Nanostructured Materials, Second Edition: Processing, Properties and Applications

Nanostructured Materials

Nanostructured Materials Processing, Properties, and Applications Second Edition

Edited by Carl C. Koch North Carolina State University, Raleigh, North Carolina

Copyright © 2007 by William Andrew, Inc. No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from the Publisher. Front cover depicts a simulation of nanocrystalline grains. Images provided by Professor Donald Brenner, North Carolina State University. Cover by Hannus Design Nanostructured® is a registered trademark of Hybrid Plastics, Inc., and is used in this book with the consent of Hybrid Plastics. Library of Congress Cataloging-in-Publication Data Nanostructured materials : processing, properties, and applications / edited by Carl C. Koch. p. cm. Includes bibliographical references and index. ISBN-13: 978-0-8155-1534-0 (978-0-8155) ISBN-10: 0-8155-1534-0 (0-8155) 1. Nanostructured materials. I. Koch, C. C. TA418.9.N35N3535 2006 620′.5–dc22 2006034353 Printed in the United States of America This book is printed on acid-free paper. 10 9 8 7 6 5 4 3 2 1 Published by: William Andrew Publishing 13 Eaton Avenue Norwich, NY 13815 1-800-932-7045 www.williamandrew.com Sina Ebnesajjad, Editor in Chief (External Scientific Advisor) NOTICE To the best of our knowledge the information in this publication is accurate; however the Publisher does not assume any responsibility or liability for the accuracy or completeness of, or consequences arising from, such information. This book is intended for informational purposes only. Mention of trade names or commercial products does not constitute endorsement or recommendation for their use by the Publisher. Final determination of the suitability of any information or product for any use, and the manner of that use, is the sole responsibility of the user. Anyone intending to rely upon any recommendation of materials or procedures mentioned in this publication should be independently satisfied as to such suitability, and must meet all applicable safety and health standards.

Materials Science and Process Technology Series Series Editor: Gary E. McGuire International Technology Center, Research Triangle Park, North Carolina Vacuum Deposition onto Webs, Films, and Foils, Charles A. Bishop, 978-0-8155-1535-7, 494 pp., 2007 Ultrananocrystalline Diamond: Synthesis, Properties, and Applications, Olga A. Shenderova and Dieter M. Gruen, eds., 978-0-8155-1524-1, 620 pp., 2006 Diffusion Processes in Advanced Technological Materials, Devendra Gupta, ed., 0-8155-1501-4, 550 pp., 2005 Handbook of Ellipsometry, Harland G. Tompkins and Eugene A. Irene, eds., 0-8155-1499-9, 886 pp., 2005 Thin Film Materials Technology, Kiyotaka Wasa et al., 0-8155-1483-2, 534 pp., 2004 Tribology of Abrasive Machining Processes, Ioan D. Marinescu, et al., 0-8155-1490-5, 751 pp., 2004 Handbook of Thin Film Deposition Processes and Techniques, 2nd Ed., Krishna Seshan, ed., 0-8155-1442-5, 657 pp., 2002 Nanostructured Materials, Carl C. Koch, ed., 0-8155-1451-4, 636 pp., 2002 Handbook of Hard Coatings, Rointan F. Bunshah, ed., 0-8155-1438-7, 568 pp., 2001 Handbook of VLSI Microlithography, 2nd Ed., John N. Helbert, ed., 0-8155-1444-1, 1022 pp., 2001 Wide Bandgap Semiconductors, Stephen J. Pearton, ed., 0-8155-1439-5, 591 pp., 2000 Handbook of Chemical Vapor Deposition, 2nd Ed., Hugh O. Pierson, 0-8155-1432-8, 506 pp., 1999 Hybrid Microcircuit Technology Handbook, 2nd Ed., James J. Licari, and Leonard R. Enlow, 0-8155-1423-9, 601 pp., 1998 Handbook of Magneto-Optical Data Recording, Terry McDaniel, and Randall H. Victora, eds., 0-8155-1391-7, 967 pp., 1997 Ultra-Fine Particles, Chikara Hayashi, R. Ueda and A. Tasaki, eds., 0-8155-1404-2, 467 pp., 1997 Handbook of Refractory Carbides and Nitrides, Hugh O. Pierson, 0-8155-1392-5, 362 pp., 1996 Diamond Chemical Vapor Deposition, Huimin Liu, and David S. Dandy, 0-8155-1380-1, 207 pp., 1995 Handbook of Compound Semiconductors, Paul H. Holloway and Gary E. McGuire, eds., 0-8155-1374-7, 936 pp., 1995 Handbook of Vacuum Arc Science and Technology, Raymond L. Boxman, Philip J. Martin, and David M. Sanders, eds., 0-8155-1375-5, 771 pp., 1995 High Density Plasma Sources, Oleg A. Popov, ed., 0-8155-1377-1, 465 pp., 1995 Molecular Beam Epitaxy, Robin F. C. Farrow, ed., 0-8155-1371-2, 790 pp., 1995 Handbook of Deposition Technologies for Films and Coatings, 2nd Ed., Rointan F. Bunshah, ed., 0-8155-1337-2, 887 pp., 1994 Contacts to Semiconductors, Leonard J. Brillson, ed., 0-8155-1336-4, 702 pp., 1993 Diamond Films and Coatings, Robert F. Davis, ed., 0-8155-1323-2, 437 pp., 1993 Electrodeposition, Jack W. Dini, 0-8155-1320-8, 381 pp., 1993 Handbook of Carbon, Graphite, Diamonds and Fullerenes, Hugh O. Pierson, 0-8155-1339-9, 419 pp., 1993 Handbook of Multilevel Metallization for Integrated Circuits, Syd R. Wilson, Clarence J. Tracy, and John L. Freeman, Jr., eds., 0-8155-1340-2, 901 pp., 1993 Handbook of Semiconductor Wafer Cleaning Technology, Werner Kern, ed., 0-8155-1331-3, 645 pp., 1993 Chemical Vapor Deposition of Tungsten and Tungsten Silicides, John E. J. Schmitz, 0-8155-1288-0, 249 pp., 1992 Chemistry of Superconductor Materials, Terrell A. Vanderah, ed., 0-8155-1279-1, 838 pp., 1992 Electrochemistry of Semiconductors and Electronics, John McHardy and Frank Ludwig, eds., 0-8155-1301-1, 375 pp., 1992

Handbook of Sputter Deposition Technology, Kiyotaka Wasa and Shigeru Hayakawa, 0-8155-1280-5, 316 pp., 1992 Handbook of Plasma Processing Technology, Stephen M. Rossnagel, Jerome J. Cuomo, and William D. Westwood, eds., 0-8155-1220-1, 547 pp., 1990 Handbook of Polymer Coatings for Electronics, 2nd Ed., James Licari, and Laura A. Hughes, 0-8155-1235-X, 408 pp., 1990 Handbook of Semiconductor Silicon Technology, William C. O’Mara, Robert B. Herring, and Lee P. Hunt, eds., 0-8155-1237-6, 815 pp., 1990 Characterization of Semiconductor Materials, vol. 1, Gary E. McGuire, ed., 0-8155-1200-7, 342 pp., 1989 Handbook of Ion Beam Processing Technology, Jerome J. Cuomo, Stephen M. Rossnagel, and Harold R. Kaufman, eds., 0-8155-1199-X, 456 pp., 1989 Diffusion Phenomena in Thin Films and Microelectronic Materials, Devendra Gupta, and Paul S. Ho, eds., 0-8155-1167-1, 604 pp., 1988 Handbook of Contamination Control in Microelectronics, Donald L. Tolliver, ed., 0-8155-1151-5, 509 pp., 1988 Ionized-Cluster Beam Deposition and Epitaxy, Toshinori Takagi, 0-8155-1168-X, 239 pp., 1988 Semiconductor Materials and Process Technology Handbook, Gary E. McGuire, ed., 0-8155-1150-7, 689 pp., 1988 Chemical Vapor Deposition for Microelectronics, Arthur Sherman, 0-8155-1136-1, 227 pp., 1987

Related Titles MEMS: A Practical Guide to Design, Analysis and Applications, Jan G. Korvink, and Oliver Paul, eds., 0-8155-1497-2, 990 pp., 2006 Adhesives Technology for Electronic Applications, James J. Licari, and Dale W. Swanson, 0-8155-1513-8, 479 pp., 2005 Coating Materials for Electronic Applications, James J. Licari, 0-8155-1492-1, 550 pp., 2003 Crystal Growth Technology, K. Byrappa and T. Ohachi, 0-8155-1453-0, 611 pp., 2003 Handbook of Hydrothermal Technology, K. Byrappa and Masahiro Yoshimura, eds., 0-8155-1445-X, 894 pp., 2001 Mechanical Alloying for Fabrication of Advanced Engineering Materials, M. Sherif ElEskandarany, 0-8155-1462-X, 260 pp., 2001 Cemented Tungsten Carbides, Gopal S. Upadhyaya, 0-8155-1417-4, 421 pp., 1998 Handbook of Ceramic Grinding and Polishing, Ioan D. Marinescu, Hans K. Tonshoff, and Ichiro Inasaki, eds., 0-8155-1424-7, 601 pp., 1998 Sol-Gel Silica, Larry L. Hench, 0-8155-1419-0, 178 pp., 1998 Supercritical Fluid Cleaning, John McHardy, and Samuel P. Sawan, eds., 0-8155-1416-6, 305 pp., 1998 Ceramic Cutting Tools, E. Dow Whitney, ed., 0-8155-1355-0, 383 pp., 1994 Ceramic Films and Coatings, John B. Wachtman, and Richard A. Haber, eds., 0-8155-1318-6, 463 pp., 1993 Corrosion of Glass, Ceramics and Ceramic Superconductors, David E. Clark, and Bruce K. Zoitos, eds., 0-8155-1283-X, 698 pp., 1992 Handbook of Industrial Refractories Technology, Stephen C. Carniglia, and Gordon L. Barna, 0-8155-1304-6, 651 pp., 1992 Advanced Ceramic Processing and Technology, vol. 1, Jon G. P. Binner, ed., 0-8155-1256-2, 431 pp., 1990 Fiber Reinforced Ceramic Composites, K. S. Mazdiyasni, ed., 0-8155-1233-3, 533 pp., 1990 Friction and Wear Transitions of Materials, Peter J. Blau, 0-8155-1196-5, 490 pp., 1989 Special Melting and Processing Technologies, G. K. Bhat, ed., 0-8155-1202-3, 1034 pp., 1989 Sol-Gel Technology for Thin Films, Fibers, Preforms, Electronics and Specialty Shapes, Lisa C. Klein, ed., 0-8155-1154-X, 429 pp., 1988

Contents

Contributors....................................................................................... Preface...............................................................................................

xv xvii

PART I PROCESSING ....................................................................

1

1 Chemical Synthesis of Nanostructured Particles and Films ........................................................................................ Shi Yu, Cheng-Jun Sun, and Gan-Moog Chow

3

1.1 Introduction .............................................................................. 1.2 Particles .................................................................................... 1.2.1 Nucleation and Growth .................................................. 1.2.2 Dispersion and Agglomeration....................................... 1.2.3 Metals ............................................................................. 1.2.4 Ceramics ......................................................................... 1.2.5 Host-Derived Hybrid Materials...................................... 1.2.6 Surfactant Membrane-Mediated Synthesis .................... 1.2.7 Cytotoxicity of Nanoparticles ........................................ 1.3 Films and Coatings................................................................... 1.3.1 Metals ............................................................................. 1.3.2 Ceramics ......................................................................... 1.4 Summary................................................................................... Acknowledgements ......................................................................... References .......................................................................................

3 5 5 6 9 16 23 25 28 30 30 32 34 35 35

2 Synthesis of Nanostructured Materials by Inert-Gas Condensation Methods ................................................................. 47 C. Suryanarayana and Balaji Prabhu 2.1 2.2 2.3 2.4 2.5 2.6

Introduction ............................................................................ Classification .......................................................................... Synthesis of Nanostructured Materials .................................. Early Studies on Inert-Gas Condensation.............................. The Principle of Inert-Gas Condensation .............................. Evaporation Techniques ......................................................... 2.6.1 Thermal Evaporation.................................................... 2.6.2 Laser Vaporization........................................................ vii

47 47 48 50 51 56 56 57

viii

Contents 2.6.3 Sputtering ..................................................................... 2.6.4 Electrical Arc Discharge .............................................. 2.6.5 Plasma Heating............................................................. 2.7 Particle Transport ................................................................... 2.8 Particle Collection .................................................................. 2.9 Nucleation and Growth .......................................................... 2.10 Limitations of the Classical Nucleation Theory .................... 2.11 Crystal Structure and Morphology ........................................ 2.12 Influence of Process Variables on Particle Size..................... 2.12.1 Inert-gas pressure ....................................................... 2.12.2 Inert-Gas Temperature................................................ 2.12.3 Inert-Gas Type............................................................ 2.12.4 Inert-Gas Flow Rate ................................................... 2.12.5 Evaporation Rate ........................................................ 2.12.6 Nozzle Diameter......................................................... 2.12.7 Chamber Size and Growth Distance.......................... 2.12.8 Effect of the Position of the Evaporation Source...... 2.12.9 Effect of Reactive Gases............................................ 2.13 Advantages of IGC................................................................. 2.14 Drawbacks of IGC ................................................................. 2.15 Recent Developments in IGC ................................................ 2.16 Conclusions ............................................................................ Acknowledgements ......................................................................... References .......................................................................................

58 59 60 60 61 62 64 65 67 68 71 72 73 74 75 75 76 76 78 79 80 83 84 84

3 Thermal Sprayed Nanostructured Coatings: Applications and Developments ......................................................................... 91 George E. Kim 3.1 Introduction .............................................................................. 91 3.2 Thermal Spray Technology ...................................................... 92 3.2.1 Types of Processes ......................................................... 93 3.2.2 Applications.................................................................... 95 3.2.3 Thermal Spray Processes ............................................... 96 3.3 Thermal-Sprayed Nanostructured Alumina–Titania Coating and United States Navy Applications ...................................... 98 3.3.1 Nanostructured Alumina–Titania Coating ..................... 98 3.3.2 United States Navy Applications ................................... 99 3.4 Development and Application of Nanostructured Titania-Based Coating for Industrial Application.................. 102 3.4.1 Background................................................................... 102 3.4.2 Failure Analyses ........................................................... 103

Contents 3.4.3 Case Study—Nanostructured Titania-base Coating of Ball Valve Components for High-pressure Acid Leach Autoclaves ......................................................... 3.5 Conclusions ............................................................................ Acknowledgements ....................................................................... References .....................................................................................

ix

106 116 116 117

4 Nanostructured Materials and Composites Prepared by Solid State Processing ................................................................. 119 H.J. Fecht and Yu. Ivanisenko 4.1 Introduction and Background................................................. 4.2 Phenomenology of Nanostructure Formation........................ 4.3 High-Energy Ball Milling and Mechanical Attrition............. 4.3.1 Examples ...................................................................... 4.3.2 Mechanism of Grain Size Reduction ........................... 4.3.3 Property–Microstructure Relationships........................ 4.4 Phase Stability at Elevated Temperatures .............................. 4.5 Severe Plastic Deformation (SPD)......................................... 4.5.1 General ......................................................................... 4.5.2 High-Pressure Torsion .................................................. 4.5.3 Cold Rolling of Thin Sheets ........................................ 4.5.4 Friction Induced Surface Modifications....................... 4.6 Summary and Outlook ........................................................... Acknowledgements ....................................................................... References .....................................................................................

119 120 122 122 130 135 140 143 143 144 159 161 164 165 165

5 Nanocrystalline Powder Consolidation Methods ..................... 173 Joanna R. Groza 5.1 Introduction ............................................................................ 5.2 Thermodynamics, Mechanisms and Kinetics of Nanocrystalline Powder Densification................................... 5.2.1 Thermodynamic and Kinetic Effects ........................... 5.2.2 Sintering Mechanisms .................................................. 5.2.3 Role of Impurities ........................................................ 5.2.4 Green Density of Nanopowders................................... 5.2.5 Pore Size and Its Effects on the Densification Behavior ....................................................................... 5.2.6 Grain Growth................................................................ 5.3 Methods for Full Densification of Nanopowders .................. 5.3.1 Characterization of Nanomaterials Densification: Density and Grain Size Measurements........................ 5.3.2 Conventional Sintering.................................................

173 175 175 178 185 186 193 196 200 200 202

x

Contents 5.3.3 Pressure Effects in Nanopowder Consolidation .......... 5.3.4 Pressure-Assisted Consolidation Methods................... 5.3.5 Nonconventional Sintering Methods............................ 5.4 Summary................................................................................. Acknowledgements ....................................................................... References .....................................................................................

206 210 213 216 217 217

6 Electrodeposited Nanocrystalline Metals, Alloys, and Composites............................................................................ 235 Uwe Erb, Karl T. Aust, and Gino Palumbo 6.1 Introduction ............................................................................ 6.2 Synthesis of Nanostructured Materials by Electrodeposition .................................................................... 6.3 Structure of Nanocrystalline Metal Electrodeposits .............. 6.4 Properties ................................................................................ 6.4.1 Properties with Weak Grain Size Dependence ............ 6.4.2 Properties with Strong Grain Size Dependence .......... 6.5 Applications............................................................................ 6.5.1 Nanocrystalline Copper for Printed Wiring Boards .... 6.5.2 Nanocrystalline Metals in Microsystem Components .................................................................. 6.6 Summary................................................................................. References .....................................................................................

235 236 239 246 247 253 276 278 280 283 283

7 Computer Modeling of Nanostructured Materials.................. 293 Donald W. Brenner 7.1 Introduction ............................................................................ 7.2 Modeling Methods ................................................................. 7.2.1 Molecular Dynamics and Monte Carlo Modeling....... 7.2.2 Atomic Potential Energies and Forces......................... 7.2.3 Multiscale Modeling .................................................... 7.3 Nanostructured Materials ....................................................... 7.3.1 Nanoparticle Properties ................................................ 7.3.2 Microstructure Modeling.............................................. 7.3.3 Sintering and Grain Growth Dynamics ....................... 7.3.4 Mechanical Deformation and Fracture ........................ 7.3.5 Shock Loading.............................................................. 7.3.6 Vibrational Properties................................................... 7.3.7 Nanoalloys .................................................................... 7.4 Prospects for Future Modeling............................................... References .....................................................................................

