewhere23'24. For simplifying the calculations related to the dependence of viscosity on molar density, we express the viscosity of CO2 as a quadratic equation of the form μ(ρ) = α*ρ2 +b*p + c*

(4)

where Λ*=1.7Χ10"' 3 Pa s m6/mol2,fc*=8.2xlO-11 Pa s nrVmol, and c = 1.8xlO"5Pa s. These values were obtained using regression analysis on data for the variation in viscosity of CO2 with molar density at 90°C25"26. Although Eq. (3) is the standard form of the continuity equation, we extend it in this work by allowing the viscosity to vary and by relating the density to pressure via a cubic equation of state.

230

·

Processing and Properties of Advanced Ceramics and Composites

Modeling of the Pressure in 1 -D Green Ceramic Bodies during Depressurization

Figure 1. One dimensional schematic of the green body showing the coordinate system and the body length. RESULTS Figure 2 shows an image of an MLC sample fabricated from an acrylic-based polymer with a polyester adipate plasticizer. The sample in Fig. 2 was first exposed to supercritical carbon dioxide for 1 h at 90°C and 10 MPa. Under these conditions, the sample experienced a weight loss of 0.8 weight%. As seen in Fig. 2, after depressurization over 8 h to ambient conditions, both horizontal and vertical cracks are present. The horizontal defects tend to be between the layers, and the largest crack is at the center of the green body.

Figure 2. Image of an MLC sample after depressurization over 8 h from conditions of supercritical extraction in carbon dioxide at 10 MPa at 90°C for 1 h. The sample lost 0.8 weight% of the initial binder content.

Processing and Properties of Advanced Ceramics and Composites

· 231

Modeling of the Pressure in 1-D Green Ceramic Bodies during Depressurization

To account for the defect formation evident in the image in Fig. 2, we next evaluate model results for different types of depressurization from conditions of supercritical extraction. At this time, it is not possible to unambiguously specify the value of the pressure or the pressure gradient that leads to failure; nonetheless, it is of interest to have at least a qualitative sense of the how the pressure varies within the green body during depressurization. Figure 3 shows the time evolution of the pressure of CO2 versus position in the sample for step changes in pressure of 40 to 20 MPa (left-hand side, LHS, of Fig. 3) and 40 to 35 MPa (right-hand side, RHS, of Fig. 3) at 90°C (see Table I for the parameters used in the simulations). For the larger step change, steep pressure profiles exist initially within the sample, and these diminish with increasing time until after 5 s the pressure becomes more uniform within the pore space. Although the pressure is largest at the center of the sample, the pressure gradients within the sample are largest near the body edges. For the smaller step change, the pressure profiles are less steep across the sample and the pressure gradients diminish more quickly with time. ou -

40CO Q_

2

40 - 35 MPa

40 - 20 MPa 0.1 s^^__|

0.1 s

Ts"^-^^ 5s

30-

a.

/ 20-

1s

5s

10- — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — I — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — -0.01 0.01 -0.005 0 0.005 x(m)

Figure 3. Pressure of CO? at 90°C versus position and time for step changes in the boundary value pressure. The left-hand side corresponds to a step change from 40 to 20 MPa, and the right-hand side corresponds to a step change from 40 to 35 MPa. Table I. List of parameters used in Figures 3-8. Values Parameter (Units) Figs. 3 - 5 Figs. 6 - 8 κ (m2)

lxlO17

lxlO17

Hm) M-)

0.02

0.02

0.15

0.15

Po(MPa)

40

40

20,35

-

-

6s,6h

step

P

(M?&)

DF

t (sorh)

232

· Processing and Properties of Advanced Ceramics and Composites

Modeling of the Pressure in 1 -D Green Ceramic Bodies during Depressurization

In earlier work in the modeling of thermal debinding27'28, we have shown that the stress within the sample is proportional to the gradient in pressure within the sample28, and that both the maximum stress and maximum pressure occur in the center of the green body. We thus define a quantity Pnx_, which is proportional to the gradient in pressure, as the ratio of the pressure at any position in the body, Px, to the pressure at the body edges, PX=±L/2, as (5) Figure 4 thus compares the values of P„iX for the large and small step changes in pressure in Fig. 3 for different times. For the large step change of 40 to 20 MPa, Pnx is initially large in the center of body at a value of ~2 and decreases across the body thickness. With increasing time, Pn,x versus position in the body is reduced and approaches a value of unity everywhere, thus indicating a uniform pressure in the green body. For a smaller step change of 40 to 35 MPa, Pnx is much smaller versus position and time as compared to the corresponding values for the larger step change.

1.00

0.96 0.01

Figure 4. Behavior of PtltX of CO2 at 90°C versus position and time for step changes in the boundary value pressure. The left-hand side corresponds to a step change from 40 to 20 MPa, and the right-hand side corresponds to a step change from 40 to 35 MPa. Figure 5 summarizes the time evolution of the pressure at the body center, Px=o, and the normalized pressure at the body center, Pn,x=o, of C0 2 for step changes from 40 to 20 MPa and 40 to 35 MPa. Both quantities decrease monotonically with increasing time. An alternative to the step mode of depressurization is to decrease the fluid density (or pressure) in the vessel linearly with time, and this can be conducted either rapidly or slowly. Figure 6 quantifies the effect of a depressurization time, tDP=6 s (LHS of Fig. 6) versus ^=6 h (RHS of Fig. 6) for values of the simulation parameters given in Table I. For a rapid rate of linear depressurization, the pressure across the sample varies spatially more strongly with time as compared to a slower linear rate of depressurization, where the pressure profiles are more uniform for all times.