293 294 294 296 297 299 299 302 307 311 316 318 319 322 324

Contents

xi

PART II PROPERTIES ................................................................. 329 8 Diffusion in Nanocrystalline Materials ..................................... 331 Wolfgang Sprengel 8.1 8.2 8.3 8.4

Introduction ............................................................................ Modeling of Interface Diffusion ............................................ Diffusion in Grain Boundaries of Metals .............................. Diffusion in Nanocrystalline Metals ...................................... 8.4.1 Specific Aspects............................................................ 8.4.2 Nanocrystalline Pure Metals ........................................ 8.4.3 Correlation between Diffusion and Crystallite Growth ........................................................ 8.4.4 The Nanocrystalline Soft-magnetic Alloys Fe73.5Si13.5B9Nb3Cu1 and Fe90Zr7B3 ............................... 8.4.5 Nanocrystalline Hard-magnetic Nd2Fe14B Compounds................................................................... 8.4.6 Diffusion of Hydrogen in Nanocrystalline Metals ...... 8.5 Diffusion in Nanocrystalline Ceramics.................................. Acknowledgements ....................................................................... References .....................................................................................

331 332 334 335 335 340 344 345 349 350 352 356 356

9 Nanostructured Materials for Gas Reactive Applications ...... 365 Michel L. Trudeau 9.1 Introduction ............................................................................ 9.2 Catalysis and Electrocatalysis ................................................ 9.2.1 Impact of Structure on Catalysis and Electrocatalysis Processes ............................................ 9.2.2 Nanostructure Design ................................................... 9.3 Gas Sensors ............................................................................ 9.3.1 Physical Principles of Semiconductor Sensors............ 9.3.2 Nanostructured Design of Sensing Materials .............. 9.3.3 Novel Nanostructured Systems .................................... 9.4 Hydrogen Storage................................................................... 9.4.1 Properties of Hydrogen-Storage Compounds .............. 9.4.2 Metallic Hydrides ......................................................... 9.4.3 Complex Hydrides........................................................ 9.4.4 Porous or Surface-Related Nanostructures .................. 9.5 Conclusion .............................................................................. Acknowledgements ....................................................................... References .....................................................................................

365 366 367 370 383 385 392 399 401 403 405 412 414 417 418 418

xii

Contents

10 Magnetic Nanoparticles and Their Applications ..................... Sara A. Majetich 10.1 Introduction .......................................................................... 10.2 Fundamental Physics of Magnetic Nanoparticles ............... 10.2.1 Bulk Ferromagnetism............................................... 10.2.2 Magnetic Clusters..................................................... 10.2.3 Molecular Magnetism .............................................. 10.2.4 Ideal Monodomain Particles Made from Ferromagnetic Materials .......................................... 10.2.5 Macroscopic Quantum Tunneling............................ 10.2.6 Surface and Interface Effects in Nanoparticles ....... 10.2.7 Magnetostatic Interactions between Nanoparticles............................................................ 10.2.8 Exchange Interactions Between Nanoparticles........ 10.3 Applications of Monodomain Magnets ............................... 10.3.1 Ferrofluids ................................................................ 10.3.2 Biomedical Applications .......................................... 10.3.3 Imaging with Magnetic Nanoparticles..................... 10.3.4 Data Storage Media.................................................. 10.3.5 Magnetoresistive Devices ........................................ 10.4 Conclusions .......................................................................... Acknowledgements ....................................................................... References ..................................................................................... 11 Magnetic Properties of Nanocrystalline Materials .................. Akihisa Inoue, Akihiro Makino, and Teruo Bitoh 11.1 Introduction .......................................................................... 11.2 Fe—M—B (M = Zr, Hf or Nb) Amorphous Alloys and their Crystallization-Induced Nanostructure........................ 11.3 Soft Magnetic Properties and Structural Analyses of Fe—M—B (M = Zr, Hf or Nb) Nanocrystalline Ternary Alloys ...................................................................... 11.4 Improvement of Soft Magnetic Properties by the Addition of Small Amounts of Solute Elements ................. 11.5 Soft Magnetic Properties and Structure of Cu-free Quaternary Fe—Zr—Nb—B Alloys ..................................... 11.6 Soft Magnetic Properties and Structure of Fe—Nb—B—P—Cu Alloys Produced in Air ...................... 11.6.1 Structure and Soft Magnetic Properties................... 11.6.2 Effect of Grain-Size Distribution and Curie Temperature of Intergranular Amorphous Phase on Soft Magnetic Properties ....................................

439 439 441 441 442 443 444 449 449 451 453 455 455 458 461 462 467 473 474 474 487 487 488

492 499 503 507 507

513

Contents 11.7 Improvement of High-frequency Permeability by the Dissolution of Oxygen in the Surrounding Amorphous Phase..................................................................................... 11.7.1 As-sputtered Structure.............................................. 11.7.2 Magnetic Properties.................................................. 11.8 Applications.......................................................................... 11.9 Conclusions .......................................................................... References .....................................................................................

xiii

520 520 524 530 533 534

12 Mechanical Behavior of Nanocrystalline Metals ..................... 537 Julia R. Weertman 12.1 Introduction .......................................................................... 12.2 Models and Computer Simulations of Mechanical Behavior of Nanocrystalline Materials ................................ 12.2.1 Models of Deformation............................................ 12.2.2 Molecular Dynamics Computer Simulations........... 12.3 Characterization of Nanocrystalline Metals ........................ 12.3.1 Density, Pores and Microcracks............................... 12.4 Mechanical Behavior............................................................ 12.4.1 Elastic Properties of Nanocrystalline Metals........... 12.4.2 Hardness, Yield and Ultimate Strengths .................. 12.4.3 Ductility of Nanocrystalline Metals......................... 12.4.4 Experiments that Shed Light on Deformation Mechanisms.............................................................. 12.5 Conclusions .......................................................................... References .....................................................................................

537 538 538 543 547 548 553 553 555 557 558 560 561

13 Structure Formation and Mechanical Behavior of Two-phase Nanostructured Materials ....................................... 565 Jürgen Eckert 13.1 Introduction .......................................................................... 13.2 Methods of Preparation........................................................ 13.2.1 Rapid Solidification Techniques .............................. 13.2.2 Mechanical Attrition................................................. 13.2.3 Devitrification of Metallic Glasses .......................... 13.3 Phenomenology of Nanostructure Formation and Typical Microstructures........................................................ 13.3.1 Rapidly Solidified Materials .................................... 13.3.2 Conventional Solidification and Devitrification of Bulk Samples ....................................................... 13.3.3 Mechanically Attrited Powders................................

565 567 567 569 574 580 581 598 617

xiv

Contents 13.4 Mechanical Properties at Room and Elevated Temperatures ........................................................................ 13.4.1 Al-Based Two-Phase Nanostructured Alloys........... 13.4.2 Mg-Based Amorphous and Nanostructured Alloys ....................................................................... 13.4.3 Zr-Based Alloys........................................................ 13.4.4 Ti-Based Alloys........................................................ 13.4.5 Mechanically Attrited Composites........................... 13.5 Summary and Outlook ......................................................... Acknowledgements ....................................................................... References .....................................................................................

628 629 635 641 651 654 662 664 665

14 Nanostructured Electronics and Optoelectronic Materials .... 677 Raphael Tsu and Qi Zhang 14.1 Introduction .......................................................................... 14.2 Physics of Nanostructured Materials ................................... 14.2.1 Quantum Confinement: Superlattices and Quantum Wells ......................................................................... 14.2.2 Dielectric Constant of Nanoscale Silicon ................ 14.2.3 Doping of a Nanoparticle......................................... 14.2.4 Excitonic Binding and Recombination Energies..... 14.2.5 Capacitance in a Nanoparticle ................................. 14.2.6 Structure, Bonds and Coordinations of Si Nanostructure: Porous Si and Si Clusters................ 14.3 Applications.......................................................................... 14.3.1 Porous Silicon .......................................................... 14.3.2 Photoluminescence in nc-Si/SiO2 Superlattices....... 14.3.3 Luminescence from Clusters.................................... 14.3.4 Semiconductor/Atomic/Superlattice: Si/O Superlattice ...................................................... 14.3.5 Amorphous Silicon/Oxide Superlattice ................... 14.3.6 nc-Si in an Oxide Matrix ......................................... 14.3.7 Electronic Applications of Si/O Superlattices ......... 14.3.8 Single Electron Transistor........................................ 14.3.9 Quantum Dot Laser.................................................. 14.4 Challenges in Quantum Dot Devices................................... 14.5 Epilogue................................................................................ Acknowledgements ....................................................................... References .....................................................................................

677 677 677 678 680 682 685 688 691 691 692 694 695 699 700 702 704 708 711 712 713 713

Index .................................................................................................... 719

Contributors

Karl T. Aust University of Toronto Toronto, Ontario, Canada

Yu. Ivanisenko University of Ulm Ulm, Germany

Teruo Bitoh Akita Prefectural University Yurihonjo, Japan

George E. Kim Perpetual Technologies, Inc. Quebec, Canada

Donald Brenner North Carolina State University Raleigh, NC, USA

Carl Koch North Carolina State University Raleigh, NC, USA

Gan-Moog Chow National University of Singapore Kent Ridge, Singapore

Sara Majetich Carnegie Mellon University Pittsburgh, PA, USA

Jürgen Eckert IFW Dresden, Institute for Complex Materials Dresden, Germany

Akihiro Makino Tohoku University Sendai, Japan Gino Palumbo Integran Co. Toronto, Ontario, Canada

Uwe Erb University of Toronto Toronto, Ontario, Canada

Balaji Prabhu University of Central Florida Orlando, Florida

Hans Fecht University of Ulm Ulm, Germany

Wolfgang Sprengel University of Stuttgart Stuttgart, Germany

Joanna Groza University of California, Davis Davis, CA, USA

Cheng-jun Sun National University of Singapore Kent Ridge, Singapore

Akihisa Inoue Tohoku University Sendai, Japan xv

xvi

Contributors

C. Suryanarayana University of Central Florida Orlando, FL, USA

Julia Weertman Northwestern University Evanston, IL, USA

Michel Trudeau Hydro-Quebec Research Institute Varennes, Quebec, Canada

Shi Yu Singapore-Massachusetts Institute of Technology Alliance Singapore

Raphael Tsu University of North Carolina—Charlotte Charlotte, NC, USA

Qi Zhang Advanced Photonix, Inc. Dodgeville, WI, USA

Preface to Second Edition

Introduction Nanostructure science and technology has become an identifiable, if very broad and multidisciplinary, field of research and emerging applications in recent years. It is one of the most visible and growing research areas in materials science in its broadest sense. Nanostructured materials include atomic clusters, layered (lamellar) films, filamentary structures, and bulk nanostructured materials. The common thread to these various material forms is the nanoscale dimensionality, i.e., at least one dimension less than 100 nm, more typically less than 50 nm. In many cases, the physics of such nanoscale materials can be very different from the macroscale properties of the same substance. The different, often superior, properties that can then occur are the driving force behind the explosion in research interest in these materials. While the use of nanoscale dimensions to optimize properties is not new, as will be outlined below, the present high visibility and definition of the field is mainly attributable to the pioneering work of Gleiter and coworkers in the early 1980s [1]. They synthesized nanoscale grain size materials by the in situ consolidation of atomic clusters. The study of clusters preceded this work by researchers such as Uyeda [2]. The International Technology Research Institute, World Technology Division (WTEC), supported a panel study of research and development status and trends in nanoparticles, nanostructured materials, and nanodevices between 1996 and 1998. The main results of this study have been published [3] and formed one of the drivers for the U.S. National Nanotechnology Initiative. This report attempted to cover the very broad field of nanostructure science and technology and included assessments of the areas of synthesis and assembly, dispersions and coatings, high surface area materials, functional nanoscale devices, bulk nanostructured materials and biologically related aspects of nanoparticles, nanostructured materials, and nanodevices. A conclusion of the report is that while many aspects of the field existed well before it was identified as a field in the last decade, three related scientific/technological advances have made it a coherent area of research. These are: xvii

Preface

xviii

1. New and improved synthesis methods that allow control of the size and manipulation of the nanoscale “building blocks,” 2. New and improved characterization tools for study at the nanoscale (e.g., spatial resolution, chemical sensitivity), and 3. Better understanding of the relationships between nanostructure and properties, and how these can be engineered. With the recent intense interest in the broad field of nanostructure science and technology, a number of books, articles, and conference proceedings have been published on this broad topic. A partial listing of these publications is given in the bibliography below, starting with the review of Gleiter in 1989. A two-fold justification was given for another book in this rapidly advancing field in the preface of the first edition. These were, first that since many areas of the field are moving rapidly with increased understanding from both experiment and simulation studies, it would appear useful to record another “snapshot” of the field. This justification is certainly true for the second edition since in the over four years since the first edition was published, many new advances have occurred and the updated chapters reflect them. The second justification for the first edition was that because this field is so broad, the book has been designed to focus mainly on those areas of synthesis, characterization, and properties relevant to application that require bulk, and mainly inorganic materials. An exception was the article by Tsu and Zhang on electronic and optoelectronic materials. The exceptions in this second edition are the updated chapter by Tsu and Zhang on the above area, and a new chapter on magnetic nanoparticles and their applications by Majetich. Before a brief description of the updated chapters, the new chapters, changes in some authorship, and the organization of the book is presented, a historical perspective will be given to suggest how the field has developed and what new information has been provided by reaching the limit of the nanoscale.

Historical Perspective Nanoscale microstructural features are not new, either in the natural world or in materials engineering. There are examples of nanoscale ferromagnetic particles found in microorganisms, e.g., 50 nm Fe3O4 in the organism A. magnetotactum [4]. A number of examples exist of improvement in mechanical properties of structural materials when a fine microstructure was developed. Early in the last century, when

Preface

xix

“microstructures” were revealed primarily with the optical microscope, it was recognized that refined microstructures, for example, small grain sizes, often provide attractive properties such as increased strength and toughness in structural materials. A classic example of property enhancement due to a refined microstructure—with features too small to resolve with the optical microscope—was age-hardening of aluminum alloys. The phenomenon, discovered by Alfred Wilm in 1906, was essentially explained by Merica, Waltenberg, and Scott in 1919 [5], and the microstrutural features responsible were first inferred by the x-ray studies of Guinier and Preston in 1938 [6]. With the advent of transmission electron microscopy (TEM) and sophisticated x-ray diffraction methods, it is now known that the fine precipitates responsible for age-hardening, in Al-4% Cu alloys, for example, are clusters of Cu atoms—Guinier-Preston (GP) Zones—and the metastable partially coherent Θ′ precipitate [7,8]. Maximum hardness is observed with a mixture of GPII (or Θ″, coarsened GP zones) and Θ′, with the dimensions of the Θ′ plates typically about 10 nm in thickness by 100 nm in diameter. Therefore, the important microstructural feature of age-hardened aluminum alloys is nanoscale. Critical length scales often determine optimum properties which are structure sensitive. Mechanical properties such as strength and hardness are typical, and as above, microstructural features such as precipitates or dispersoids are most effective when their dimensions are nanoscale. In ferromagnetic materials, the coercive force has been found to be a maximum if spherical particles (e.g., Fe3C in Fe) which act as domain wall pinners have a diameter about equal to the domain wall thickness, i.e., about 50 nm [9]. Similarly, in type II superconductors, it has been found that fluxoid pinning, which determines the magnitude of the critical current density, is most effective when the pinning centers typically have dimensions of the order of the superconducting coherence length for a given material. For the high field superconductors, the coherence length is usually about 10–20 nm, and indeed the commercial superconductors have pinning centers that approximate these dimensions. In Nb3Sn, the grain boundaries are the major pinning sites and optimum critical current densities are obtained when the grain sizes are about 50 nm [10]. Many other examples could be given of the long term use of nanoscale materials in fields such as catalysis.

Organization As was done in the first edition of this book, Part I covers the important synthesis/processing methods for the production of nanocrystalline

xx

Preface

materials. Part II focuses upon selected properties of nanostructured materials. Potential, or existing, applications of nanocrystalline materials are described as appropriate throughout the book. Chapter 1, “Chemical Synthesis of Nanostructured Particles and Films,” by Yu, Sun, and Chow, is an updated version of Ch. 1 of the first edition by Chow and Kurihara. The chemical methods for nanoparticle synthesis described include aqueous, polyol, sonochemical, precursor, organometallic, hydrolysis, solvothermal, and sol-gel methods. The cytotoxicity of nanoparticles is discussed in this updated chapter. Other methods discussed are host-derived hybrid materials, surfactant membrane mediated synthesis, and a variety of films and coatings. Chapter 2, “Synthesis of Nanostructured Materials by Inert-Gas Condensation Methods,” by Suryanarayana and Prabhu, is new in this edition. It covers the important technique that was used by Gleiter and co-workers that stimulated the present field of nanocrystalline materials. This chapter reviews the principles of the inert-gas condensation method, explains the synthesis of nanophase materials via this technique, and discusses the process parameters that influence the constitution and particle size of the product phase. Chapter 3 by Kim is an updated version of the chapter by Lau and Lavernia now entitled “Thermal Sprayed Nanostructured Coatings: Applications and Development.” Thermal sprayed nanostructured coatings are a prime example of a method that has already matured to the point of application. Thermal sprayed nanostructured oxide coatings in particular have been shown to be practically advantageous for both military and industrial applications. After reviewing the technology of thermal spray a number of applications are described. Chapter 4 by Fecht and Ivanisenko, “Nanostructured Materials and Composites Prepared by Solid State Processing,” is the updated version of the chapter by Fecht in the first edition. The methods described, such as mechanical attrition and other severe plastic deformation methods have become popular methods to produce nanocrystalline materials from the “top-down.” The promise to scale up from laboratory to industrial quantities is one of the advantages of these methods. The mechanisms believed responsible for this nanocrystalline synthesis as well as the stability of the nanocrystalline microstructures at elevated temperatures are reviewed. A major problem with nanocrystalline materials made in particulate form is the requirement for consolidation into bulk for most applications. Chapter 5 by Groza is an updated version of her chapter in the first edition. “Nanocrystalline Powder Consolidation Methods” are reviewed and include conventional sintering methods as well as a variety of full-density

Preface

xxi

consolidation techniques. The challenge in processing nanocrystalline powders is to fully densify them without losing the initial metastable features (nanoscale grain size and, sometimes, metastable phases). The stateof-the-art in consolidation of nanocrystalline powders is presented and the remaining challenges are discussed. While chapters 1, 2, and 4 describe processing methods for nanocrystalline materials that result in particulates that require subsequent compaction, i.e. “two-step” processing, there are one-step processing methods available that eliminate the need for compaction with its attendant problems. A notable and commercially attractive one-step method is electrodeposition. Pioneers in this field, Erb, Aust, and Palumbo, update their former chapter into chapter 6, “Electrodeposited Nanocrystalline Metals, Alloys, and Composites.” This chapter describes the processing methods as well as the structure and properties of the electrodeposited nanostructured materials. Recent breakthroughs in the mechanical and magnetic properties of electrodeposited nanocrystalline materials are presented. A variety of applications for electrodeposited nanocrystalline coatings are reviewed. Computer simulation of nanomaterials comprises “virtual processing” and so was included in Part I of the first edition. The chapter in the first edition was written by Professor Phil Clapp who has subsequently retired. Chapter 7, now entitled “Computer Modeling of Nanostructured Materials,” has been written by Professor Donald Brenner. This chapter describes the various modeling techniques including molecular dynamics and Monte Carlo modeling, atomic potential energies and forces, and multiscale modeling. The modeling of nanoparticle properties, microstructure, sintering and grain growth dynamics, mechanical deformation, and nanoalloys are reviewed. Part II of the book deals with selected properties of nanocrystalline materials. Chapter 8, “Diffusion in Nanocrystalline Materials,” is an updated version of the chapter by Wurschum, Brossmann, and Schaefer in the first edition. This chapter is written by Sprengel, a colleague of the former authors. This chapter reviews the data for diffusion in nanocrystalline materials. It describes modeling of interface diffusion, diffusion in grain boundaries of metals, and then gives examples of diffusion behavior for a variety of nanocrystalline materials including pure metals, soft magnetic materials, hard magnetic materials, ceramics, and diffusion of hydrogen in nanocrystalline metals. Chapter 9, “Nanostructured Materials for Gas Reactive Applications,” is an updated version by Trudeau of his chapter in the first edition which brings in important results since the first chapter was written. This large

xxii

Preface

important field is reviewed with examples from catalysis and electrocatalysis, semiconductor gas sensors, and hydrogen storage materials. Of special interest is the sensitivity to nanocrystalline structure. It is speculated that reducing the nanocrystallite size below 10 nm may have more dramatic effects on such properties as catalysis. Chapter 10 is a new chapter in this edition. Majetich reviews “Magnetic Nanoparticles and Their Applications.” After a brief introduction to the phenomenon of ferromagnetism, an in-depth description of the physics of monodomain ferromagnetic particles is given. Applications based upon magnetic nanoparticles are discussed and include such topics as magnetic recording media, spin valve devices, and tunnel junction structures as possible magnetic random access memory. Both current and future applications based on magnetic nanoparticles are described in terms of their basic properties, and the material challenges are identified. Chapter 11 by Inoue, Makino, and Bitoh is an updated and expanded version of the chapter by Inoue and Makino in the first edition. “Magnetic Properties of Nanocrystalline Materials” focuses on the soft magnetic properties of bulk ferromagnetic nanocrystalline alloys prepared by the crystallization of amorphous precursors. The formation of nanogranular bcc and amorphous structures in the Fe—Zr—Nb—B—Cu, Fe—Zr— Nb—B, Fe—Nb—B—P—Cu, and Fe—Hf—O systems are described along with their superior soft magnetic properties and their engineering applications. Chapter 12, “Mechanical Behavior of Nanocrystalline Metals” by Weertman, is an updated review of this dynamic field of research. She brings in the new results from both experimental studies and the simulation of mechanical behavior by molecular dynamics calculations. An experimental breakthrough is the observation in some nanocrystalline materials of both high strength and good ductility. Computer simulation has allowed access to the smallest nanocrystalline grain sizes that are difficult to attain experimentally without the introduction of processing artifacts. Chapter 13, “Structure, Formation, and Mechanical Behavior of TwoPhase Nanostructured Materials” by Eckert, is updated from his chapter in the first edition. The methods used to produce bulk two-phase nanostructured materials are described. The mechanical behavior of such materials is then discussed. Of special interest in this updated chapter is the report of research from the author’s laboratory of enhanced plasticity in a Ti-base alloy with a nanocrystalline matrix and micron-scale ductile dendrites. This material exhibited both high strength and good ductility.

Preface

xxiii

The subject of “functional” nanostructured materials for electronic and optoelectronic applications is a large and important area. While this field is not stressed in this book, it was felt that a chapter outlining some of the important features of this area should be included. Tsu and Zhang have updated their chapter from the first edition, entitled “Nanostructured Electronic and Optoelectronic Materials.” Functional nanocrystalline materials, typically thin films or quantum dots, are covered. An in-depth treatment of several topics related to Si semiconductors is given. This includes the physics of nanostructured materials which covers the dielectric constant, the capacitance, doping and exiton binding energies of a nanoparticle. Possible devices requiring nanoscale features are described. Such devices are light emitting diodes (LEDs) and quantum field effect transistors (QD-FETs).

References 1. Gleiter, H., Progress in Materials Science, 33:223–315 (1989). 2. Uyeda, R., Progress in Materials Science, 35:1–96 (1991). 3. Siegel, R.W., Hu, E., and Roco, M.C., (eds), Nanostructure Science and Technology, Kluwer Academic Publishers, Dordrecht, Netherlands (1999). 4. Kirschvink, J.L., Koyayashi-Kirschvink, A., and Woodford, B.J., Proc. Nat’l Acad. Sci., USA, 89:7683–7687 (1992). 5. Merica, P.D., Waltenburg, R.G., and Scott, H., Bulletin AIME, June: 913 (1919). 6. Guinier, A., Nature, 142:569 (1938); Preston, G.D., ibid, 570. 7. Silcock, J.M., Heal, T.J., and Hardy, H.K., J. Institute of Metals, 82:239 (1953–54). 8. Cohen, J.B., Metall. Trans. A., 23A:2685 (1992). 9. Swisher, J.H., English, A.T., and Stoffers, R.C., Trans. ASM, 62:257 (1969). 10. Scanlan, R.M., Fietz, W.A., and Koch, E.F., J. Appl. Phys., 46:2244 (1975).

Bibliography Gleiter, H., Nanocrystalline Materials, Progress in Materials Science, 33:223–315 (1989). Siegel, R.W., Nanostructured Materials-Mind Over Matter, NanoStructured Materials, 3:1 (1993). Hadjipanayis, G.C., and Siegel, R.W., Nanophase Materials: SynthesisProperties-Applications, Kluwer Press, Dordrecht, Netherlands (1994). Gleiter, H., Nanostructured Materials: State of the Art and Perspectives, NanoStructured Materials, 6:3 (1995). Edelstein, A.S., and Cammarata, R.C., (eds.), Nanomaterials: Synthesis, Properties, and Applications, Institute of Physics, Bristol (1996).

xxiv

Preface

Suryanarayana, C., and Koch, C.C., Nanostructured Materials, in NonEquilibrium Processing of Materials, edited C. Suryanarayana, Pergamon, Elsevier Science Ltd., Oxford, UK (1999) p. 313. Dekker Encyclopedia of Nanoscience and Nanotechnology, Marcel Dekker Inc., New York, NY (2004).

Carl C. Koch Raleigh, North Carolina

October, 2006

PART 1 PROCESSING

1 Chemical Synthesis of Nanostructured Particles and Films Shi Yu,1 Cheng-Jun Sun,2 and Gan-Moog Chow1,2 1

Molecular Engineering of Biological and Chemical Systems, SingaporeMassachusetts Institute of Technology Alliance (SMA), Singapore

2

Department of Materials Science and Engineering, National University of Singapore, Singapore

1.1 Introduction The performance and properties of materials depend on atomic structure, composition, microstructure, defects, and interfaces which are controlled by thermodynamics and kinetics of the synthesis and processing. Nanostructured materials, often characterized by a physical dimension (such as particle size or grain size) of less than 100 nm, attract much interest due to their unique properties compared to conventional materials. Current advances in synthesizing and processing of functional materials for high technology emphasize the bottom-up approach to assemble atoms, molecules, and particles, from the atomic or molecular scale to the macroscopic scale. The tailor-designed arrangement of atoms from the nanoscale to the macroscale for optimized properties may be realized by materials chemistry. Increasing recent interests have been found in chemical synthesis and processing of nanostructured materials.[1–2,3,4,5,6,7,8,9,10,11,12,13,14] Chemical synthesis of materials may be conducted in solid, liquid, or gaseous state. The traditional solid-state approach involves grinding and mixing of solid precursors, followed by heat treatment at high temperatures to facilitate diffusion-controlled chemical reactions to obtain the final products. Mixing and grinding steps are usually repeated throughout the heating cycle, with great efforts to mix materials at the nanoscale and provide fresh surfaces for further chemical reactions. Grain growth, if not prevented, occurs at elevated temperatures resulting in undesirable large grain size. Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 3–46 © 2007 William Andrew, Inc.

3

4

Processing

Material diffusion in liquid or gas is, advantageously, many orders of magnitude larger than in the solid phase, allowing for synthesis of nanostructures at lower temperatures. Reduced reaction temperatures discourage detrimental grain growth. Many functional materials can be synthesized in aqueous or nonaqueous solutions. Water, for example, is one of most common solvents. There are three general classes of aqueous reactions: acid/base reaction, precipitation, and reduction/oxidation (redox). The reactants may be solids, liquids, or gases in any combination, in the form of single elements or multi-component compounds. A multi-element compound often acts as precursor where the components of the final product are in a “mixture” with atomicscale mixing. Many precursors may be prepared by precipitation. In precipitation, the mixing of two or more reactant solutions leads to formation of insoluble precipitate or a gelatinous precipitate. Caution and care must always be taken in handling reactants and precursors, reaction by-products and post-reaction wastes, particularly when complex and hazardous chemicals are involved. Special procedures may be required to remove any impurities from the products and to avoid postsynthesis contamination. Although many laboratory-scale reactions may be scaled up economically to produce large quantities of materials, the laboratory-scale reaction parameters are not necessarily linearly related to those of large-scale reactions. Parameters such as temperature, pH, reactant concentration, and time ideally should be correlated with factors such as supersaturation, nucleation and growth rates, surface energy, and diffusion coefficients, in order to ensure the reproducibility of reactions. Chemistry is based on the manipulation of atoms and molecules, and indeed has a very long history in the synthesis of materials comprising nanostructures. The fields of colloids and catalysts are such examples. The recent popularity of “nanoscience” not only revitalized the use of many “old” chemical methods, but also motivated many “new” and “modified” ones to be developed for the synthesis of nanostructured materials. The wide scope of chemical synthesis and processing of nanostructured materials spans structural, optical, electronic, magnetic, biological, catalytic, and biomedical materials. A comprehensive review of every aspect of this field is not possible in this chapter. The previous overview chapter[15] has been revised here and updated, with addition of new topics. Some of the topics covered in the earlier review are not addressed here. The current overview highlights current advances in chemical synthesis of nanostructured particles and films. Selected examples mainly are metals, ceramics, and hybrid materials with novel magnetic and optical properties for highdensity magnetic data storage and biomedical applications. The synthesis of monodisperse nanoparticles is emphasized due to the importance of

1: Chemical Synthesis, Shi et al.

5

size-dependent properties and feasibility of particle organization to form two-dimensional and three-dimensional superlattices. For other materials, interested readers are encouraged to consult the chapter in the previous edition.[15] This chapter is organized according to the class of material and type of synthetic approach. However, due to the fact that many advanced materials are hybrid and are prepared using multidisciplinary techniques, a clear distinction is not always possible. Cited references,[1–14] archival journals,[16] and latest conference proceedings may be consulted for further details.

1.2 Particles 1.2.1 Nucleation and Growth In a solution or mixture, chemical reagents or precursors react to form stable nuclei followed by the growth of particles. Reactants can be solids or liquids and sometimes gases. Aqueous or nonaqueous solvents are used. Precipitation of solids in solution has been well studied.[17,18] For coprecipitation of multicomponent particles, attention is required to control the conditions to achieve chemical homogeneity of the final product. Different ions may precipitate under different conditions of pH and temperatures with different solubility product constants. After a reagent such as a reducing or oxidizing agent is added to the reactant solution or mixture, chemical reactions occur and the solution becomes supersaturated with the product. The supersaturation drives the chemical system to deviate from the minimum free energy configuration. The state of thermodynamic equilibrium is restored by condensation of nuclei of the reaction product. Homogeneous nucleation does not involve foreign species as nucleating aids. Heterogeneous nucleation however allows formation of nuclei on foreign species. Kinetic factors compete with the thermodynamics of the system in a growth process.[19] Factors such as reaction rates, transport rates of reactants, accommodation, removal, and redistribution of matter compete with influences of thermodynamics in particle growth. The reaction and transport rates are affected by concentration of reactants, temperature, pH, and order of introduction of and degree of mixing of reagents. The structure and crystallinity of particles may be influenced by reaction rates and impurities. Factors such as supersaturation, nucleation and growth rates, colloidal stability, and recrystallization and aging processes have effects on the particle size and microstructure. Supersaturation generally shows predominant influence on the morphology of precipitates. At low

6

Processing

supersaturation, the particles are small, compact, and well-formed, and the shape depends on crystal structure and surface energies. At high supersaturation, large and dendritic particles form. At even higher supersaturation, smaller but compacted, agglomerated particles form.[18] The interface-controlled growth of a small particle in solution becomes diffusion-controlled after the particle exceeds a critical size.[20]

1.2.2 Dispersion and Agglomeration If the formation of all nuclei occurs at nearly the same time in a supersaturated solution, subsequent growth of the nuclei results in formation of particles with a very narrow size distribution, provided that subsequent secondary nucleation does not occur.[21] Homogeneous nucleation as a single event requires the use of proper concentrations of reagents. Foreign nuclei should be removed before reaction to prevent heterogeneous nucleation that may otherwise result in a wide size distribution of particles. A narrow size distribution may be maintained as long as agglomeration and Ostwald ripening of particles do not occur in solution. The synthesis of stable colloids and dispersion of agglomerated particles have been extensively investigated.[22] Colloids and sols refer to the dispersion of particles (with particle sizes less than 100 nm) within a continuous fluid matrix. The small particles approach and then separate from each other by Brownian motion, and as a result, settling out of particles from solution does not occur. Note that random agglomeration between particles may still occur by Brownian motion. Agglomerates of small particles or particles with size larger than 100 nm tend to settle out of solution. In aqueous solvents, particles with a surface oxide layer or a hydrated surface may become charged under appropriate conditions. Electrostatic repulsion, with a force proportional to the inverse of second power of separation distance, occurs between two particles carrying the same charge. The attractive van der Waals force is proportional to the inverse of the distance with an exponent of 3–6. The net attractive or repulsive force between the particles in such a suspension is the sum of the electrostatic repulsion and the attractive van der Waals forces. The DLVO theory (Derjaguin, Landau, Verwey, and Overbeek) describes the effects of attraction and repulsion of particles as a function of separation distance.[23] On the DLVO plot of potential energy versus the separation distance of particles, there exists a positive potential energy peak, which separates the negative potential energy of the primary minimum and secondary minimum. The height of the potential energy peak must be ≥25 millivolts

1: Chemical Synthesis, Shi et al.

7

(corresponding to the thermal energy of Brownian motion at 20°C) at ambient conditions, in order for a dispersion of particles to remain stable. In an appropriate solvent, an electric double layer is formed surrounding the particle. The stable distance of particle separation depends not only on the charges of particles, but also the concentration of other ions in the diffuse region of the double layer. When there is a sufficient number of such ions or ions with multiple charges in the diffuse layer, the charge repulsion will be neutralized. The collapse of the double layer leads to particle contacts and agglomeration.[23] Nanostructured particles possess large surface areas and often form agglomerates as a result of attractive van der Waals forces and the tendency of the system to minimize the total surface or interfacial energy. Coagulation refers to the formation of strong, compact aggregates (corresponding to the primary minimum on the DLVO plot of potential energy versus particle separation), and flocculation refers to the formation of a loose network of particles (corresponding to the secondary minimum on the DLVO plot). Agglomeration of particles may occur during any of the following stages: synthesis, drying, handling, or subsequent processing. In many applications and processing where dispersed particles or stabilized dispersions are required, undesirable agglomeration in each synthesis and processing step must be prevented. To produce unagglomerated particles, surfactants may be used to control the dispersion during chemical synthesis, or disperse as-synthesized agglomerated fine particles. A surfactant, usually an organic compound, lowers the surface or interfacial tension of the medium in which it is dissolved. A surfactant is a surface-active agent that need not be completely soluble and may decrease surface or interfacial tension by spreading over the surface. It has an amphipathic structure in that solvent, i.e., a lyophobic (solvent repulsive) and lyophilic group (solvent attractive). Surfactants are classified as anionic, cationic, zwitterionic (bearing both positive and negative charges), or non-ionic (bearing no charges). The effectiveness of a surfactant is measured by the maximum reduction in surface or interfacial tension by the surfactant, whereas, surfactant efficiency refers to the surfactant concentration that is needed to reduce the surface or interfacial tension by a certain amount from that of the pure solvents. For example, water and oil may be dispersed in each other provided a suitable surfactant is used to stabilize the microemulsion. The surfactant establishes itself at and defines the boundary between the two liquids. The relative quantity of a surfactant determines the amount of surface that can be covered and, therefore, the extent to which the size and number of droplets of one liquid is dispersed in the other. When the major component is apolar (oil),

8

Processing

the dispersion is one in which the water (polar) phase forms the droplets or reverse micelles. The polar head group of the surfactant is pointing inward toward the water phase while the hydrocarbon tail is pointing outward into the oil phase. The radius of the water droplet is related to the amount of water and surfactant. Repulsive interparticle forces are needed to prevent particle agglomeration during synthesis. A common method is to disperse the particles by electrostatic repulsion resulting from interactions between the electric double layers surrounding the particles. This may be achieved by adjusting the pH of the solution or adsorbing charged surfactant molecules on the particle surfaces. Such stabilization is generally effective in dilute systems of aqueous or polar organic media, and is very sensitive to the pH and effects of other electrolytes in the solution. At the isoelectric point, the pH where the particles have no net surface charges, agglomeration may occur. The isoelectric point varies for different materials. In most nonaqueous solvents without significant ionization, electrostatic repulsion has a lesser contribution to stabilization of particles. Another approach to dispersion involves the steric forces produced by adsorbed surfactant on particle surfaces. The lyophilic, nonpolar chains of surfactant molecules extend into the solvent and interact with each other. The interactions between nonpolar chains are subject to much less van der Waals attraction and provide a steric hindrance to interparticle approach. To optimize steric stabilization, the size of surfactant molecules must be large enough to be a barrier without entangling each other. When the particles approach one another, the stretched-out lyophilic chains of the adsorbed surfactant are forced into a smaller spatial confinement. This interaction leads to a thermodynamically unfavorable decrease of the entropy of the system, thus, the particles will be prevented from approaching each other by this entropic repulsion. Entropic stabilization becomes even more significant when the temperature of the dispersion is increased. Steric stabilization may occur in the absence of electric barriers and is effective in both aqueous and nonaqueous media. It is also less sensitive to impurities or trace additives than electrostatic stabilization and is particularly effective in dispersing high concentrations of particles. Dry, high-surface-area powders agglomerate by van der Waals forces and hydrogen bonds. When these agglomerates need to be used in a dispersed form during subsequent processing, deagglomeration can be achieved by breaking the agglomerates using methods such as milling or ultrasonication in an appropriate solvent containing a suitable surfactant for dispersion.[22] The deagglomerated powders may then be carried in a

1: Chemical Synthesis, Shi et al.