Processing and Properties of Advanced Ceramics and Composites

·

233

Modeling of the Pressure in 1-D Green Ceramic Bodies during Depressurization

50 2.5

40 co

30 X

1.5

20

c Q.

10 H

0.5

0.01

0.1

10

t(s)

Figure 5. Behavior of Px=() (left-hand axis) and Pn,.x=o (right-hand axis) of C0 2 at 90°C versus time for step changes in the boundary value pressure of 40 to 20 MPa and 40 to 35 MPa. tDP = 6s

J

OS

J

ΤΓ

1-^-

1 -1

1

,

tDP = 6h

J

Oh

1h

J^—π

1

!

_!

2h

!

4h

i ,

(

ι

1

0.005

,

0.01

x (m)

Figure 6. Pressure of C0 2 at 90°C versus position and time for linear changes in the boundary value pressure. The left-hand side corresponds to a depressurization time of 6 s, and the righthand side corresponds to a depressurization time of 6 h. Figure 7 shows how Pnx varies spatially and temporally for the rapid and slow linear rates of depressurization shown in Fig. 6. Now, in contrast to the behavior seen in Fig. 4, Pnx is initially near unity throughout the sample, and then increases with increasing time during depressurization; once again, however, the maximum value of Pnx is always at the center of the body. Although the trends for the time evolution of P,ltX are different as compared to Fig. 4, the magnitude of P„tX is smaller for linear depressurization as compared to a step change in pressure.

234

· Processing and Properties of Advanced Ceramics and Composites

Modeling of the Pressure in 1 -D Green Ceramic Bodies during Depressurization

For the slower rate of linear depressurization (RHS of Fig. 7), the values P„iX are much smaller than for the faster rate of depressurization (LHS of Fig. 7). 1.5 I

J

tDP-6s

!

tDP - 6 h

As

/^^\

I

f

1.0016

1.0012

1.2 1.0004

f -0.01

os -0.005

;

oh

0

0.005

I 0.01

x (m)

Figure 7. Behavior of PtliX of C0 2 at 90°C versus position and time for linear changes in the boundary value pressure. The left-hand side corresponds to a depressurization time of 6 s, and the right-hand side corresponds to a to a depressurization time of 6 h. Figure 8 summarizes the evolution of Px=0 and Pn,x=o of CO2 versus time for rapid and slow linear rates of depressurization. For the case with depressurization lasting 6 s, although there is a gradual reduction in the pressure at the center of the sample with time, the value of Pn,x=o increases. Although the same qualitative effect is observed in the same two quantities, with a slower linear rate of depressurization, the overall magnitude is much lower. CONCLUSIONS Defects have been shown to arise in green ceramic bodies following supercritical extraction of binder and such defects are postulated to occur during depressurization from conditions of supercritical extraction. The origin of the defects is ascribed to pressure gradients that arise during depressurization. The temporal and spatial evolution of these pressure gradients are modeled with Darcy's Law for flow in porous media for both step changes and linear changes in the boundary value pressure. In general, the pressure gradients are larger and take longer to dissipate for larger step changes in pressure or for more rapid linear changes in the boundary value pressure. ACKNOWLEDGEMENT This work was supported in part by the NSF/ICURC Center for Dielectric Studies.

Processing and Properties of Advanced Ceramics and Composites

· 235

Modeling of the Pressure in 1-D Green Ceramic Bodies during Depressurization

0

1

| ,__ |

en J

ou

40 1o 30 CL

t(h) 3

2 j

|

4 ,_

5

|

,

!

f t = 6s

t DP ==

0-

| V 1.8

6rNv

,r

'

1.6 M

o

S

6

,

20 -

CL

Ρ=68 10 -

t

1

1.2

_L

\ΌΡ^Γ

0 -

0

·4 3

2

- ^

7

'

1 0.8

l

3

4

5

6

t(s)

Figure 8. Behavior οΐΡχ=ο (left-hand axis) and Pn.x=o (right-hand axis) of C 0 2 at 90°C versus time for rapid and slow linear changes in the boundary value pressure. REFERENCES [ K. V. Shende, D. S. Krueger and S. J. Lombardo, Supercritical Extraction of Binder Containing Poly(vinyl butyral) and Dioctyl Phthalate from Barium Titanate-Platinum Multilayer Ceramic Capacitors, J. Mater. Sci.: Mater. Electron., 12, 637-643 (2001). 2 R. V. Shende, D. S. Krueger and S. J. Lombardo, Binder Removal by Supercritical Extraction from BaTiU3-Pt Multilayer Ceramic Capacitors, Ceramic Transactions, Vol. 129, Innovative Processing and Synthesis: Ceramics, Glasses and Composites V, eds. J.P. Singh, N.P. Bansal, A. Bandyopadhyay, and L. Klein (American Ceramic Society, Westerville, OH) (2002) 165174. 3 R. V. Shende and S. J. Lombardo, Supercritical Extraction with Carbon Dioxide and Ethylene of Poly(vinyl butyral) and Dioctyl Phthalate from Multilayer Ceramic Capacitors, J. Supercrit. Fluids, 23, 153-162(2002). 4 R. V. Shende, M. Kline and S. J. Lombardo, Effects of Supercritical Extraction on the Plasticization of Poly(vinyl butyral) and Dioctyl Phthalate Films, J. Supercrit. Fluids, 28, 113120(2004). 5 R. V. Shende, T. R. Redfern and S. J. Lombardo, Defect Formation during Supercritical Extraction of Binder from Green Ceramic Components, J. Am. Ceram. Soc, 87, 1254-58 (2004). 6 D. W. Matson and R. D. Smith, Supercritical Fluid Technologies for Ceramic-Processing Applications, J. Am. Ceram. Soc, 72, 871-881 (1989). 7 S. Nakajima, S. Yasuhara, and M. Ishihara, Method of Removing Binder Material from a Shaped Ceramic Preform by Extracting with Supercritical Fluid, US Patent No. 4,731,208, March 15, 1988.