9

liquid for further processing such as injection molding and polymer-based casting. Though surfactants may be used to stabilize particles against agglomeration, the presence in the final product may adversely affect the properties of the materials. Subsequent removal of surfactants from the particle surface could result in undesirable agglomeration and particle growth under certain conditions. Careful consideration is warranted in optimizing the properties and control of particle dispersion.

1.2.3 Metals Interest in nanostructured metals and semiconductors arises from their unique electronic and physical properties and diverse high-tech applications including high-density magnetic recording, catalysis, pharmaceuticals, and medical diagnosis. Many references are available in the literature.[24,25,26] Here, emphasis is given to recent advances in this area. Aqueous Methods. Water has a high permittivity which makes it a good solvent for polar or ionic compounds. Therefore, many chemical reactions occur in aqueous media. Precious, elemental metal nanoparticles for catalytic and biomedical applications, such as Au, Ag, Pt, and Pd nanoparticles may be prepared by adding liquid reducing agents to aqueous solutions of respective salts in the presence of a stabilizer. The choice of reducing agents may drastically affect the nucleation rate and particle growth, which in turn influences the particle size and size distribution. Commonly used reducing agents include sodium borohydride, hydrazine, sodium citrate, and alcohols.[25,27] For example, crystalline Ag nanowires were prepared by reducing AgNO3 with sodium citrate in the presence of NaOH at 100°C. The quantity of NaOH was an important factor in determining the morphology of the final product.[28] Sizecontrolled synthesis of chemically clean Ag nanoparticles was also reported using reduction of silver oxide by hydrogen gas in water. Particles with a diameter between 15 and 200 nm were synthesized by varying the reaction time. The advantages of this method are the easy scale-up for production of naked particles with long-term stability.[29] However, the reaction must be handled with great caution because hydrogen gas is explosive when mixed with air in concentrations larger than 4%. The particle size and morphology of metal nanoparticle may be controlled by choosing suitable capping agents and varying the ratio

10

Processing

of concentrations of capping agent to metal salts. For example, monodisperse silver nanocubes with a mean edge length of 55 ± 5 nm were synthesized in water by a modified silver mirror reaction at 120°C. nHexadecyltrimethylammonium bromide (HTAB), as an ionic surfactant, played a key role in the formation of the nanocubes. Its micelles directed the silver metal to nucleate and grow into nanoparticles other than the usual silver mirror. An increase in the molar ratio of HTAB/[Ag(NH3)2]+ led to an obvious shape evolution of Ag nanoparticles from spheres to cubes, due to the anisotropic adsorption (molar ratio dependent) of the surfactant on the silver crystal faces.[30] Although the aqueous approach to making metal powders is not new, its use in the synthesis of metal nanoparticles requires special attention to avoid undesirable contaminated products. Impurities such as salts and other reaction by-products may not be completely removed, even by repeated washing procedures, if they are entrapped inside the particles or agglomerates during a fast and ill-controlled reaction. Because of the high reactivity of metal nanoparticles due to large surface area, special care must be taken during washing and filtering of the nanoparticles to avoid undesirable hydrolysis or oxidation. Subsequent drying often requires vacuum-assisted procedures to avoid oxidation. Polyol Method. The polyol method has been used to make finely dispersed single elemental metal particles such as Cu, Ni, Co, and others in the micron and submicron size range.[31–32,33,34] In this method, precursor compounds such as oxides, nitrates, and acetates are either dissolved or suspended in ethylene glycol, diethylene glycol, or 1,2-propanediol. The polyol acts as both solvent and reducing agent. The mixture is heated to reflux between 160°C and 194°C. During the reaction, the precursors are reduced and metal particles precipitate out of solution. Submicron size particles may be synthesized by increasing the reaction temperature or inducing heterogeneous nucleation via the addition of foreign nuclei or forming foreign nuclei in-situ. A higher temperature favors the nucleation step and this, in turn, favors the monodispersity of particles when more nuclei are formed. Nanocrystalline particles such as Ni, Co, Pt, Ag, Au, Co—Ni, and Fe—Ni have been recently synthesized using this method.[35–36,37,38,39,40,41,42] The morphology of nanostructured metal synthesized by the polyol method may be controlled by choosing appropriate capping agents. Facetselective capping agents promote the abundance of a particular shape by selectively interacting with a specific crystallographic facet via chemical adsorption. A typical example is the shape-controlled polyol synthesis of Ag nanoparticles. The primary reaction involves the reduction of AgNO3

1: Chemical Synthesis, Shi et al.

11

with ethylene glycol in the presence of a capping agent poly(vinyl pyrrolidone) (PVP) at 160°C. The morphology and dimensions of Ag are found to strongly depend on reaction conditions such as temperature, the concentration of AgNO3, and the molar ratio between the repeating unit of PVP and AgNO3. When the temperature is reduced to 120°C or increased to 190°C, the product is dominated by nanoparticles with irregular shapes. To obtain Ag nanocubes, the initial concentration of AgNO3 has to be higher than 0.1 M, otherwise Ag nanowires are the major product. Silver nanocubes of various dimensions can be obtained by controlling the growth time.[36,43] Moreover, the addition of a small amount of hydrochloric acid improves the monodispersity and shape perfection of synthesized Ag nanocubes. It is suggested that hydrochloric acid plays an important role in selectively etching and dissolving twinned silver nanoparticles that form at the earlier stages of reaction. The presence of protons further slows down the reduction reaction and thereby facilitates the formation of single-crystal seeds.[44] Nanostructured metal with different morphologies may also be synthesized by controlling the reduction kinetics in the polyol process. For example, morphological control over Pt nanoparticles is realized by varying the amount of NaNO3 added to a polyol process, where H2PtCl6 is reduced by ethylene glycol to form PtCl42− and Pt at 160°C. As the molar ratio between NaNO3 and H2PtCl6 is increased from 0 to 11, the morphology of Pt nanoparticles evolves from irregular spheroids with rounded profiles to tetrahedra and octahedra with well-defined facets. It is proposed that nitrate is reduced to nitrite by PtCl42− in the early stage of the synthesis, and the nitrite can then form stable complexes with both Pt(II) and Pt(IV) species. As a result, the rate of reduction of Pt precursors by ethylene glycol is significantly reduced. The change in reaction kinetics alters the growth rates associated with different crystallographic directions of the Pt nanocrystals and ultimately leads to formation of different morphologies.[39] The presence of a trace amount of Fe species (FeCl2 or FeCl3) in a polyol synthesis can also alter the growth kinetics of Pt nanostructures and hence morphology. Depending on the way the Fe species and oxygen (from air) are supplied to the reaction system, Pt nanostructures in the form of spheres, star-shaped particles, branched multipods, and nanowires are prepared as the major product for each run of synthesis.[40] Metal nanoparticles have also been synthesized using a modified polyol process. The modification includes addition of other solvents and sodium hydroxide. In the synthesis of monodisperse Co nanoparticles, cobalt acetate tetrahydrate, oleic acid, and diphenylether (DPE) are mixed and

12

Processing

heated to 200°C under N2. Trioctylphosphine (TOP) is added at 200°C, and the mixture is heated to 250°C. When a solution of 1,2-dodecanediol in DPE is injected into the mixture at 250°C, the color of the reaction solution changes from blue to black, indicating the formation of Co nanoparticles. After size-selective precipitation, monodisperse Co nanoparticles are obtained. The particle size of Co nanoparticles can be controlled by tailoring the concentration or composition of stabilizers. For example, increasing the concentration of oleic acid and TOP by a factor of 2 yields nanoparticles having average diameters between 3 and 6 nm, while substituting tributylphosphine (TBP) for TOP increases the average size to 10–13 nm. Substituting cobalt acetate tetrahydrate with nickel acetate tetrahydrate, using oleic acid, TBP and tributylamine (TBA) as stabilizers, Ni nanoparticles with average diameters ranging from 8 to 13 nm are synthesized. When a mixture of cobalt acetate tetrahydrate and nickel acetate tetrahydrate is used, Co/Ni alloy nanoparticles form.[45] Monodisperse FePt nanoparticles with an average particle size of 3 nm are synthesized by heating a solution of platinum(II) acetylacetonate, iron(II) acetylacetonate, 1,2-hexanedecanediol, oleic acid, and oleylamine in octylether solution at 286°C for 30 min.[46] Compared with commonly used ethylene glycol or glycerol, using long-chain 1,2-diols may prevent the formation of insoluble particles because the precursors and generated nanoparticles are well dispersed during reaction. In the preparation of Ni—Fe nanoparticles in ethylene glycol (EG) (Fig. 1.1), strong alkaline solution is essential for disproportionation of Fe. The addition of NaOH results in a competition between the formation of metal hydroxide and that of metal EG complex. In the Figure 1.1 A TEM micrograph of presence of NaOH, the metal FeNi nanoparticles synthesized by the polyol method. Reprinted with hydroxide formation is predominpermission from [47]. Yin, H., and ant with only a small amount of Chow, G. M., Electroless Polyol complex precipitated from solution. Deposition of FeNi-based Powders The formation of Ni(OH)2 inhibited and Films, J. Mater. Res., the reduction of Ni(II) to Ni, 18:180–187 (2003). © 2003, Materials Research Society. because Ni(OH)2 has a more nega-

1: Chemical Synthesis, Shi et al.

13

tive reduction potential than Ni(II). On the other hand, the formation of Fe(OH)2 leads to disproportionation of Fe(II), which is an alternative way to synthesize Fe. This disproportionation can significantly increase the yield of Fe. The composition of Ni—Fe particles is independent of reaction time. However, the particles with reaction time of 60 min show small size with narrow size distribution compared to particles obtained in shorter or longer time. It was suggested that the formation of a metal–EG complex acts as a reservoir of solute and prevents further metal particle nucleation during the growth process.[35,47] Compared to aqueous methods, the polyol approach results in the synthesis of metal nanoparticles protected by surface-adsorbed glycol, thus minimizing the problem of oxidation. The use of a nonaqueous solvent such as the polyol also further reduces the problem of hydrolysis of fine metal particles that often occurrs in the aqueous case. Sonochemical Methods. Ultrasound has been used in chemical synthesis of nanostructured materials. High energy sonochemical reactions, without any molecular coupling of the ultrasound with the chemical species, are driven by the formation, growth, and collapse of bubbles in a liquid. This acoustic cavitation involves a localized hot spot of temperature of about 5000 K, a pressure of ∼1800 atm and a subsequent cooling rate of about 109 K/s, due to the implosive collapse of a bubble in the liquid.[48] Volatile precursors in low-vapor-pressure solvents are used to optimize the yield. Ultrasonic irradiation is carried out with an ultrasound probe, such as a titanium horn operating at 20 kHz. When water is sonicated, the very high temperatures and pressures of collapsing gas bubbles lead to thermal dissociation of water vapor into *OH and H* radicals.[49] In the presence of a primary alcohol (RCH2— OH; R==H or alkyl group), the formation of RCHOH* radicals during sonication was proposed. Metal nanoparticles may be prepared by reducing prospective salts with the reducing species H* and RCHOH* radicals. For example, Au nanoparticles in the size range of 9–25 nm are synthesized by reduction of tetrachloroauric(III) acid in the presence of aliphatic alcohols and sodium dodecyl sulfate in aqueous solutions using 20 kHz ultrasound. The temperature of the reaction solution was maintained at 20 ± 5°C. The extent of reduction of AuCl4− is found to be dependent on the concentration of the alcohol at the bubble–water interface. The particle size can be tailored by varying the alcohol concentration in solution and the hydrophobicity of the alcohol. The particle size decreases with increasing alcohol concentration and alkyl chain length. It is suggested that as alcohol is adsorbed onto Au in aqueous solution, the adsorbed molecules stabilize the particles at a smaller size and prevent further growth of the

14

Processing

colloids. The amount of alcohol adsorbed depends on the alcohol used and its concentration.[50] Other noble-metal nanoparticles with a narrow size distribution, such as Pt and Pd, are prepared by sonochemical reduction of aqueous solution containing H2PtCl6 or K2PdCl4 in the presence of PVP as a capping agent.[51] Precursor Methods. The conventional approach to preparing alloys and composites is to first grind and mix the solid precursors using some mechanical means, and then carry out appropriate chemical reactions to obtain final products. The communition and mixing in the solid state are generally limited to the submicron level. Consequently, material diffusion in chemical synthesis is limited to this spatial scale, which has a direct influence on the time and temperature of reactions and the final chemical homogeneity of the product. With great efforts such as high energy milling, solid state mixing of constituents at the atomic scale is possible. If the precursors are mixed at the atomic or molecular level, the synthetic reactions may then be carried out at shorter times and reduced temperatures due to the shorter distance for material diffusion. Intimate contact of constituents at the atomic scale also provides a better means to control the stoichiometry and homogeneity of the final product. These advantages motivate the synthesis of precursor materials which have the constituents as atomic neighbors (for example, as in a compound). These precursors are subsequently subjected to thermochemical reactions to synthesize alloys and composites with improved properties compared to the same materials obtained by traditional solid-state reactions. Organometallic Methods. An organometallic compound is an organic compound containing a metal, in which a metal atom is bonded directly to a carbon atom. Organometallic compounds are advantageous chemical precursors since the constituents, in molecular proximity to each other, may be decomposed at relatively low temperatures to form the final product desired. The biggest disadvantage of this approach is that most of the reactions involve air-sensitive reactants, therefore, a glove box or schlenck line technique must be used. Because of the air-sensitive nature of some of the reactants, greater care must also be taken in preparation of solvents and the choice of atmospheres. Generally, organometallic routes produce only small amounts of material. There has been a renewed interest in the synthesis of monodisperse magnetic nanoparticles with unique properties such as high magnetization. Monodisperse Co nanoparticles are prepared by pyrolysis of Co2(CO)8 in diphenylether (DPE). An as-prepared solution of Co2(CO)8 in DPE is injected into a hot solution containing DPE and surfactants

1: Chemical Synthesis, Shi et al.