236

· Processing and Properties of Advanced Ceramics and Composites

Modeling of the Pressure in 1-D Green Ceramic Bodies during Depressurization

8

T. Miyashita, Y. Ueno, H. Nishio, and S. Kubodera, Method for Removing the Dispersion Medium from a Molded Pulverulent Material, US Patent No. 4,737,332, April 12, 1988. E. Nishikawa, N. Wakao and N. Nakashima, Binder Removal from Ceramic Green Body in the Environment of Supercritical Carbon Dioxide With and Without Entrainers, J. Supercrit. Fluids, 4, 265-269(1991). ,0 T. Chartier, M. Ferrato and J. F. Baumard, Supercritical Debinding of Injection Molded Ceramics, J. Am. Ceram. Soc, 78, 1787-92 (1995). U T. Chartier, E. Delhomme and J. Baumard, Mechanisms of Binder Removal Involved of Injection Molded Ceramics, J. De Physique III, 7, 291-302 (1997). 12 T. Chartier, E. Delhomme, J. F. Baumard, P. Marteau, P. Subra and R. Tufeu, Solubility in Supercritical Carbon Dioxide, of Paraffin Waxes Used as Binders for Low-Pressure Injection Molding, bid. Eng. Chem. Res., 38, 1904-10 (1999). ,3 F. Bordet, T. Chartier, J. F. Baumard, The Use of Co-Solvents in Supercritical Debinding of Ceramics, J. European Ceram. Soc, 22, 1067-72 (2002). 14 T. Chartier, F. Border, E. Delhomme, J. F. Baumard, Extraction of Binders from Green Ceramic Bodies by Supercritical Fluid: Influence of the Porosity, J. European Ceram. Soc, 22, 1403-09(2002). I5 H. Darcy, Les Fontaines Publiques de La Ville de Dijon, Victor Dalmont, Paris, 1856. 16 K. Kannan and K. R. Rajagopal, Flow Through Porous Media Due to High Pressure Gradients, App. Math. Comput., 199, 748-59 (2008). 17 G. G. Stokes, On the Theories of Internal Friction of Fluids in Motion, and of the Equilibrium and Motion of Elastic Solids, Trans. Camb. Phil. Soc, 8, 287-305 (1845). l8 C. Barus, Isotherms, Isopiestics and Isometrics Relative to Viscosity, Am. J. Sci., 45, 87-96 (1893). 19 P. W. Bridgman, The Physics of High Pressure, MacMillan, New York, 1931. 20 E. C. Andrade, Viscosity of Liquids, Nature, 125, 309-310 (1930). 21 S. J. Lombardo and Z. C. Feng, Pressure Distribution during Binder Burnout in ThreeDimensional Porous Ceramic Bodies with Anisotropic Permeability, J. Mater. Res., 17, 143440 (2002). 22 M. E. Harr, Mechanics of Paniculate Media, McGraw Hill, New York, 1997. 23 D.-Y. Peng and D. B. Robinson, A New Two-Constant Equation of State, Ind. Eng. Chem. Fundam., 15,59-64(1976). 24 J. M. Smith, H. C. Van Ness and M. M. Abbott, Introduction to Chemical Engineering Thermodynamics, Sixth Ed., McGraw Hill, New York, 2001. 25 A. Fenghour, W. A. Wakeham and V. Vesovic, The Viscosity of Carbon Dioxide, J. Phys. Chem. Ref Data, 27, 31 -44 (1998). 26 V. Vesovic, W. A. Wakeham, G. A. Olchowy, J. W. Sangers, J. T. R. Watson and J. Millat, The Transport Properties of Carbon Dioxide, J. Phys. Chem. Ref Data, 19, 763 -770 (1990). 27 J. W. Yun, S. J. Lombardo, Effect of Decomposition Kinetics and Failure Criteria on BinderRemoval Cycles from Three-Dimensional Porous Green Bodies, J. Am. Ceram. Soc, 89, 176183(2006). 28 Z. C. Feng, B. He, and S. J. Lombardo, Stress Distribution in Porous Ceramic Bodies During Binder Burnout, J. ofAppl. Mech., 69, 497-501 (2002). 9