15

(oleic acid and tributylphosphine) at ∼200°C under N2. Thermal decomposition of Co2(CO)8 results in nucleation of Co and release of CO gas. The reaction solution was held at 200°C for 15–20 min, allowing the nanoparticles to grow. The solution was cooled, and the nanoparticles were isolated by size-selective precipitation. Generally, a higher temperature and larger ratio of metal-precursor to surfactants produces larger nanoparticles. Fe nanoparticles are synthesized by replacing Co2(CO)8 with Fe(CO)5 in such a procedure at ∼250°C.[45] Face centered tetragonal (fct) FePt alloys have potential application in high-density magnetic information storage due to the large uniaxial magnetocrystalline anisotropy and good chemical stability. The chemical synthesis strategy combines reduction of platinum acetylacetonate by 1,2-hexadecanediol and thermal decomposition of iron pentacarbonyl. Both chemical reactions are initiated by refluxing a mixture containing metal precursors, dioctylether, oleic acid and oleylamine at ∼300°C for 30 min under airtight conditions. The composition of the resulting alloys can be adjusted by controlling the molar ratio of Fe(CO)5 to platinum salt. For example, a 3 : 2 molar ratio of iron pentacarbonyl to platinum acetylacetonate gives Fe48Fe52, and a 2 : 1 molar ratio produces Fe52Pt48. The particle size of FePt alloys can be tailored to be from 3 to 10 nm in diameter with a standard deviation of less than 5%. This involves first growing 3 nm seed particles in situ followed by adding more reagents to allow the existing seeds to grow to the desired size.[52] However, the assynthesized FePt nanoparticles have a face centered cubic (fcc) structure and are superparamagnetic. To obtain the ordered fct phase (the so-called Ll0 structure), the as-synthesized nanoparticles typically have to be heated to ≥550°C. Heat treatment at these temperatures leads to undesirable agglomeration of particles and a dramatic increase in particle size. Direct synthesis of fct FePt nanoparticles is reported using Collman’s reagent, Na2Fe(CO)4, as a reducing agent for Pt(II). In this method, a 1 : 1 molar ratio of platinum acetylacetonate to Na2Fe(CO)4, and a surfactant oleylamine are sonicated and then refluxed in tetracosane at 389°C under an inert atmosphere. Magnetic measurements of samples produced directly in solution show a coercivity of 1300 Oe at 290 K and 3100 Oe at 10 K.[53] Monodisperse CoPt3 nanocrystals are synthesized via simultaneous reduction of platinum acetylacetonate and thermodecomposition of cobalt carbonyl in different coordinating mixtures in the presence of 1-adamantanecarboxylic acid. The average particle size can be varied from 1.5 to 7.2 nm by controlling reaction conditions and types of coordinating mixture. As-synthesized CoPt3 particles are single crystal, however, with chemically disordered fcc structure.[54]

16

Processing

1.2.4 Ceramics Chemical methods such as precipitation[12,55] and sol–gel processing can be used to synthesize ceramic nanoparticles. As-synthesized powders, depending on the synthesis technique used, may require subsequent heat treatment for dehydration, removal of organics, and controlled crystallization to form oxides with desirable structure and crystallite size. Hydrolysis. Precipitation from solution generally involves formation of an insoluble hydroxide which can then be converted to its oxide by heat-assisted dehydration. Metal (hydrous) oxide particles are synthesized by forced hydrolysis involving controlled deprotonation of hydrated cations. For example, by heating suitable metal salts with a defined amount of water in diethylene glycol (DEG), various metal oxides such as CoO, SiO2, TiO2, Fe2O3, ZnO, and Nb2O5 nanoparticles with a mean particle diameter between ∼30 and 200 nm are obtained. The concentrations of the metal precursor and water are important in controlling the particle size.[59–60,61] Water in the metal hydrates is used as an alternative water source for the polyol-mediated preparation. Transition metal ferrite nanoparticles are synthesized by heating transition metal hydrates in DEG in the presence of alkaline hydroxide under an argon atmosphere. Complexation of the transition metal cations with DEG in the presence of alkaline hydroxide enables control over the rate of their hydrolysis. The growth of nanoparticles can be terminated by adding long-chain carboxylic acid, which binds to their surface as a capping ligand.[62,63] Magnetite nanoparticles can also be synthesized by hydrolysis of hydrated ferric salt in 2-pyrollidone at the boiling temperature of 2-pyrollidone under a nitrogen atmosphere. It is proposed that thermal decomposition of 2-pyrrolidone results in carbon monoxide and azetidine. Azetidine catalyzes the hydrolysis of FeCl3·6H2O to form ferric oxide hydroxide (FeOOH). The FeOOH is then partially reduced by CO and dehydrates to form Fe3O4.[64] Organometallic Methods. A complete separation of nucleation from growth is crucial for synthesis of monodisperse nanoparticles. A general scheme for preparing monodisperse (standard deviation, s < 5%) ceramic nanoparticles involves producing a single, short nucleation event followed by slower growth on the nuclei formed. Semiconductor nanoparticles are synthesized by injecting thermally unstable organometallic precursors into a hot solution containing colloidal stabilizers.[65–66,67,68,69,70,71,72,73] Size selective precipitation may be used to provide monodisperse nanoparticles. For synthesis of CdTe nanoparticles, CdO powder is dissolved in [13,56–57,58]

1: Chemical Synthesis, Shi et al.

17

tetradecylphosphonic acid (TDPA) and trioctylphosphine oxide (TOPO) at 300°C under Ar flow. The solution is cooled to 270°C, and tellurium stock solution (Te powder dissolved in TOP) is injected. After injection, nanoparticles grow at 250°C to reach the desired size. Replacing the tellurium stock solution with selenium or sulfur stock solution, CdSe or CdS, results in nanoparticles with narrow size distribution (standard deviation, ∼10 %) without any size-sorting.[67] The shape of CdSe and other semiconductor nanocrystals can be controlled, producing dots, rods, rice-shaped particles, tetrapods, or other elongated shapes. The monomer concentration in the growth solution is the determining factor in shapecontrol and shape-evolution. Elongated shapes can be transformed into spherical shapes if the monomer concentration in the solution is reduced to a certain level, whereas spherically shaped nanoparticles can grow to elongated shapes by simply increasing the monomer concentration.[74,75] This injection-based synthetic method, however, is not suitable for producing high-quality nanoparticles on a large scale (e.g. tens of kilograms), since the rapid injection of precursors is very difficult to achieve. Moreover, the limitations of mass transfer in a big reactor further diminish the merits of the injection method. To overcome this obstacle, a single-step synthetic method that does not require the injection of precursors has been developed for the synthesis of high-quality CdS nanoparticles. Heating at 120°C with gentle stirring under vacuum (30 mTorr), cadmium acetate hydrate, sulfur, myristic acid, tetraethylthiuram disulfides, and 2,2′dithiobisbenzothiazole are dissolved in octadecene (ODE). The resulting solution is then heated to 240°C under Ar flow. Small nanoparticles appear when the reaction temperature reaches 240°C. As the particles grow, a narrow size distribution (standard deviation, ∼7%) can be achieved. There is no detectable new nucleation occurring during particle growth. The narrow particle size distribution can be maintained for at least 12 h. It was suggested that tetraethylthiuram disulfides and 2,2′-dithiobisbenzothiazole play an important role in achieving this separation of nucleation from growth by affecting the reactivity of the precursors (i.e., cadmium myristate and S). In addition, the greater amount of 2,2′-dithiobisbenzothiazole leads to a smaller number of stable nuclei. With the same amount of precursors, this results in nanoparticles with a larger final size.[76] Another successful approach to achieve a complete separation of nucleation from growth is the large-scale synthesis of monodisperse iron oxide nanoparticles without using the size-sorting process via thermal decomposition of iron-oleate complex.[77–78,79,80] The iron–oleate complex is prepared by reacting iron chlorides and sodium oleate. For the synthesis of monodisperse iron oxide nanoparticles, the iron–oleate complex and oleic

18

Processing

Figure 1.2 12-nm magnetite nanocrystals. The TEM image clearly demonstrates that the nanocrystals are highly uniform in particle-size distribution. Inset is a photograph showing a Petri dish containing 40 g of the monodisperse magnetite nanocrystals, and a US one-cent coin for comparison. Reprinted with permission from [79], Park, J., An, K. J., Hwang, Y. S., Park, J. G., Noh, H. J., Kim, J. Y., Park, J. H., Hwang, N. M., and Hyeon, T., Ultralarge-scale Syntheses of Monodisperse Nanocrystals, Nat. Mater., 3:891–895 (2004). © 2004 Nature Publishing Group.

acid are dissolved in octadecene at room temperature. The reaction mixture is heated to and maintained at 320°C for 30 min. The resulting solution containing iron oxide nanoparticles is then cooled to room temperature, and ethanol is added to the solution to precipitate the nanoparticles (Fig. 1.2).[79] Through demonstrating one-nanometer-scale size-controlled synthesis of monodisperse iron oxide nanoparticles, it has been proposed that the iron–oleate complex exclusively contributes to growth, instead of participating in both nucleation and growth processes.[80] Solvothermal Method. In solvothermal techniques, the reaction mixture is heated above the boiling point of the solvent in an autoclave or other closed system and the sample is exposed to steam at high pressures. The reactions may be carried out in water or in any other solvent (e.g. methanol, ethanol, polyol). When water is used as a solvent, the process is described as hydrothermal. Compared with synthesis routes at atmospheric pressure, the increased reaction temperature in the solvothermal technique may lead to an accelerated crystal growth accom-

1: Chemical Synthesis, Shi et al.

19

panied by a narrow particle size distribution and better crystallinity. For example, the precipitation of CdTe under hydrothermal conditions yields nanoparticles with a narrow size distribution and enhanced photoluminescence (PL) quantum yield (QY). Hydrothermally treated, a mixture of NaHTe and CdCl2 solution in the presence of thiols as stabilizing agents at a desired temperature (160 and 180°C), leads to formation of cubic zinc blende structured CdTe nanoparticles with a size distribution of about 10%. The average particle size could evolve from 2 to 4 nm by extending the reaction time. As-synthesized CdTe nanoparticles with a PL QY of up to 35% were directly used as biological labels without any postpreparative treatment.[81] A general strategy utilizing the solvothermal technique was reported as a unified approach to the synthesis of a large variety of nanoparticles including noble metal, ceramic, and polymer nanoparticles.[82] In this method, sodium linoleate (or another sodium stearate), linoleic acid (or another fatty acid), and ethanol are added to the aqueous solution of metal ions in an autoclave tube under agitation. At a certain temperature, it is proposed that a cation exchange between the metal ion and sodium linoleate results in formation of metal linoleate. Decomposition of the insitu generated metal linoleate, under designated reaction conditions, can yield oxide (e.g. TiO2 and ZnO), ferrite (e.g. Fe3O4 and CoFe2O4), and titanate (e.g. BaTiO3 and SrTiO3) nanoparticles. Alternatively, in the presence of other anion species such as S2− (S2− was supplied by Na2S or (NH4)2S), Se2− (Se2− was generated by the reduction of SeO32− by N2H4) or F− (F− was provided from NaF or NH4F), various functional nanoparticles, such as CdS, MnS, PbS, Ag2S, CuS, ZnS, CdSe, ZnSe, YF3, LaF3, or NaYF4, are effectively synthesized. Sol–Gel Methods. As early as the mid-1800s, it was reported that silicon tetrachloride, when left standing in an open container, hydrolyzed and turned into a gel.[83–84] After this time, biologists did much work with gels and colloids. In the early 1930s, aerogels were discovered.[85] Since the 1950s, sol–gel techniques have been used for phase equilibrium studies that have led to synthesis and processing of ceramics.[86–87,88,89] Sol–gel processes can be used to make various types of materials such as powders, films, fibers, and monoliths. The conventional sol–gel process involves hydrolysis and condensation of metal alkoxides. Metal alkoxides have the general formula M(OR)x and an alkoxide ion is the conjugate base of an alcohol. The general synthesis of metal alkoxides involves the reaction of metal species (a metal, metal hydroxide, metal oxide, or metal halide) with an alcohol. Metal alkoxides are good precursors because they readily undergo hydrolysis that replaces an alkoxide with a hydroxide

20

Processing

group from water and a free alcohol is formed. Once hydrolysis has occurred the sol can react further and condensation (polymerization) occurs, leading to gel formation. In condensation two hydrolyzed fragments join together, releasing either an alcohol or water. Condensation occurs by either nucleophilic substitution or nucleophilic addition. The parameters to be controlled in a sol–gel process are solvent, temperature, precursors, catalysts, pH, additives, and mechanical agitation. These factors can influence the kinetics, growth reactions, hydrolysis, and condensation reactions.[58] The solvent influences the kinetics and conformation of the precursors, and the pH affects the hydrolysis and condensation reactions. Acidic conditions favor hydrolysis, which means that fully or nearly fully hydrolyzed species are formed before condensation begins. Under acidic conditions there is a low crosslink density which yields a denser final product when the gel collapses. On the other hand, basic conditions promote condensation reactions to begin before completion of hydrolysis. The pH also affects the isoelectric point and the stability of the sol, in turn, influencing the aggregation and particle size. By controlling the reaction rates of hydrolysis and condensation, the structure and properties of the gel may be tailored. Sol–gel methods can be used to prepare pure, stoichiometric, dense, equiaxed, and uniform particles. For example, uniform anatase TiO2 particles with average particle size of 5 to 30 nm were prepared by controlled hydrolysis of a titanium triethanolamine (TEOA) complex.[90,91] It was found that the pH value of the sol–gel was a decisive factor for controlling the final size, shape, and phase of the particle. The effects of surface chemistry on the morphology and phase stability of titanium dioxide nanoparticles was investigated using a thermodynamic model based on surface free energies and surface tensions. It was suggested that surfaces representing acidic and alkaline conditions had a significant influence on both the shape of the nanocrystals and the anatase-to-rutile transition size.[92] Nanostructured AlN powders for thermal management in electronics applications were synthesized by nitridation of oxide precursor powders.[93,94] The oxide powders were prepared by hydrolysis of aluminum tri-sec-butoxide at room temperature to favor precipitate formation instead of gel formation. The dry precursor (AlOOH) powders were calcined and subsequently nitrided in ammonia at temperatures up to 1100°C for 10 h. Since small particles favored the diffusion-controlled nitridation kinetics, nanostructured AlN powders were synthesized at temperatures of 400 to 600°C lower than that used in conventional carbothermal nitridation or direct nitridation of large-grained oxide or aluminum

1: Chemical Synthesis, Shi et al.

21

powders. Nitride formation was favored when the oxide precursor powders were either amorphous or highly disordered. Commercially available nanocrystalline alumina powders could not be fully nitrided under the same nitridation conditions used for the sol–gel derived amorphous precursors. As-synthesized AlN powders were consolidated by pressureless sintering, showing better densification compared to commercial coarse-grained AlN powder under the same conditions. However, fully dense nanostructured AlN bulk solids could not be obtained by pressureless sintering due to adverse agglomeration of powders with a mixed pore size. The sol–gel process is particularly attractive for synthesis of multicomponent particles with binary or ternary compositions using double alkoxides (two metals in one molecule), or mixed alkoxides (with mixed metaloxane bonds between two metals). Atomic homogeneity is not easily achieved by coprecipitating colloidal hydroxides from a mixture of salt solutions since it is difficult to construct double metaloxane bonds from metal salts.[57] Hybrid materials such as metal-oxide, organics-oxide can be prepared using the sol–gel approach. For example, controlled nanoheterogeneity can be achieved in metal/ceramic nanocomposites.[95] A solution of metal salt, (RO)3Si(CH2)3A (A is functional organic group) and Si(OR)4 is used in sol–gel processing. The ethylene diamine derivative such as (RO)3Si(CH2)3NHCH2CH2NH2(DIAMO) is used to form stable complexes with most transition metals. The complexation of metal ions and anchoring of the resulting metal complexes to the oxide matrix takes place. The oxidation of the metal complex containing gels is then carried out at high temperatures and results in formation of composites of nanoscale metal oxide particles in an oxide matrix. Reduction of metal oxide particles in hydrogen provides metal–ceramic nanocomposite powders such as Cu (<5 nm) in silica. The metal particles, a few nanometers in size with a very narrow size distribution even for high metal loading, are statistically distributed in the oxide matrix without any agglomeration, as a result of anchoring the metal complexes to the oxide matrix. The narrow particle size distribution could not be achieved if the sol–gel processing is performed without complexation of metal ions. Nanocomposite silica glassy layers (host) containing copper-based species (guest) have also been prepared by the sol–gel route, using ethanolic solutions of tetraethoxysilane (Si(OC2H5)4, TEOS) and copper(II) acetate (Cu(CH3COO)2·4H2O) in a single-step process, followed by subsequent annealing under different atmospheres. A homogeneous distribution of copper-containing clusters in the sol–gel silica matrix is observed.

22

Processing

This opens the possibility of obtaining thin films of silica–copper nanocomposites with tailored compositional and microstructures.[96] Hybrid organic–inorganic materials can also be prepared by the sol–gel approach.[97,98,99,100,101–102] These materials can be divided into three types: 1. organics used to impregnate a porous oxide gel; 2. organics incorporated into the liquid sol–gel solution and subsequently trapped in the oxide matrix after gelation; and 3. organics which react with the sol–gel precursors in solution and subsequently form a chemical bond to the oxide after gelation.[103] Other nanostructured hybrid materials such as nanoconfined proteins and enzymes are prepared by encapsulation of biomolecules in sol–gel processing.[104,105] The dopant biomolecules serve as templates for the formation of self-specific pores around them. The stable biomolecules are accessible to the external environment due to the porous nature of the oxide matrix, and biorecognition of the hybrid materials is feasible.[105] Multicomponent gels can be thermochemically converted to form nanocomposite nitride powders. For example, a precomposite gel, prepared by reacting iron chloride hexahydrate, urea, and boric acid in water under strongly basic conditions at 150°C, is reacted in ammonia at 500°C to obtain FexN/BN (x = 3 or 4) nanocomposite powder.[106] Similarly, AlNxBN100−x (0 ≤ x ≤ 100) nanocomposite powders are synthesized by pyrolysis of the precomposite gels (with aluminum chloride hexahydrate as the Al source) in ammonia.[107] The compositional and thermal effects on these AlN–BN nanocomposite powders were studied.[108] Grain growth of AlN (both normal and abnormal) and BN occurred. Both crystallites retained their nanometer size up to 1600°C. The normal grain growth of AlN phase was impeded by the second phase of immobile inclusion particles of BN. When normal grain growth of AlN was inhibited by inclusions, only large AlN grains with boundary curvatures larger than average could move past the inclusions. As a result, larger grains experienced exaggerated grain growth. These results indicate that initial crystallization of AlN in the composite powders produced grains with a mixed size distribution. Non-Hydrolytic Sol–Gel Methods. In non-hydrolytic sol-gel methods, the traditional hydrolysis and condensation reactions are replaced by direct condensation reactions or an ester elimination reaction.[109] The reaction may be direct condensation of metal halide and alkoxides[110] or condensation of a metal alkoxide and a metal carboxylate. The ester elimination reaction involves reacting an alcohol and a metal acetate[111] or alkoxide[112]. A non-hydrolytic sol–gel process was also used for the synthesis of organic–inorganic hybrids.[113] Non-hydrolytic reactions do not

1: Chemical Synthesis, Shi et al.

23

involve water or polar solvents. In the traditional sol–gel methods, the condensation reaction may be reversible and special efforts are required to completely remove water or polar solvents.

1.2.5 Host-Derived Hybrid Materials Materials with unique microstructures or nanostructures, either found in nature or synthesized, may be used as hosts or structural templates for synthesis of nanoparticles for various applications. Zeolites, aluminosilicates with a three-dimensional cage-like framework of tetrahedra of SiO4 and AlO4, have ordered nanoscale pore systems that may be used as nanoreactors. Examples include stabilization of organic dyes for fabrication of energy transfer and storage systems, preparation of host–guest hybrid catalyst systems and encapsulation of conducting or semiconducting nanoscale wires and clusters. Zeolite crystals may also be used as building blocks for self assembly of complex hierarchical thin films or three-dimensional structures, involving a combination of host–guest interactions (ranging between dispersive, electrostatic, and covalent).[114] Ordered mesoporous silicates are synthesized from calcination of aluminosilicate gels in the presence of surfactants, where silicate forms inorganic walls between ordered surfactant micelles.[115] These materials have high surface areas with easy access to uniform pores that can be further tuned beyond the range of zeolites. Mesoporous materials are now available in the size range of 2 to 50 nm. The synthesis and properties of mesoporpus materials have been reviewed.[116–117,118] The catalytic, optical, and magnetic properties of mesoporous materials with nanoparticles are discussed in reference 85. For catalytic applications, commercial microporous zeolites are limited by their pore sizes. Ordered mesoporous materials with pore diameters of 2–30 nm may show good properties for catalytic conversion of bulky reactants. The catalytic activity and hydrothermal stability are relatively low, limiting their use in industrial catalytic reactions such as petroleum cracking. The relatively low catalytic activity and hydrothermal stability may be related to the amorphous nature of the mesoporous walls. Methods to improve catalytic activity and hydrothermal stability of mesoporous materials have been reviewed.[119] These include supported sulfated zirconia; sulfated zirconia with tetragonal crystalline phase; aluminosilicates synthesized in alkaline media or strongly acidic media; titanosilicates with catalytically active titanium species and high-temperature synthesis of silica-based materials using fluorocarbonhydrocarbon surfactant mixtures.