Processing and Properties of Advanced Ceramics and Composites

· 237

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

MODELS OF THE STRENGTH OF GREEN CERAMIC BODIES AS A FUNCTION OF BINDER CONTENT AND TEMPERATURE Stephen J. Lombardo# and Rajiv Sachanandani Department of Chemical Engineering University of Missouri Columbia, MO 65211, USA ^Department of Mechanical and Aerospace Engineering University of Missouri Columbia, MO 65211, USA ABSTRACT The strength of ceramic green bodies is evaluated for binder distributed in two ways between ceramic particles: pendular state bonds and coated state bonds. Two models for the strength of the binder are then utilized. In the first, the strength of the binder is taken as a constant, and in the second, the yield strength of the binder is represented as a function of temperature and the activation energy for polymer segment motion. INTRODUCTION The main role of binder in the fabrication of ceramic components is to impart strength to green bodies in order to aid in handling. Such binder may consist of one or more organic compounds or polymers, and the binders are presumed to reside between particles in the pore space of the green body. Such binder, however, ultimately must be removed from the green body, and this most often occurs by thermal degradation of the organic species. During the thermal removal of binder [1-3], however, the pressure buildup arising from the decomposition of binder leads to stress in the green body, which may ultimately lead to failure. Earlier modeling efforts have been able to quantify both the buildup of pressure and the occurrence of stress [4-7]. A common feature of these models, however, is that little is said about how the strength of the green body changes during the binder removal heating cycle, when both the volume fraction of binder and the binder material properties are changing. As a consequence of both changes in temperature and binder volume fraction, the failure mode of the ceramic green body may change as well. At low temperatures and/or low binder volume fractions, failure may arise primarily from a brittle mode of failure. At high temperatures and/or high volume fractions of binder, the failure mode may be more plastic in nature. In this work, two models are evaluated to describe the strength of green ceramic bodies [8-10]. In the first model—which may correspond to brittle failure—the strength of the binder is taken as constant and then the effects of binder loading and solids loading on the green strength are examined. For the case of plastic failure, the first model is modified to include the effects of temperature on the yield strength of polymers. These two models are then applied to two cases for the distribution of binder between the ceramic particles: pendular state bonds and coated state bonds. The strength of the green body is then evaluated for different volume fractions of solid and binder.

239

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

THEORY Onoda [9] has developed a model for the strength of porous bodies by postulating that failure occurs through the organic bonds between particles. The model then incorporates the idea that the tensile strength of the green body depends on the strength of individual bonds and the number of bonds. For an individual bond, the tensile strength depends on the binder content at the particle neck, and the force, b\ required to break a bond is given by [9,10]: (1)

F = aBA

where A is the cross-sectional area of binder at the bridge between the particles and OB is the cohesive or adhesive strength of the binder. At the neck between particles, the binder may be distributed in different ways depending on the degree of wetting of the binder on the surface of the particles. Figure 1(A) shows a nonwetting binder for which the contact angle between the binder and the particle is >90 ; such nonwetting behavior would presumably result in poor bonding between the particles. For the case of binder that wets the particles (a contact angle of <90 ), the binder will be pulled into the neck region by capillary action as shown by Fig. 1(B); this is referred to as pendular state bonds. If the wetting binder is viscous and is present in sufficient amount, it would tend to coat the particles as is depicted by Fig. 1(C); this is referred to as coated state bonds. Models are next developed for the strength of binder distributed in the pendular and coated states, which are the most relevant for ceramic processing.


(B) Binder layer

(C) Particle boundary

Figure 1 Distribution of binder between ceramic particles for: (A) non-wetting binder, (B) wetting binder (pendular state bonds), (C) fully coated particles (coating state bonds). Adapted from Ref. [9].

240

·

Processing and Properties of Advanced Ceramics and Composites

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

To proceed further, expressions for the area between particles in Eq. 1 are needed. It is first assumed that the particles are monosized spheres of radius, r, and that every neck between two particles has an equal amount of binder. For binder distributed at the neck of particles in the form of pendular bridges, the cross-sectional area is given by Onoda [9] as

4WV.T W2{v,

(2)

where Vs is the total volume of solid ceramic in the body, Vj, is the volume occupied by the binder, and k is the average number of touching neighbors around each particle. Assuming that the green body consists of only three components, namely ceramic, binder, and pores, and that dimensional changes of the green body are small during binder removal, then volume is conserved. The volumes on Eq. 2 can thus be replaced with volume fractions as es + sb + ε = 1

(3)

where Eb is the volume fraction of binder, ε is the volume fraction of pores, and es is the volume fraction of solids. With this relationship, Eq. (2) then becomes ,

/

\ 1/2

In circumstances where the binder completely coats the particles with a uniform thickness, e.g., coated state bonds, the cross-sectional area of binder at the neck is given by:

Onoda [9] then combined these two expressions for the area between particles with Rumpfs model [8] for describing the tensile strength of particulate bodies consisting of randomly packed monosized spheres of uniform size: σ

= Αί!ζ£Μ 32

{πτ2)

(6)

where k is the mean coordination number of spheres around a given sphere. Several relations between k and ε have been derived [8,11], and one of the simplest is that given by Rumpf [8]: * = -

Processing and Properties of Advanced Ceramics and Composites

(7)

· 241

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

Model I - Constant Strength of Binder To obtain an expression for the green strength of a composite body consisting of ceramic particles and binder, Rumpf s [8] Eq. 6 for the tensile strength of randomly packed spheres is combined with Eq. 1. Then, Onoda's [9] derived relationships, Eqs. 4 and 5, for the area of pendular and coated state bonds are used, and the resulting expression for the green strength for pendular state bonds is

and for coated state bonds is

(9)

^τίττ^^Η

16 (\-ss-eh) {ej We note that these two expressions differ by a constant and in the dependence of strength on the volume fractions of binder and solid. Model II - Temperature Dependent Yield Strength of Binder To describe binder strength that depends on temperature, Eyring [12] has suggested a model which is based on the idea that a segment of polymer must overcome an energy barrier in moving from one position to another in a solid. This activated process occurs with a forward and reverse rate across the energy barrier. The important point established by Eyring [12] is that the application of a shear stress modifies the barrier height, so that in the direction of stress, the rate of segment jumping, formerly slow, now becomes fast enough to give rise to measurable strain rate. Further analysis leads to that the shear stress in a tensile test is the maximum shear stress at yield, aR/2, and the final result is that the strain rate at yield, <s\·, is based on Arrhenius behavior with an activation energy, AH\ ;,=;oexp^]exp[^r]

(.0)

where V is the activation volume, R is the gas constant, and εο is a constant. Rearrangement of Eq. 10 leads to the yield strength as Γ

ση=Τ\ L

ΔΗ

r. ΛΊ

+ 2.303/? log \ε*)\

When this result is combined with Eqs. 8 and 9, the expression for the yield strength for pendular state bonds is,

242

· Processing and Properties of Advanced Ceramics and Composites

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

3>/3^

(ε,+ε„)

(εΎ*

f 2

8 vo-*.-*»)UJ U*

ΔΗ

( · λ

+ 2.303/?log

(12)

and for coated particles is σ =

3π (εχ+ε„)

ίεΛί

2

AH

ί · \

— + 2.303/? log ε0

(13)

RESULTS AND DISCUSSION Model I - Constant Strength of Binder From Eqs. (8) and (9), it is evident that the strength of the green body is dependent on the volume fractions of binder and solids and on the strength of binder. Such a situation could, for example, represent brittle failure of the green ceramic body at low temperature. To proceed further, the binder is assumed to have a constant tensile yield strength of 60-65 MPa [13] at room temperature—this range is appropriate for polycarbonate. For the case of pendular state bonds, Fig. 2 shows that the green strength increases with both the volume fraction of binder and with the solids loading. For coated state bonds (see Fig. 3), similar trends are observed as in Fig. 2, but the precise values of the green body strength depend on the details of the model. 250

0.05

0.1

Figure 2 Strength (Eq. 8) versus binder volume fraction at different volume fractions of solids for pendular state bonds.

Processing and Properties of Advanced Ceramics and Composites

· 243

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

450 -1 400 j 350 ] 300 J

250

2

j

— 200 H Ό J

150 J 100 j 50 j 0 i-r-r 0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Figure 3 Strength (Eq. 9) versus binder volume fraction at different volume fractions of solids for coated state bonds. Model II - Temperature Dependent Yield Strength of Binder For the case where the yield strength of binder depends on the temperature, Fig. 4 shows the yield strength versus temperature for different strain rates. With increasing temperature, the strength appears to decrease almost linearly with temperature. For a constant temperature, decreasing the strain rate leads to a weaker body. In addition to the dependence on the temperature, both the activation energy and activation volume can influence the yield strength. Figure 5 shows that the yield strength of the binder is sensitive to both of these values. Figures 6 and 7 show the results for the yield strength of the green body when the models of Rumpf [81 and Eyring [12] are combined for pendular state bonds (Eq. 12) and for coated state bonds (Eq. 13), respectively. As seen earlier, the strength of the green body decreases with binder content in the green ceramic, and with increases in the temperature. Once again, however, differences in the green strength of the body are observed depending on how the binder is distributed in the green body.

244

· Processing and Properties of Advanced Ceramics and Composites

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

300

320

340

360 380 Temperature (K)

400

420

Figure 4 Yield strength (Eq. 11) of polycarbonate versus temperature for different strain rates with AH = 309,000 kJ/kmol, έ0= 2.067* 1030 s"1, V*= 3.9 m 3 /kmol.

ΔΗ = 400 kJ/mol V = 3 m /kmol

τ loo

ΔΗ = 350 kJ/mol

<§, 80

V* = 5 m' /kmol

*

60

ΔΗ = 300 kJ/mol

40

V* = 7 mTkmol

20 0 280

300

320

340 360 Temperature (K)

380

400

Figure 5 Yield strength (Eq. 11) versus temperature at constant strain rate of 0.1s" and at different activation energies and activation volumes.

Processing and Properties of Advanced Ceramics and Composites

• 245

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

i <4u

:

120

:

c =

05

T == 300K

/

100 \ -a 80

'

:

350 K

/

OH

^r 60 -: D

-

40 ;

^y

y

^ ...--"" 400 K

20 0

s ss

/

: ■H-pTi

0.05

0.1

o.i5

o.:

0.25

0.3

0.35

Figure 6 Strength (Eq. 12) versus binder volume fraction at constant strain rate of 0.01s"1 and AH= 309,000 kJ/kmol, έ0= 2.067* 1030 s_1and V = 3.9 m3/kmol for pendular state bonds.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Figure 7 Strength (Eq. 13) versus binder volume fraction at constant strain rate of 0.01 s"1 and AH= 309,000 kJ/kmol, έ0 = 2.067*1030 s'and V*= 3.9 m3/kmol for coated state bonds.