24

Processing

Self assembly of inorganics, surfactant, and optically active species lead to unique hierarchical structures with higher concentrations of optical materials compared to traditional methods. A high concentration of isolated dye molecules can be incorporated in the mesoporous materials. Ordered pores provide a high surface area for better dispersion of optically active constituents and faster diffusion in optical sensor applications.[120,121] Examples include a fast-response optical pH sensor, host systems for laser dyes, and photochromic molecules. Ordered arrays of nanostructured diluted magnetic semiconductor nanostructures, combining semiconducting and magnetic properties, are potential devices in magneto- or spin electronics. In these materials, the cation A of a binary semiconductor AB is randomly substituted by a magnetic ion M. Mn-doped II–VI semiconductor nanoparticles were synthesized in mesoporous silica. The band-gap energy and magnetic properties depended on the particle size and dopant concentrations.[122] Well-dispersed, uniform CoO (3 nm size) particles were partially embedded in mesoporous silica-based MCM-41 using a one-pot surfactant-assisted aqueous procedure involving cationic surfactant cetyltrimethylammonium bromide (structural directing agent) and molecular atrane complexes of Co and Si (inorganic hydrolytic precursors). Magnetic measurements confirmed the superparamagnetic behavior of CoO, eliminating the possibility of homogeneous distribution of Co(II) ions in the Si framework.[123] Magnetic nanoparticles confined in the pores of these mesoporous materials tend to be superparamagnetic due to small particle size. Uniform α-Fe2O3 particles (120 nm in diameter) were chemically coated with a dense silica layer (tunable thickness from 10 to 50 nm). A mesoporous silica shell was then formed using simultaneous sol–gel polymerization of tetraethoxysilane (TEOS) and n-octadecyltrimethoxysilane (C18TMS) followed by removal of the organic group. The hematite cores were reduced in H2 and N2 to produce Fe3O4/Fe, which were encapsulated within the silica layer with the mesoporous silica shell. The pores were randomly arranged in the mesoporous silica shell. A commercial drug, introduced into the pores with an uptake amount of 12 wt %, could be subsequently released. The room-temperature magnetization curve showed a larger hysteresis, due to the large size of encapsulated magnetic particles.[124] To date, the magnetic properties of mesoporous nanocomposite appear to be less readily exploited for applications as compared to those of the optical and catalytic particles. Other materials, such as self-assembled block copolymer systems, may also be used in self-assembly of materials. For example, well-ordered pho-

1: Chemical Synthesis, Shi et al.

25

tonic-bandgap structures at visible and near-IR frequencies have been produced. The size and shape of periodic microstructures of block copolymers may be controlled by molecular weight and relative composition, and blending with homopolymers or plasticizers. Optical properties of block-copolymer-based photonic crystals may be further enhanced by incorporating inorganic nanoparticles.[125] In meeting the challenges of organizing nanostructures across various length scales by self-assembly, the understanding and mimicking of the process of biomineralization in nature has been proposed as a potential enabling step. Biomineralization occurs via an insoluble organic template or soluble molecules that control crystallization, growth, and self-assembly. The possibilities of biomimetic mineralization, though in its infancy, for the generation of mesoscale order via self-assembly of inorganic nanoparticles has been discussed.[126] For example, in an attempt to prepare L10 FePt for high density magnetic media at low temperatures, a phage display methodology was used to identify peptide sequences that specifically bind to ferromagnetic L10 FePt and control the precipitation of FePt nanoparticles. The gP8 engineered phage (as template) is mixed with FeCl2, and H2PtCl6. The mixture is then reduced by NaBH4. The FePt alloy nanoparticles (∼4 nm in diameter) show some degree of chemical ordering of L10 phase of FePt. These particles are ferromagnetic at room temperature with coercivity of ∼1 kOe.[127]

1.2.6 Surfactant Membrane-Mediated Synthesis Surfactants are commonly used to stabilize nanoparticles or colloids against aggregation in solution. They can adsorb on the surface or form an envelope around the particle to provide either electrostatic or steric repulsion. Surfactants may also be used in post-synthesis processing to disperse the agglomerated particles. Deagglomeration is accomplished by breaking the agglomerates by milling or ultrasonication in a suitable solvent and surfactant.[22] Surfactants, as self-assembled structures can be used as a reactor to synthesize nanoparticles.[128,129] In an appropriate solution, for example, an aqueous medium, surfactant molecules orient themselves so that contact of the nonpolar tails of the molecules with the solvent is minimized. The polar headgroups of the molecules are attracted to water by electrostatic and hydrogen-bond interactions. These interactions allow the molecules to self-assemble into membrane structures with minimum energy configuration. Self-assembly has been used to form monolayer films,

26

Processing

Langmuir–Blodgett films, micelles, reversed micelles, vesicles, and tubules. For example, glycerol monooleate, a lipid with a single fatty acid tail and a glycerol headgroup, forms bicontinuous cubic phases with sizecontrolled aqueous channels that are connected in a three-dimensionally periodic network. Pd nanoparticles (4 nm) were synthesized by a polyol type reaction of a solution of tetrachloropalladate and glycerol monooleate.[130] The hydroxyl groups in the glycerol headgroup reduce the metal salt to metal at room temperature. The narrow aqueous channels of the bicontinuous phase constrained the particle size of Pd. Membrane structures such as reverse micelles and vesicles can be used as nanoreactors to synthesize metal and ceramic nanoparticles.[131,132] Precipitation occurs inside the hollow compartments of the membrane structures. Compared to bulk precipitation, better chemical homogeneity is achieved since the reactions occur in a better controlled environment. The size of the membrane structures can be manipulated to control the particle size and size distribution. The particles are encapsulated inside the membrane and are prevented from agglomeration with other particles by the membrane which acts as a barrier. Since it is possible to incorporate functional groups on vesicle surfaces, nanoparticles may be carried inside functionalized membranes for targeted applications. Reverse micelles (sometimes known as water-in-oil microemulsions) are a single layer of surfactant molecules entrapping solubilized water pools in a hydrocarbon solvent. The size of the water pool depends on the amount of entrapped water at a given surfactant concentration. Nanoscale metal colloids such as Ag[133], Co[134], Au—Ag[135] and semiconductors such as CdS[136], CdSe[137] are made using the reverse micelle method. Ceramic particles can be prepared by sol–gel reactions in microemulsions.[138,139] Monodisperse spherical silica particles containing homogeneously dispersed Ag quantum dots are synthesized via a controlled photochemical reduction of silver ions during hydrolysis of tetraethoxysilane in a microemulsion.[140] The morphology of the composite particles strongly depends on the time of exposure to UV light. Depending on the timing, Ag quantum dots can be directed to different annuli within the SiO2 spheres, as well as onto the SiO2 sphere surfaces. Ternary and quaternary nanostructured mixed oxides such as BaTiO3, BaZrO3, SrTiO3, and BaxSr1−xTiO3 are also synthesized in microemulsions.[141] Vesicles are closed bilayer membrane assemblies. Multilamellar vesicles have diameters in the range of 100–800 nm. Single bilayer vesicles are 30–60 nm in diameter. Sonication or extrusion can be used to control the vesicle size. Extrusion, however, is limited to the preparation of only

1: Chemical Synthesis, Shi et al.

27

a small quantity of vesicles. Vesicles are more stable than micelles. Nanoscale oxide particles may be synthesized using the vesicle-mediated approach. Metal ions and lipid mixtures form vesicles upon sonication. After removal of exogenous ions, anions are added and allowed to diffuse through the membrane layers and intravesicular precipitation occurs. Due to preferential anion diffusion across the membrane, generally only oxides such as Ag2O,[142] Fe3O4,[143] and Al2O3[144] have been prepared. Dispersed nanocrystalline metal particles can be prepared using polymerized phospholipid vesicles.[145,146] The non-cross-linked polymerization of vesicles results in many individual polymer chains in the membrane structure. This enhances the structural integrity of vesicles and provides breaks in the polymer network through which both anions and cations can diffuse across the polymerized membranes (thus removing the restriction to only oxide synthesis as in the case of non-polymerized vesicles). In this approach, vesicles made from mixed phospholipids (negatively charged and zwitterionic) are UV-polymerized. Positively charged Pd ions are selectively and chemically attached to the negatively charged headgroup molecule on the surface of the interior compartment of membrane. After adding Au salt and a reducing agent to the solution, Au ions diffuse across the membrane. The interior membrane-bound Pd species act as catalysts for initiation of autocatalytic electroless metallization of Au, which leads to the formation of nanocrystalline Au particles inside the polymerized vesicles. When unpolymerized vesicles are used to make nanoscale Au particles using this electroless method,[147] Pd catalysts are bound both to the internal and external membrane surfaces and electroless metallization occurrs on external surfaces of the vesicles. The osmotic pressure (built up inside unpolymerized vesicles due to the difference in salt concentrations across the membranes) and the external electroless reaction weaken the structural integrity of these vesicles, which eventually rupture. In this case, the lipid molecules serve as a dispersant during the synthesis of Au particles. The dispersion property of the nanocrystalline Au particles depends on the thermal disordering of the phospholipid mixture, which becomes more effective as a dispersant with increasing temperature. Many multiplytwinned particles (MTP) are formed in reactions carried out at 25°C and 40°C for 3 h (Fig. 1.3). The formation of MTP by rapid quenching in vapor deposition is well known. Vapor deposition is generally a “far-fromequilibrium” process. The condensation of supersaturated vapor into a solid has a maximum departure from equilibrium (ΔG) of approximately 160 kJ/Na, where Na is Avogardro’s number. The precipitation of solid from a supersaturated liquid has a ΔG of 8 kJ/Na.[148] Although the deposition

28

Processing

Figure 1.3 A HRTEM micrograph of gold colloids synthesized using unpolymerized vesicles. An example of an intercalated structure.

kinetics in solution chemistry differ significantly from that of the vapor phase, MTPs of Au can be synthesized in both cases. Semiconductor crystallites of 2–10 nm have bulk-like internal lattices but their optical spectra show resolved discrete features. In bulk species, the electronic energy is continuous but, as the particle size gets below 6 nm, the electronic levels are discrete due to quantum confinement of the carriers. This leads to third-order nonlinear optical effects. Agglomeration of particles must be avoided in preparation of these materials by dispersing the particles in polymers,[149] inside a multilayer,[150,151] pores,[152] and vesicles[153,154]. For example, the vesicle’s internal diameter controls the particle size of cubic CdS. The absorption spectrum of CdS using phoshatidylcholine had exciton energy shifted to shorter wavelengths.[154]

1.2.7 Cytotoxicity of Nanoparticles Besides their inherent chemical composition, the properties of nanoparticles depend not only on their size but also their microstructure and surface coating. These in turn are controlled by the synthesis and processing conditions. Despite their significant scientific interests and promising potential in many applications, the potential environmental and health impacts of nanoparticles have raised concerns, toxicity being a critical factor in evaluating the utility of these materials (especially for in-vivo biomedical applications).[155–156,157,158] While cytotoxicity of a bulk

1: Chemical Synthesis, Shi et al.

29

material might be well documented, more work has to be done to clarify potential cytotoxicity of its nanostructured counterpart with or without surface coating. Few reports have addressed the extent to which particle size and surface coating influence the toxicological properties.[159–160,161,162,163,164,165] For example, CdSe-core semiconductor quantum dots (QDs) may be useful as an alternative to fluorescent dyes for use in biological imaging, due to their bright fluorescence, narrow emission, broad UV excitation, and high photostability. However, their potential cytotoxicity remains unclear. Using primary hepatocytes as a liver model, it was found that CdSe-core QDs were toxic under certain conditions modulated by processing parameters during synthesis, exposure to ultraviolet light, and surface coatings. The data suggested that cytotoxicity correlates with the liberation of free Cd2+ ions due to deterioration of the CdSe lattice. When appropriately coated, CdSe-core QDs can be rendered nontoxic and used to track cell migration and reorganization in vitro.[160] As a capping agent, oleic acid has been widely used in the synthesis of monodisperse nanoparticles and forms a coating on as-synthesized nanoparticles. Using the Neuro-2A cell line as a model, the influence of particle size on cytotoxicity of nickel ferrite nanoparticles with or without oleic acid coating was investigated. For nickel ferrite particles without oleic acid, cytotoxicity was independent of particle size within the given mass concentrations and surface areas accessible to the cells. Oleic-acidcoated nickel ferrite nanoparticles showed significantly higher cytotoxicity than their uncoated counterparts. Moreover, for coated particles, it was noted that large particles (150 ± 50 nm diameter) showed a higher cytotoxicity than smaller particles (10 ± 3 nm diameter). Two possible explanations for this size effect were suggested. First, the surface energy of small particles is higher than that of large particles. The different surface energies may have different effects on the surfactant adsorption and conformation, which may alter the chemical reactivity of the surfactant per unit surface areas of particle. In this way, the same surfactant may behave differently when it interacts with cells. Another possible explanation involves consideration of different interaction areas of large and small particles, as shown in Fig. 1.4. The effective interaction area over which a large particle can access the cell is greater than that for a small particle. Within this specific area, there are more function groups on individual large particles. Thus each large particle exerts a stronger stimulus on the cell. However, if an equal mass of particles with different sizes is considered, the significantly larger number of small particles increases the number of interaction points that are randomly distributed around the cell. However, due to the likelihood of a smaller number of functional groups

Processing

30 Oleic acid Alkyl end and

COOH end

Cell surface

Interaction area

Figure 1.4 Sketch of the various interaction areas of an individual particle with one layer oleic acid coating, in which a single large particle possesses larger interaction areas with more function groups. The large number of small particles increases the number of interaction points that are randomly distributed on the cell, but there are less functional groups on each interaction point. Reprinted with permission from [167], Yin, H., Too, H. P., and Chow, G. M., The effects of particle size and surface coating on the cytotoxicity of nickel ferrite, Biomaterials, 26:5818–5826 (2005). © 2005 Elsevier Science.

on individual small particles, each interaction point exerts a weak stimulus on the cell. The experimental results seem to suggest that the total sum of weak stimuli at different locations was not as toxic as one localized strong stimulus from a bigger particle.[165]

1.3 Films and Coatings 1.3.1 Metals Nanostructured metal coatings can be deposited by aqueous and nonaqueous methods. Both electrolytic and electroless plating have been used. Aqueous Electroplating. In electrodeposition, the deposition of a pure metal or alloy from a electrolyte solution occurs on the cathode when an external current is applied to the plating system. An electrically conductive substrate is required. Nanostructured metal coatings of pure metals, alloys and composites are deposited using electrodeposition.[166–167,168] Nanostructured grains are deposited when the plating variables such as bath composition, pH, temperature, and current density are controlled so that nucleation of new grains is favored over grain growth.

1: Chemical Synthesis, Shi et al.

31

Aqueous Electroless Deposition. In the electroless approach,[171] electrons are generated by chemical reactions without applied external current as in the case of electroplating. Unlike electrodeposition, electrical conductivity of the substrate is not required. Electroless deposition can occur by the following mechanisms: 1. deposition by ion or charge exchange; 2. deposition by contacting the metal to be coated; and 3. autocatalytic deposition on catalytically active surfaces from solutions containing reducing agents. In an autocatalytic electroless process, a noncatalytic surface, on which electroless metal is to be deposited, is initially coated with catalyst particles. The catalyst particles are typically colloids such as Pd (with diameter ≥2 nm) encapsulated in a tin oxide shell. The catalyst is first adsorbed on the substrate surface, followed by chemical stripping of the protective surface oxide to expose the catalytic Pd core. Reduction of a soluble metal ion or complex by a soluble reducing agent present in the plating solution leads to deposition of metal atoms at the surface. Metal ions are reduced by electrons provided by reducing agents. Each deposited metal layer subsequently becomes the catalyst for deposition of the next layer, hence the name autocatalytic metallization. For example, electroless metallization is used to deposit soft magnetic CoNiP thin films (phosphorous or boron was in the composition of the plating solution).[169] Films with optimized soft magnetic properties can be obtained by controlling the pH of the plating bath, rather than by varying the concentrations of metal ions and reducing agent in the bath. While the increase of metal and reducing agent ion concentrations in the plating solution results in an initial increase and subsequent decrease of both magnetic moment and coercivity of the CoNiP films, the increase of pH is found to increase the magnetic moment and simultaneously decrease the coercivity. Using the electroless approach, nanostructured metal coatings are also deposited on self-assembled biomolecular structures. Rhapidosomes, protein tubules which are about 17 nm in diameter and 400 nm in length, are electrolessly metallized using molecular catalysts instead of conventional colloidal catalysts. The tubule surfaces are initially catalyzed by treating with a (PdCl4)2− or Pd(Ac)2 solution. Several amino acid residues such as cysteine, histidine, and tyrosine can reduce the catalyst molecules to Pd crystallites (3 nm) on the rhapidosome surface. Randomly oriented nanocrystals of Ni (about 10 nm) are deposited on the surfaces by this electroless plating method.[170] Self-assembled phospholipid hollow tubules, with an average diameter of 0.5 μm and 50 to 80 μm long, are interesting materials due to their large shape anisotropy. The tubules are

32

Processing

metallized by electroless plating with nanoscale Ni. The Ni-coated tubules are magnetically aligned and cast into a polymer matrix to form a composite. The magnetic anisotropy of electroless metallized Ni or permalloy tubules (with metal grain size about 2–4 nm, depending on metallization conditions) and the alignment of metallized tubules in the magnetic tubule–polymer composite have been investigated.[171] The metallized tubules were used to fabricate an ungated vacuum field emission cathode structure for generation of a macroscopic electron beam current. The local electric field enhancement was achieved by exploiting the large aspect ratio of the metallized tubules, the radius of curvature, and the thickness of metal coatings at the edge of the metallized tubules.[172,173] An aqueous, electroless “oxidative-soak-coating” deposition has been used to fabricate MnO2 and CeO2 films on glass substrates at 313 to 333 K.[174] Film deposition proceeded via oxidation of Mn2+ and Ce3+ ions in homogeneous solutions and then heterogeneous nucleation of MnO2 and CeO2 on the substrates. Nonaqueous Electroless Deposition. Aqueous electrodeposition and electroless deposition are not attractive approaches to metallization of substrates which can be detrimentally hydrolyzed or oxidized in aqueous solution. A nonaqueous coating process capable of producing fine-grained deposits has been recently developed using the polyol method. For example, Cu coatings have been deposited on AlN substrates,[175] and Ni— Co[176] and Ni[177] on Cu substrate. Unlike traditional aqueous electroless metallization, this process does not require the adsorption of catalysts on electrically insulative substrates to initiate metallization. Enhanced saturation magnetization was observed in films with granular nanostructured Ni in Cu(II)—O—C matrix. The saturation magnetization of Ni—Cu complex films at room temperature was 112 emu/g, which was about twice that of bulk Ni (54.4 emu/g). The observed shifted Curie temperature, large anisotropy field, and strong temperature dependence of saturation magnetization seemed to support the hypothesis that the enhancement was caused by the magnetic interaction between the Ni particles and the amorphous Cu complex.[177]

1.3.2 Ceramics Nanostructured ceramic oxide films and coatings can be deposited using sol–gel type methods.[178] Sol–gel processing is especially adaptable for film formation. Films and coatings represent the earliest commercial use of sol–gel processing. Sol–gel techniques offer the following advan-

1: Chemical Synthesis, Shi et al.

33

tages in coatings: control of microstructure, pore size, and surface area. By controlling these parameters, the film properties can be tailored. Thin films use very little raw material and can be processed quickly, and very large and irregularly shaped surfaces may also be coated. Dense pinholefree layers can be prepared at low temperatures using sol–gel processing. Porous films can be prepared by changing the reaction conditions. This approach is particularly useful in obtaining homogeneous multicomponent coatings. To avoid cracking caused by large capillary stress during evaporation of solvent in the drying process, either slow evaporation (a slow process) or supercritical drying (a fast process) is used. As-deposited oxide coatings are typically amorphous. Thermal and thermochemical post-synthesis treatment are needed to obtain nanostructured oxide, carbide, or nitride coatings. Hybrid coatings can be fabricated by doping the sol with material of a different phase, followed by gelation and densification. For example, nanocomposite films of noble-metal nanoparticles in a dielectric matrix such as Au/BaTiO3 demonstrate desirable optical properties. Noble-metal nanoparticles (such as Au) exhibit unique optical properties because of the excitation of Mie plasmon resonances or surface plasmon resonance (SPR). The optical responses at SPR wavelength lead to the use of these nanocomposites in applications including surfaceenhanced Raman spectroscopy, sensors, and all-optical switching devices. Nanocomposite thin films of Au/BaTiO3 with different Au concentrations have also been prepared by a sol–gel process. Gold doping lowers the crystallization temperature of as-synthesized amorphous BaTiO3 and enhances its crystallinity in post-deposition annealing. The Au—BaTiO3 interface was also investigated and no alloying occurred between Au and BaTiO3. The evolution of Au surface plasmon resonance spectra with increasing annealing temperature was observed in 10 mol% Au/BaTiO3 thin films.[179] Sol–gel films may be deposited by spraying, dip coating, and spin coating. Viscosity of the sol can be increased by aging that may affect the deposition. The amount of porosity in a sol–gel-derived film can be controlled by the pH. A higher porosity is achieved when base-catalyzed sols are used. For example, nanoporous thin films of silica are prepared by dipcoating a silica sol onto a substrate and drying at or near room temperature.[180] The film thickness depends on dipping speed, aging, ratio of water to precursor, and pH. Drying the films near room temperature preserves the porosity of the gel. As the ratio of water to TEOS was increased under acidic conditions, there is an increase in film thickness, due to fast hydrolysis under acidic conditions. At a lower ratio of water to precursor,

34

Processing

continued hydrolysis occurrs in the film due to atmospheric moisture. Langmuir–Blodgett techniques may be used to prepare either closepacked or nanoporous TiO2 thin films of a known thickness and porosity. Using successive compression and expansion cycles at different pressures, monolayers with different areal particle density are prepared. The porosity is controlled by controlling the average spacing between the crystallites in the monolayer.[181] A low-voltage electrochemical method for depositing titania thin films from sol–gel monomers with entrapped organic molecules has been reported.[182] The film thickness (ranging between 20 and 1000 nm) can be controlled by changing either the potential or the duration of its application. Because large capillary stress during solvent evaporation can cause cracking, the sol–gel process has been traditionally used for preparing thin films and coatings. To prepare thick coatings, the problems of shrinkage and cracking and the limitation of coating thickness can be mitigated by increasing particle loading in the sol–gel process.[183] This approach involves dispersing large ceramic powders in sol–gel solution, and applying the mixture onto the substrate by various techniques such as dipping and spraying. Coatings with thicknesses up to 200 μm are fabricated. The sol–gel film forms strong bonds to both oxide powders and substrates by interaction with functionalized surface hydroxyl groups on the oxide powders and the oxide layer of substrates. The strong bond reduces cracking. The shrinkage problem associated with the conventional sol–gel approach is minimized due to the high loading of ceramic powders. A twodimensional sol–gel process is also used to fabricate thick films of TiO2.[184] In this process, the traditional sol–gel hydrolysis and condensation reactions take place at an air–water interface. The gel films formed can then be deposited onto substrates using Langmuir–Blodgett techniques. Dense, crack-free, and transparent TiO2 composite films (0.9 μm in thickness) have been fabricated using a methylcellulose-assisted sol–gel process in a single-step at room temperature. Nanostructured TiO2 film prepared by this method is one of promising candidates for applications in areas ranging from optics to gas sensors via solar energy.[185]

1.4 Summary Nanostructured particles, films, and coatings can be synthesized using solution chemistry. The ability to manipulate atoms and molecules in the liquid phase provides a powerful arsenal for synthesis of tailor-designed nanomaterials using a bottom-up approach. Significant progress in chem-

1: Chemical Synthesis, Shi et al.

35

ical synthesis and processing of nanostructured materials has been made. Active transdisciplinary efforts in chemical synthesis and processing of nanostructured materials continue to be pursued. Fundamental understanding of the interfacial interactions of these high-surface area materials, particularly the interfacial stability of hybrid materials, is essential in order to design and control the properties of the materials.

Acknowledgements GMC thank the Office of Naval Research (USA) and Academic Research Funds, National University of Singapore for support of this work. SY and CJS were supported by the SMA and NUS Research Fellowships, respectively.

References 1. Ke, Y. C. and Stroeve, P., Polymer-Layered Silicate and Silica Nanocomposites, Elsevier, Boston (2005). 2. Yang, P. D., (ed.), The Chemistry of Nanostructured Materials, World Scientific, Singapore, London, (2003). 3. Nalwa, H. S., (ed.), Handbook of Organic–Inorganic Hybrid Materials and Nanocomposites, American Scientific Publishers, California (2003). 4. Baraton, M. I., (ed.), Synthesis, Functionalization and Surface Treatment of Nanoparticles, American Scientific Publishers, California (2003). 5. Nalwa, H. S., (ed.), Nanoclusters and Nanocrystals, American Scientific Publishers, California (2003). 6. Koch, C. C., (ed.), Nanostructured Materials: Processing, Properties and Potential Applications, New York, Noyes Publications (2002). 7. Klabunde, K. J., (ed.), Nanoscale Materials in Chemistry, WileyInterscience, New York (2001). 8. Ying, J. Y., (ed.), Nanostructured Materials, Academic, San Diego (previously published as: Advances in chemical engineering, v. 27.) (2001). 9. Hodes, G., (ed.), Electrochemistry of Nanomaterials, Wiley-VCH, Weinheim (2001). 10. Chow, G. M., Ovid’ko, I., and Tsakalakos, T., (eds.), Nanostructured Films and Coatings, NATO Science Series, 3. High Technology, Vol. 78, Kluwer Academic Publishers, Dordrecht (2000). 11. Fendler, J. H., (ed.), Nanoparticles and Nanostructured Films: Preparation, Characterization and Application, Wiley-VCH, Weinheim (1998). 12. Chow, G. M., and Gonsalves, K. E., Particle Synthesis by Chemical Routes, (A. S. Edelstein, and R. C. Cammarata, eds.), Nanomaterials: Synthesis, Properties, and Applications, Institute of Physics Publishing, Bristol and Philadelphia, pp. 55–71 (1996).

36

Processing

13. Klein, L. C., Processing of Nanostructured Sol-Gel Materials, (A. S. Edelstein, and R. C. Cammarata, eds.), Nanomaterials: Synthesis, Properties, and Applications, Institute of Physics Publishing, Bristol and Philadelphia, pp. 147–164 (1996). 14. Chow, G. M., and Gonsalves, K. E., (eds.), Nanotechnology: Molecularly Designed Materials, American Chemical Society Symposium Series 622, Washington, DC (1996). 15. Chow, G. M., and Kurihara, L. K., Chemical Synthesis and Processing of Nanostructured Powders and Films, (Carl. C. Koch, ed.), Nanostructured Materials: Processing, Properties and Potential Applications, Noyes Publications, pp. 3–50 (2002). 16. See, for example, papers in J. Mater. Chem., Chem. Comm., J. Am. Chem. Soc., Adv. Mater., Chem. Mater., Biomaterials, and Langmuir, etc. 17. Nielsen, A. E., Kinetics of Precipitation, Pergamon Press, London, NY (1964). 18. Walton, A. G., The Formation and Properties of Precipitates, Robert Krieger Publishing Company, Huntington, NY (reprint edition) (1979). 19. Lagally, M. G., An Atomic-Level View of Kinetic and Thermodynamic Influences in The Growth of Thin Films: A Review, Jpn. J. Appl. Phys., 32:1493–1501 (1993). 20. Turnbull, D., The Kinetics of Precipitation of Barium Sulfate From Aqueous Solutions, Acta Metall., 1:684–691 (1953). 21. LaMer, V. K., and Dinegar, R. H., Theory, Production, and Mechanism of Formation of Monodispersed Hydrosols, J. Am. Chem. Soc., 72:4847–4854 (1950). 22. For example, see Shanefield, D. J., Organic Additives and Ceramic Processing: with Applications in Powder Metallurgy, Ink, and Paint, Kluwer Academic Publishers, Boston, Dordrecht, London (1996). 23. Derjaguin, B. V., and Landau, J. D., Acta Physicochim., URSS 14:633 (1941); Verwey, E. J. W., and Overbeek, J. T. G., Theory of The Stability of Lyophobic Colloids, Elsevier, Amsterdam (1948). 24. Schmid, G., (ed.), Nanoparticles: from Theory to Application, Wiley-VCH, Weinheim (2004). 25. Masala, O., and Seshadri, R., Synthesis Routes for Large Volumes of Nanoparticles, Annu. Rev. Mater. Res., 34:41–81 (2004). 26. Trindade, T., O’Brien, P., and Pickett, N. L., Nanocrystalline Semiconductors: Synthesis, Properties, and Perspectives, Chem. Mater., 13:3843–3858 (2001). 27. Nersisyan, H. H., Lee, J. H., Son, H. T., Won, C. W., and Maeng, D. Y., A New and Effective Chemical Reduction Method for Preparation of Nanosized Silver Powder and Colloid Dispersion, Mater. Res. Bull., 38:949–956 (2003). 28. Caswell, K. K., Bender, C. M., and Murphy, C. J., Seedless, Surfactantless Wet Chemical Synthesis of Silver Nanowires, Nano Lett., 3:667–669 (2003). 29. Evanoff, D. D., and Chumanov, G., Size-controlled Synthesis of Nanoparticles. 1. “Silver-only” aqueous Suspensions via Hydrogen Reduction, J. Phys. Chem. B, 108:13948–13956 (2004).

1: Chemical Synthesis, Shi et al.

37

30. Yu, D. B., and Yam, V. W. W., Controlled Synthesis of Monodisperse Silver Nanocubes in Water, J. Am. Chem. Soc., 126:13200–13201 (2004). 31. Fievet, F., Lagier, J. P., and Figlarz, M., Preparing Monodisperse Metal Powders in Micrometer and Submicrometer Size by The Polyol Process, Mater. Res. Bull., 14:29–34 (1989). 32. Kurihara, L. K., Chow, G. M., and Schoen, P. E., Nanocrystalline Metallic Powders and Films Produced by The Polyol Method, Nanostruct. Mater., 5:607–613 (1995). 33. Chow, G. M., Kurihara, L. K., Kemner, K. M., Schoen, P. E., Elam, W. T., Ervin, A., Keller, S., Zhang, Y. D., Budnick, J., and Ambrose, T., Structural, Morphological, and Magnetic Study of Nanocrystalline Cobalt-Copper Powders Synthesized by The Polyol Process, J. Mater. Res., 10:1546–1554 (1995). 34. Chow, G. M., Schoen, P. E., and Kurihara, L. K., Nanostructured Metallic Powders and Films Via an Alcoholic Solvent Process, Navy Case No. 76,572, US Patent No. 5,759,230 (June 2, 1998). 35. Yin, H., and Chow, G. M., Anomalous Electroless Polyol Deposition of FeNi Powders and Films, J. Electrochem. Soc., 149:C68–C73 (2002). 36. Sun, Y. G., and Xia, Y. N., Large-scale Synthesis of Uniform Silver Nanowires through A Soft, Self-seeding, Polyol Process, Adv. Mater., 14:833–837 (2002). 37. Chakroune, N., Viau, G., Ricolleau, C., Fievet-Vincent, F., and Fievet, F., Cobalt-based Anisotropic Particles Prepared by the Polyol Process. J. Mater. Chem., 13:312–318 (2003). 38. Wiley, B., Herricks, T., Sun, Y. G., and Xia, Y. N., Polyol Synthesis of Silver Nanoparticles: Use of Chloride and Oxygen to Promote the Formation of Single-crystal, Truncated Cubes and Tetrahedrons, Nano Lett., 4:1733–1739 (2004). 39. Herricks, T., Chen, J. Y., and Xia, Y. N., Polyol Synthesis of Platinum Nanoparticles: Control of Morphology with Sodium Nitrate, Nano Lett., 4:2367–2371 (2004). 40. Chen, J. Y., Herricks, T., and Xia, Y. N., Polyol Synthesis of Platinum Nanostructures: Control of Morphology through the Manipulation of Reduction Kinetics, Angew. Chem. Int. Edit., 44:2589–2592 (2005). 41. Wiley, B., Sun, Y. G., Chen, J. Y., Cang, H., Li, Z. Y., Li, X. D., and Xia, Y. N., Shape-controlled Synthesis of Silver and Gold Nanostructures. MRS Bull., 30:356–361 (2005). 42. Garcia-Gutierrez, D. I., Gutierrez-Wing, C. E., Giovanetti, L., RamalloLopez, J. M., Requejo, F. G., and Jose-Yacaman, M., Temperature Effect on the Synthesis of Au—Pt Bimetallic Nanoparticles, J. Phys. Chem. B, 109:3813–3821 (2005). 43. Sun, Y. G., and Xia, Y. N., Shape-controlled Synthesis of Gold and Silver Nanoparticles, Science, 298:2176–2179 (2002). 44. Im, S. H., Lee, Y. T., Wiley, B., and Xia, Y. N., Large-scale Synthesis of Silver Nanocubes: The Role of HCl in Promoting Cube Perfection and Monodispersity, Angew. Chem. Int. Edit., 44:2154–2157 (2005). 45. Murray, C. B., Sun, S. H., Doyle, H., and Betley, T., Monodisperse 3d Transition-metal (Co, Ni, Fe) Nanoparticles and Their Assembly into Nanoparticle Superlattices, MRS Bull., 26:985–991 (2001).

38

Processing

46. Liu, C., Wu, X. W., Klemmer, T., Shukla, N., Yang, X. M., Weller, D., Roy, A. G., Tanase, M., and Laughlin, D., Polyol Process Synthesis of Monodispersed FePt Nanoparticles, J. Phys. Chem. B, 108:6121–6123 (2004). 47. Yin, H., and Chow, G. M., Electroless Polyol Deposition of FeNi-based Powders and Films, J. Mater. Res., 18:180–187 (2003). 48. Suslick, K. S., Sonochemistry, Science, 247:1439–1445 (1990). 49. Riesz, P., and Kondo, T., Free-Radical Formation Induced by Ultrasound and Its Biological Implications, Free Radical Bio. Med., 13:247–270 (1992). 50. Caruso, R. A., Ashokkumar, M., and Grieser, F., Sonochemical Formation of Gold Sols, Langmuir, 18:7831–7836 (2002). 51. Fujimoto, T., Terauchi, S., Umehara, H., Kojima, I., and Henderson, W., Sonochemical Preparation of Single-dispersion Metal Nanoparticles from Metal Salts, Chem. Mater., 13:1057–1060 (2001). 52. Sun, S. H., Murray, C. B., Weller, D., Folks, L., and Moser, A., Monodisperse FePt nanoparticles and Ferromagnetic FePt Nanocrystal Superlattices, Science, 287:1989–1992 (2000). 53. Howard, L. E. M., Nguyen, H. L., Giblin, S. R., Tanner, B. K., Terry, I., Hughes, A. K., and Evans, J. S. O., A Synthetic Route to Size-controlled fcc and fct FePt Nanoparticles, J. Am. Chem. Soc., 127:10140–10141 (2005). 54. Shevchenko, E. V., Talapin, D. V., Rogach, A. L., Kornowski, A., Haase, M., and Weller, H., Colloidal Synthesis and Self-assembly of CoPt3 Nanocrystals, J. Am. Chem. Soc., 124:11480–11485 (2002). 55. Gallagher, P. K., Chemical Synthesis, in: Engineered Materials Handbook, Vol. 4, Ceramics and Glasses, ASM International, USA, pp. 52–64 (1991). 56. Shoup, R. D., Sol-Gel Processes, in: Engineered Materials Handbook, Vol. 4, Ceramics and Glasses, ASM International, USA, pp. 445–452 (1991). 57. Pierre, A. C., Sol-Gel Processing of Ceramic Powders, Am. Ceram. Soc. Bull., 70:1281–1288 (1991). 58. Brinker, C. J., and Sherrer, G. W., Sol-Gel Science, Academic Press, NY (1990). 59. Feldmann, C., and Metzmacher, C., Polyol Mediated Synthesis of Nanoscale MS Particles (M = Zn, Cd, Hg), J. Mater. Chem., 11:2603–2606 (2001). 60. Feldmann, C., and Jungk, H. O., Polyol-mediated Preparation of Nanoscale Oxide Particles, Angew. Chem. Int. Edit., 40:359–362 (2001). 61. Feldmann, C., Polyol-mediated Synthesis of Nanoscale Functional Materials, Adv. Funct. Mater., 13:101–107 (2003). 62. Caruntu, D., Remond, Y., Chou, N. H., Jun, M. J., Caruntu, G., He, J. B., Goloverda, G., O’Connor, C., and Kolesnichenko, V., Reactivity of 3d Transition Metal Cations in Diethylene Glycol Solutions. Synthesis of Transition Metal Ferrites with the Structure of Discrete Nanoparticles complexed with Long-chain Carboxylate Anions, Inorg. Chem., 41:6137–6146 (2002). 63. Caruntu, D., Caruntu, G., Chen, Y., O’Connor, C. J., Goloverda, G., and Kolesnichenko, V. L., Synthesis of Variable-sized Nanocrystals of Fe3O4 with High Surface Reactivity, Chem. Mater., 16:5527–5534 (2004).