246

· Processing and Properties of Advanced Ceramics and Composites

Strength of Green Ceramic Bodies as a Function of Binder Content and Temperature

CONCLUSIONS The effect of the binder content, binder distribution, solids loading, and temperature on the strength of ceramic green bodies has been examined. The distribution of binder is depicted in two limiting cases as either being in the pendular state between particles or as surrounding particles completely. The strength of the binder is taken as either a constant or is represented as a yield strength that depends on temperature. The green strength of ceramic bodies decreases monotonically with decreasing solids content, with decreasing binder loading, and with increasing temperature, although differences do exist depending on how the binder is distributed in the green body. REFERENCES [I] R. M. German, "Theory of Thermal Debinding," Int. J. Powder Metall., 23, 237-245 (1987). [2] J. A. Lewis, "Binder Removal From Ceramics," Annual Rev. Mater. Sci., 27, 147-173 (1997). [3] P. Calvert and M. Cima, "Theoretical Models for Binder Burnout," J. Am. Ceram. Soc, 73 [3] 575-579 (1990). [4] G. Y. Stangle and I. A. Aksay, "Simultaneous Momentum, Heat and Mass Transfer With Chemical Reaction in a Disordered Porous Medium: Application to Binder Removal From a Ceramic Green Body," Chem. Eng. Sci., 45, 1719-1731 (1990). [5] D.-S. Tsai, "Pressure Buildup and internal Stresses During Binder Burnout: Numerical Analysis", AlChEJ., 37, 547-554 (1991). [6] Z. C. Feng, B. He, and S. J. Lombardo, "Stress Distribution in Porous Ceramic Bodies During Binder Burnout," J. Appl Mech., 69 497-501 (2002). [7] S. J. Lombardo and Z. C. Feng, "Analytical Method for the Minimum Time for Binder Removal from Three-Dimensional Porous Green Bodies," J. Mater. Res., 18 [11] 27172723 (2003). [8] H. Rumpf, The Strength of Granules and Agglomerates: International Symposium on Agglomeration, Interscience, London, UK, 1962. [9] G. Onoda, "Theoretical Strength of Dried Green Bodies with Organic Binders," J. Am. Ceram. Soc, 59 [5] 236-239 (1976). [10] S. A. Uhland, R. K. Holman, S. Moissette, M. J. Cima, and E. M. Sachs, "Strength of Green Ceramics with Low Binder Content," J. Am. Ceram. Soc, 84 [12] 2809-2818 (2001). [II] W. O. Smith, P. D. Foote, and P. K. Busang, "Packing of Homogeneous Spheres," Phys. Rev., 34 [\\] 1271-1274(1929). [12] N. G. McCrum, C. P. Buckley, and C. B. Bucknall, Principles of Polymer Engineering, Oxford Science Publications, New York, 1997, p. 172-174. [13] ASM Handbooks, Volume 19, Fatigue and Fracture , ASM International, USA, 1996.

Processing and Properties of Advanced Ceramics and Composites

· 247

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

FINITE ELEMENT MODELING OF STEEL WIRE DRAWING THROUGH DIES BASED ON ENCAPSULATED HARD PARTICLES Daniel J. Cunningham, Erik M. Byrne, Ivi Smid, John M. Keane1 Pennsylvania State University, University Park, PA, USA 1 Allomet Corporation, North Huntingdon, PA, USA ABSTRACT The process of pulling a wire through a series of conical shaped dies which incrementally reduces its cross sectional area is known as wire drawing. These wire drawing dies are subjected to extremely high stresses while at the same time expected to survive long service lifetimes. Finite element modeling is used to model the interactions of these materials throughout the wire drawing process. These models show that during the drawing process the wire at the exit of the die can reach local stresses of roughly 150% of its yield strength. The required drawing force is monitored at varying approach angles and varying friction coefficients. At low approach angles the drawing force required is 50-60% larger than at high approach angles. Fluctuations of 5-10% of the drawing load are seen in all cases and are found to have a dependency on geometry, material properties, and drawing speed. In conventional drawing dies the standard approach angles are 10-15°. For the first time, finite element analysis has confirmed these approach angles as the largest angles where significant fluctuation can be prevented, and at the same time the smallest allowable angles without risking wire rupture due to high drawing forces. INTRODUCTION The first hard metals were patented in 1923 and were comprised of tungsten carbide and cobalt [1]. These early materials were primarily marketed for wire drawing dies and other highwear applications. In the past three quarters of a century many improvements have been made, but the basic principle remains the same [2-5]. Most of these changes can be attributed to the decreasing powder particle size and complying with stricter quality control standards. The microstructure of Tough Coated Hard Particles (TCHP) is unique in that it is comprised of hard "core" particles surrounded by a tough tungsten carbide cobalt shell. By combining A1203, WC, and Co in this way, a new composite material is produced which has an extremely high wear resistance and still maintains a high toughness [6]. Wire drawing is the process in which a wire is pulled through a series of dies in order to reduce its cross sectional area. TCHP is an impressive candidate for wire drawing due to its inherent wear resistance; as its surface wears, new wear resistant material resurfaces. At the exit region of the die, the stress in the wire exceeds its yield strength and local plastic deformation takes place. When the wire is pulled through a die, small dislocations in the wire are shifted and the material is work hardened. Wires, such as those used in radial car tires, are pulled through a series of eight to fifteen dies. The drawing speed and the reduction ratio along with other factors become important in maintaining control in the plastic region. If either factor produces excessive stress, then uncontrolled deformation can occur. This uncontrolled deformation can result in wire breakage or an elongated wire that does not meet final dimensional tolerances. The wire die itself is another factor influencing the success of wire drawing runs. Die longevity is crucial in industrial applications because of the costs related to die failure. Die wear and fracture are common modes of failure. It is important to consider both the die and the wire

249

Modeling of Steel Wire Drawing through Dies Based on Encapsulated Hard Particles

when simulating the wire drawing process. To date there have been only limited finite element simulations of this process due to its computationally demanding nature [7]. FINITE ELEMENT MODELING Wire Drawing Assumptions • Material Properties are uniform for both the die and wire materials. Material properties for TCHP have been generalized to model the wire die as a continuum material. • Loading above the yield strength of the steel wire is modeled using a bilinear kinematic response. • Although the wire is modeled using an inelastic response, the wire die is modeled as an elastic material. This simplification is made since peak stresses during drawing are below 70% of the rupture strength of the composite hard metal wire die. Die Modeling Procedure In order to accurately model the behavior of TCHP under the complex loading conditions found in wire drawing, ANSYS 11.1 finite element modeling software is used [8]. The FE model uses Plane 183 2-D triangular elements containing mid-nodes. Mid-nodes allow for a quadratic response in displacement. The element size increases with increasing distance from the die-wire interface, as seen in fig. 1. A fine mesh is present at the die-wire interface to prevent problematic stress irregularities that can arise if there are too few contact nodes. Contact surfaces are assigned on all inner faces of the wire die as well as all exposed faces of the wire. A standard Lagrange solver is used to minimize penetration between the die and the wire mesh. To further achieve this goal, an automatic contact adjustment is set to close gaps and reduce penetration throughout the duration of the solution. This FE model is performed using a transient analysis with 3000 elements (fig. 2). The bottom face of the wire is loaded with a normal stress oriented in the drawing direction. In order to accurately calculate the large plastic deformations of the wire 1000 sub-steps are used to model 30% of the wire being pulled through

Figure 1. Assembly of TCHP particle model (left). FEM meshing of wire die with increased mesh density near die-wire interface (right). Wire: Ew=193GPa, oy=285MPa, Tangent Modulus=2.5GPa, Die: Ed=850GPa

250

· Processing and Properties of Advanced Ceramics and Composites

Modeling of Steel Wire Drawing through Dies Based on Encapsulated Hard Particles

Figure 2. Von Mises stress distributions surrounding exit of wire die (left). Plastic strain deformation in drawn wire (right). the die. Due to large plastic deformations in the wire and its non-linear material properties, it is necessary to use stabilized finite elements. A constant energy dissipation value of 0.1 is used. Since the potential energy in the system is much greater than this ratio its effects are minimal. Micro-scale Particle Modeling Modeling Procedure Due to the composite nature of the die material, a particulate model of a TCHP is simulated. A cross section of a sphere is placed inside several spherical shells in order to model the encapsulated A1203 core particles inside of the WC-Co matrix. The compositions of these layers are shown in fig. 1. The transition layer is comprised of an even weight percent mixture of Co and WC. This layer provides a more uniform stress distribution by removing the abrupt interface of pure Co and WC. Mesh sizes vary between layers depending on thicknesses. The interfaces between these layers are assumed to be in complete contact throughout the simulation RESULTS DISCUSSION Micro-scale Modeling Analyzing the composite model of TCHP produces some interesting results. When tensile stresses are applied to the die, high stress regions can be seen at the interface between the transition layer and the WC-Co matrix. The lowest stresses are seen in the pure Co binder layer. This is due to the significantly lower elastic modulus of the Co as compared to the WC. This influence gives the structure its ability to resist shock. Also, stress distributions throughout the structure in the direction parallel to the applied stress are similar to the distribution in the perpendicular direction. Since tensile stresses are generally the cause of failure in drawing dies this relationship between applied tensile stress and inter-particle stresses can be used to predict particle separation. Stress States in the Die-Wire System The resultant stresses and permanent strains are studied at various stages in the drawing process. Figures 2 and 3 show the stress distribution after a significant amount of wire has been drawn. It can be seen that there is a small stress concentration equaling almost 150% of the wire's yield strength. At increased distances the stress is reduced to below the yield strength.

Processing and Properties of Advanced Ceramics and Composites

· 251

Modeling of Steel Wire Drawing through Dies Based on Encapsulated Hard Particles

Furthermore, fig. 3 shows that the stresses are distributed up the wire a distance equal to the width of the wire. These stress regions are nearly equal to the wire's yield strength. Lastly, residual stress fluctuations can be seen throughout the length of the wire after it has exited the die. As expected, the permanent strain has a similar pattern to the stresses shown in fig. 2. Severe plastic deformation of 17% takes place locally at the die exit. This deformation decays throughout the width of the wire coming to a uniform strain at the midsection. The sharp edge at the die exit is responsible for this effect. Similar to the stress state, there is plastic deformation in the wire before it reaches the die exit. Plastic deformation only occurs when there is direct contact between the die and the wire. These deformations are not only deforming the wire to a narrower diameter, they are also deforming the wire upwards. The upwards deformation effect takes place at a distance equal to roughly half the wire diameter in the drawing direction. This phenomenon varies with the drawing speed, approach angle, reduction ratio, and many other factors. A tradeoff must be made between the speed and the reduction ratio of the die so that the stresses and plastic deformations can be controlled. If uncontrolled plastic deformation occurs within the wire there can be warping or even fracture. This is one of the reasons why the die material is required to be much harder than wire material. Any deformations in the die itself would most likely result in die cracking and immediate failure. Table 1. Wire Drawing Drawing Speed Final Diameter Reduction Percentage Bearing Length Exit Angle

Specifications. 3 m/s 1 mm 12% 40% of diameter 90°

To determine the effects of each of these parameters, they are studied individually while the rest of the die-wire system is held constant. The permanent parameters for the FE model are displayed in table 1. The wire is drawn through the die at approach angles ranging from 6° to 40°. At low approach angles there is a 50-60% increase in the required drawing load as shown in fig. 3. This increase is attributed to the increase in contact area at low approach angles. The amount of die-wire contact area with a 6° approach angle is 52% higher than with an approach angle of 40°. To further indentify this relationship, a study is done on the effects of the frictional coefficient on the required drawing force. It can be seen in fig. 3 that the relationship between drawing force and friction coefficient is not linear. As the friction coefficient is increased beyond 0.3 its effects become less influential. When the drawing loads are studied over time it can be seen that very low approach angles require more drawing force, there is less of an oscillation in that frequency. The amplitude of these oscillations varies from 5-10% of the overall drawing load and it can be seen that patterns in the wave shapes reoccur throughout drawing. This could be due to the presence of a combination of several frequencies taking place in the structure. Further analyses of these oscillations are needed to determine exactly what influences them and how they can be minimized.

252

· Processing and Properties of Advanced Ceramics and Composites

Modeling of Steel Wire Drawing through Dies Based on Encapsulated Hard Particles

Figure 3. Variance of drawing force on approach angle at 3 m/s (left). Variance of drawing force on friction coefficient at 3 m/s and 22° approach angle (right). CONCLUSIONS Stress concentrations at the die exit become extremely high compared to the drawing stress applied to the wire. Both stress and plastic strain fluctuate during drawing. Modeling of TCHP as a composite structure shows how these stresses can translate to a separation along the particle-matrix interface. The composite model can be used to predict failure resulting from this separation caused by both the applied stress and resulting stress fluctuations. The required drawing force is found to be dependent on both the approach angle and the friction coefficient. In all cases it is shown that stress fluctuations in the drawing force occur and are affected by changes in geometry, material properties and drawing speeds. Further work is needed to determine how to minimize these fluctuations and such increase the drawing speed. 5. REFERENCES [1]

R.E. Toth, I. Smid. "Tough-Coated Hard Powders for Hardmetals of Novel Properties"; 2001 International Conference on Powder Metallurgy & Particulate

[2]

P. Ettmayer, "Hardmetals and Cermets"; Annual Review of Material Science, 1989, vol.19, pp. 145-164

[3]

R.E Toth, J. Keane, I. Smid, P. Ettmayer, "Tough coats on hard powders — a revolution in the making"; Metal Powder Report, 2003, vol. 58, no. 9, pp 14

[4]

R.M. German, I. Smid, et al., "Liquid phase sintering of tough coated hard particles," Int. J. Refractory Metals and Hard Materials, 2005, vol. 23, no. 4-6, pp 267

[5]

B. Roebuck, C. Hamann, E. Bennet, "The Palmqvist Test for Tough Hardmetals," HM, 2005, vol. 89, Proc. 16th Int. Plansee Seminar

[6]

R.E. Toth, J.M. Keane, I. Smid, "New TCHP Drawing Dies with Unprecedented Performance"; Wire Journal International, 2008, vol. 4, no. 1, pp. 68-72

[7]

C.J. Luis, J. Leon, R. Luri. "Comparison between finite element method and analytical methods for studying wire drawing process"; Journal of Materials Processing Technology, 2005, pp 1218-1225

Processing and Properties of Advanced Ceramics and Composites

· 253

Modeling of Steel Wire Drawing through Dies Based on Encapsulated Hard Particles

[8]

ANSYS* Academic Research, Release 11.1, ANSYS, Inc.

[9]

E. Van der Giessen and A. Needleman, "Micromechanics Simulations of Fracture," Annual Reviews of Materials Research, 2001, vol. 32, pp 141-162

[10] A.E. Giannakopoulos, P.L. Larsson, et al, "Analysis of Vickers indentation," International Journal of Solids and Structures, 1994, vol. 31, no. 19, pp 2679, 1994 [11] T.L. Anderson, Fracture Mechanics Fundamentals and Applications, Boca Raton, FL, Taylor & Francis Group, 2005 [12] H. Raffii-Tabar, L.H., and M. Cross, "A multi-scale atomistic-continuum modeling of crack propagation in a two-dimensional macroscopic plate"; Journal of Physics: Condensed Matter, 1997, vol. 10, pp 2375-2387

254

· Processing and Properties of Advanced Ceramics and Composites

Processing and Properties of Advanced Ceramics and Composites Edited by Narottam P. Bansal and J. P. Singh Copyright O 2009 The American Ceramic Society.

Author Index

Agaogullari, D., 77 Alexander, D, J., 171 Arin, M., 131 August, C , 215 Borisovich, P. K., 45 Bruce, R. W., 3 Byrne, E. M., 249 Chang, C. L, 143 Chen, M. W., 107 Chin, E. S. C , 107 Choi, S. R., 171 Cunningham, D. J., 249 David, A., 3 Duman, I., 77 Ertekin, A., 205 Faucett, D. C , 171 Feng, C. R., 3 Fliflet, A. W., 3 Flores-Garcia, J. C , 25 Gairola, A., 15 Genelin, C. L, 193 Gold, S. H., 3

Haber, R.,215 Hill, D., 205 Imam, M. A.,, 3 Ishikawa, T., 181 Izui, H., 93 Jitianu, M., 215 Kayihan, A. B., 131 Keane, J. M., 249 Kell, J.,153 Kimura, H., 119 Ko, C , 143 Krishnamurthy, K., 229 Lara-Curzio, E., 35 Leal-Cruz, A. L., 25, 35 Li, S., F., 93 Lin, S., 53 Lombardo, S. J., 229, 239 McCrabb, H., 153 Mihajlovna, D. E., 45 Okano, M., 93 Paliwal, B., 107 255

Author Index

Peascoe, R., 35 Pech-Canul, M. I., 25, 35

Singh, R. P., 65 Smid, I., 249 Sridhar, N., 205

Qadri, S. B., 3 Rahman, A., 65 Ramanathan, S., 143 Ruggles-Wrenn, M. B., 193 Sachanandani, R., 239 Sano, T., 107 Scott, C , 205 Selig, J., 53 Sevik, S., 131 Shivashankar, S. A., 15 Siebach, K. L, 3

Trejo, R. M., 35 Tsuchiya, M., 143 Umarji, A. M., 15 Villafuerte, J., 161 Wang, J., 161 Watanabe, T., 93 Zhang, W. H., 93 Zunjarra, S. C , 65

256 · Processing and Properties of Advanced Ceramics and Composites