1: Chemical Synthesis, Shi et al.

39

64. Li, Z., Sun, Q., and Gao, M. Y., Preparation of Water-soluble Magnetite Nanocrystals from Hydrated Ferric Salts in 2-pyrrolidone: Mechanism Leading to Fe3O4, Angew. Chem. Int. Edit., 44:123–126 (2005). 65. Murray, C. B., Norris, D. J., and Bawendi, M. G., Synthesis and Characterization of Nearly Monodisperse Cde (e = S, Se, Te) Semiconductor Nanocrystallites, J. Am. Chem. Soc., 115:8706–8715 (1993). 66. Green, M., and O’Brien, P., The Synthesis of III-V Semiconductor Nanoparticles Using Indium and Gallium Diorganophosphides as Single-molecular Precursors, J. Mater. Chem., 14:629–636 (2004). 67. Peng, Z. A., and Peng, X. G., Formation of High-quality CdTe, CdSe, and CdS Nanocrystals Using CdO as Precursor, J. Am. Chem. Soc., 123:183–184 (2001). 68. Zhong, X. H., Feng, Y. Y., Knoll, W., and Han, M. Y., Alloyed ZnxCd1−xS Nanocrystals with Highly Narrow Luminescence Spectral Width, J. Am. Chem. Soc., 125:13559–13563 (2003). 69. Bailey, R. E., and Nie, S. M., Alloyed Semiconductor Quantum Dots: Tuning the Optical Properties without Changing the Particle Size, J. Am. Chem. Soc., 125:7100–7106 (2003). 70. Lin, S. L., Pradhan, N., Wang, Y. J., and Peng, X. G., High Quality ZnSe and ZnS Nanocrystals Formed by Activating Zinc Carboxylate Precursors, Nano Lett., 4:2261–2264 (2004). 71. Nair, P. S., Fritz, K. P., and Scholes, G. D., A Multiple Injection Method for Exerting Kinetic Control in the Synthesis of CdSe Nanorods, Chem. Commun., 2084–2085 (2004). 72. Qu, L. H., and Peng, X. G., Control of Photoluminescence Properties of CdSe Nanocrystals in Growth, J. Am. Chem. Soc., 124:2049–2055 (2002). 73. Qu, L. H., Peng, Z. A., and Peng, X. G., Alternative Routes Toward High Quality CdSe Nanocrystals, Nano Lett., 1:333–337 (2001). 74. Peng, X. G., Mechanisms for the Shape-control and Shape-evolution of Colloidal Semiconductor Nanocrystals, Adv. Mater., 15:459–463 (2003). 75. Peng, Z. A., and Peng, X. G., Nearly Monodisperse and Shape-controlled CdSe Nanocrystals via Alternative Routes: Nucleation and Growth, J. Am. Chem. Soc., 124:3343–3353 (2002). 76. Cao, Y. C., and Wang, J. H., One-pot Synthesis of High-quality Zinc-blende CdS Nanocrystals, J. Am. Chem. Soc., 126:14336–14337 (2004). 77. Jana, N. R., Chen, Y. F., and Peng, X. G., Size- and Shape-controlled Magnetic (Cr, Mn, Fe, Co, Ni) Oxide Nanocrystals via A Simple and General Approach, Chem. Mater., 16:3931–3935 (2004). 78. Yu, W. W., Falkner, J. C., Yavuz, C. T., and Colvin, V. L., Synthesis of Monodisperse Iron Oxide Nanocrystals by Thermal Decomposition of Iron Carboxylate Salts, Chem. Commun., 2306–2307 (2004). 79. Park, J., An, K. J., Hwang, Y. S., Park, J. G., Noh, H. J., Kim, J. Y., Park, J. H., Hwang, N. M., and Hyeon, T., Ultra-large-scale Syntheses of Monodisperse Nanocrystals, Nat. Mater., 3:891–895 (2004). 80. Park, J., Lee, E., Hwang, N. M., Kang, M. S., Kim, S. C., Hwang, Y., Park, J. G., Noh, H. J., Kini, J. Y., Park, J. H., and Hyeon, T., One-nanometerscale Size-controlled Synthesis of Monodisperse Nagnetic Iron Oxide Nanoparticles, Angew. Chem. Int. Edit., 44:2872–2877 (2005).

40

Processing

81. Zhang, H., Wang, L. P., Xiong, H. M., Hu, L. H., Yang, B., and Li, W., Hydrothermal Synthesis for High-quality CdTe Nanocrystals, Adv. Mater., 15:1712–1715 (2003). 82. Wang, X., Zhuang, J., Peng, Q., and Li, Y. D., A General Strategy for Nanocrystal Synthesis, Nature, 437:121–124 (2005). 83. Ebelmen, J., J. Ann. Chim. Phys. Series 3, 57, and 311 (1846). 84. Graham, T., On the Properties of Silicic Acid and Other Analogous Colloidal Substances, J. Chem. Soc. London, 17:318–327 (1864). 85. Kistler, S. S., Coherent Expanded Aerogels and Jellies, Nature, 127:741(1931). 86. Roy, R., and Osborn, E. F., The System Al2O3—SiO2—H2O, Am. Mineral., 39:853–885 (1954). 87. DeVries, R. C., Roy, R., and Osborn, E. F., Phase Equilibria in The System CaO—TiO2—SiO2., J. Am. Ceram. Soc., 38:158–171 (1955). 88. Roy, R., Aids in Hydrothermal Experimentation: II, Methods of Making Mixtures for Both “Dry” and “Wet” Phase Equilibrium Studies, J. Am. Ceram. Soc., 39:145–146 (1956). 89. Roy, R., Gel Route to Homogeneous Glass Preparation, J. Am. Ceram. Soc., 52:344 (1969). 90. Sugimoto, T., Zhou, X. P., and Muramatsu, A., Synthesis of Uniform Anatase TiO2 Nanoparticles by Gel-sol Method 3. Formation Process and Size Control, J. Colloid Interf. Sci., 259:43–52 (2003). 91. Sugimoto, T., Zhou, X. P., and Muramatsu, A., Synthesis of Uniform Anatase TiO2 Nanoparticles by Gel-sol method 4. Shape Control, J. Colloid Interf. Sci., 259:53–61 (2003). 92. Barnard, A. S., and Curtiss, L. A., Prediction of TiO2 Nanoparticle Phase and Shape Transitions Controlled by Surface Chemistry, Nano Lett., 5:1261–1266 (2005). 93. Kurihara, L. K., Chow, G. M., Choi, L. S., and Schoen, P. E., Chemical Synthesis and Processing of Nanostructured Aluminum Nitride, (T. P. Battle and H. Henein, eds.), Processing and Handling of Powders and Dusts, The Minerals, Metals, and Materials Society, pp. 3–12, Warrendale, PA (1997). 94. Kurihara, L. K., Chow, G. M., and Schoen, P. E., Nanostructured Ceramic Nitride Powders and A Method of Making The Same, US Navy Case No. 77,219, US Patent No. 5,876,682 (March 2, 1999). 95. Schubert, U., Schwertfeger, F., and Gorsmann, C., Sol-gel Materials with Controlled Nanoheterogeneity, in: Nanotechnology: Molecularly Designed Materials, (G. M. Chow, and K. E. Gonsalves, eds.), Ch. 25, pp. 366–381, American Chemical Society Symposium Series 622, Washington, DC (1996). 96. Armelao, L., Barreca, D., Bottaro, G., Mattei, G., Sada, C., and Tondello, E., Copper-Silica Nanocomposites Tailored by the Sol-Gel Route, Chem. Mater., 17:1450–1456 (2005). 97. Wen, J. Y., and Wilkes, G. L., Organic/inorganic Hybrid Network Materials by the Sol-gel Approach, Chem. Mater., 8:1667–1681 (1996). 98. Loy, D. A., and Shea, K. J., Bridged Polysilsesquioxanes-Highly Porous Hybrid Organic-Inorganic Materials, Chem. Rev., 95:1431–1442 (1995). 99. Wight, A. P., and Davis, M. E., Design and Preparation of Organicinorganic Hybrid Catalysts, Chem. Rev., 102:3589–3613 (2002).

1: Chemical Synthesis, Shi et al.

41

100. Vioux, A., Le Bideau, J., Mutin, P. H., and Leclercq, D., Hybrid Organicinorganic Materials Based on Organophosphorus Derivatives, Top Curr. Chem., 232:145–174 (2004). 101. Shea, K. J., and Loy, D. A., Bridged Polysilsesquioxanes. Molecular-engineered Hybrid Organic-inorganic Materials, Chem. Mater., 13:3306–3319 (2001). 102. Cerveau, G., Corriu, R. J. P., and Framery, E., Nanostructured Organic-inorganic Hybrid Materials: Kinetic Control of the Texture, Chem. Mater., 13:3373–3388 (2001). 103. Mackenzie, J. D., Hybrid Organic-Inorganic Materials, in: Hybrid OrganicInorganic Composites, (J. E. Mark, C. Y. C. Lee, and P. A. Bianconi, eds.), Ch. 17, pp. 226–236, American Chemical Society Symposium Series 585, Washington, DC (1995). 104. Dave, B. C., Dunn, B., Valentine, J. S., and Zink, J. I., Nanoconfned Proteins and Enzymes: Sol-Gel Based Biomolecular Materials, in: Nanotechnology: Molecularly Designed Materials (G. M. Chow, and K. E. Gonsalves, eds.), Ch. 24, pp. 351–365, American Chemical Society Symposium Series 622, Washington, DC (1996). 105. Gill, I., Bio-doped Nanocomposite Polymers: Sol-gel Bioencapsulates, Chem. Mater., 13:3404–3421 (2001). 106. Gonsalves, K. E., Chow, G. M., Zhang, Y., Budnick, J. I., and Xiao, T. D., Iron Nitride/Boron Nitride Magnetic Nanocomposite Powders, Adv. Mater., 6:291–292 (1994). 107. Xiao, T. D., Gonsalves, K. E., and Strutt, P. R., Synthesis of Aluminum Nitride/Boron Nitride Materials, J. Am. Ceram. Soc., 76:987–992 (1993). 108. Chow, G. M., Xiao, T. D., Chen, X., and Gonsalves, K. E., Compositional and Thermal Effects on Chemically Processed AlN-BN Nanocomposite Powders, J. Mater. Res., 9:168–175 (1994). 109. Vioux, A., Nonhydrolytic Sol-gel Routes to Oxides, Chem. Mater., 9:2292–2299 (1997). 110. Joo, J., Yu, T., Kim, Y. W., Park, H. M., Wu, F. X., Zhang, J. Z., and Hyeon, T., Multigrarn Scale Synthesis and Characterization of Monodisperse Tetragonal Zirconia Nanocrystals. J. Am. Ceram. Soc., 125:6553–6557 (2003). 111. Joo, J., Kwon, S. G., Yu, J. H., and Hyeon, T., Synthesis of ZnO Nanocrystals with Cone, Hexagonal Cone, and Rod Shapes via Non-hydrolytic Ester Elimination Sol-gel Reactions, Adv. Mater., 17:1873–1877 (2005). 112. Pinna, N., Neri, G., Antonietti, M., and Niederberger, M., Nonaqueous Synthesis of Nanocrystalline Semiconducting Metal Oxides for Gas Sensing, Angew. Chem. Int. Edit., 43:4345–4349 (2004). 113. Hay, J. N., and Raval, H. M., Synthesis of Organic-inorganic Hybrids via the Non-hydrolytic Sol-gel Process, Chem. Mater., 13:3396–3403 (2001). 114. Bein, T., Zeolitic Host-Guest Interactions and Building Blocks for the SelfAssembly of Complex Materials, MRS Bull., 30:713–720 (2005). 115. Kresge, C. T., Leonowicz, M. E., Roth, W. J., Vartuli, J. C., and Becks, J. S., Ordered Mesoporpous Molecular-sieve Synthesized by a Liquidcrystal Template Mechanism, Nature, 359:710–712 (1992). 116. Shi, J. L., Hua, Z. L., and Zhang, L. X., Nanocomposites from Ordered Mesoporous Materials, J. Mater. Chem., 14:795–806 (2004).

42

Processing

117. Bronstein, L. M., Nanoparticles Made in Mesoporous Solids, Top. Curr. Chem., 226:55–89 (2003). 118. Stein, A., Melde, B. J., and Schroden, R. C., Hybrid Inorganic-organic Mesoporous Silicates-nanoscopic Reactors Coming of Age, Adv. Mater., 12:1403–1419 (2000). 119. Xiao, F. S., Ordered Mesoporous Materials with Improved Stabilityand Catalytic Activity, Top. Catal., 35:9–24 (2005). 120. Scott, B. J., Wirnsberger, G., and Stucky, G. D., Mesoporous and Mesostructured Materials for Optical Applications, Chem. Mater., 13:3140–3150 (2001). 121. Wirnsberger, G., Yang, P., Scott, B. J., Chmelka, B. F., and Stucky, G. D., Mesostructured Materials for Optical Applications: from Low-k Dielectrics to Sensors and Lasers, Spectrochim. Acta A, 57:2049–2060 (2001). 122. Brieler, F. J., Grundmann, P., Fröba, M., Chen, L., Klar, P. J., Heimbrodt, W., Krug von Nidda, H-A., Kurz, T., and Loidl, A., Comparison of the Magnetic and Optical Properties of Wide-Gap (II,Mn)VI Nanostructures Confined in Mesoporous Silica, Eur. J. Inorg. Chem., 2005:3597–3611 (2005). 123. El Haskouri, J., Cabrera, S., Gomez-Garcia, C. J., Guillem, C., Latorre, J., Beltran, A., Beltran, D., Marcos, M. D., and Amoros, P., High Cobalt Content Mesoporous Silicas, Chem. Mater., 16:2805–2813 (2004). 124. Zhao, W., Gu, J., Zhang, L., Chen, H., and Shi, J., Fabrication of Uniform Magnetic Nanocomposite Spheres with a Magnetic Core/Mesoporous Silica Shell Structure, J. Am. Ceram. Soc., 127:8916–8917 (2005). 125. Yoon, J., Lee, W., and Thomas, E. L., Self-Assembly of Block Copolymers for Photonic-Bandgap Materials, MRS Bull., 30:721–726 (2005). 126. Colfen, H., and Yu, S. H., Biomimetic Mineralization/Synthesis of Mesoscale Order in Hybrid Inorganic-Organic Materials via Nanoparticle Self-Assembly, MRS Bull., 30:727–735 (2005). 127. Reiss, B. D., Mao, M., Solis, D. J., Ryan, K. S., Thomson, T., and Belcher, A. M., Biological Routes to Metal Alloy Ferromagnetic Nanostructures, Nano Lett., 4:1127–1132 (2004). 128. Fendler, J. H., Atomic and Molecular Clusters in Membrane Mimetic Chemistry, Chem. Rev., 87:877–899 (1987). 129. Shchukin, D. G., and Sukhorukov, G. B., Nanoparticle Synthesis in Engineered Organic Nanoscale Reactors, Adv. Mater., 16:671–682 (2004). 130. Puvvada, S., Baral, S., Chow, G. M., Qadri, S. B., and Ratna, B. R., Synthesis of Palladium Metal Nanoparticles in the Bicontinuous Cubic Phase of Glycerol Monooleate, J. Am. Chem. Soc., 116:2135–2136 (1994). 131. Schwuger, M. J., Stickdorn, K., and Schomacker, R., Microemulsions in Technical Processes, Chem. Rev., 95:849–864 (1995). 132. Lopez-Quintela, M. A., Synthesis of Nanomaterials in Microemulsions: Formation Mechanisms and Growth Control, Curr. Opin. Colloid In., 8:137–144 (2003). 133. Maillard, M., Giorgio, S., and Pileni, M. P., Silver Nanodisks, Adv. Mater., 14:1084–1086 (2002). 134. Petit, C., Taleb, A., and Pileni, M. P., Self-organization of Magnetic Nanosized Cobalt Particles, Adv. Mater., 10:259–261 (1998).

1: Chemical Synthesis, Shi et al.

43

135. Chen, D. H., and Chen, C. J., Formation and Characterization of Au—Ag Bimetallic Nanoparticles in Water-in-oil Microemulsions, J. Mater. Chem., 12:1557–1562 (2002). 136. Simmons, B. A., Li, S. C., John, V. T., McPherson, G. L., Bose, A., Zhou, W. L., and He, J. B., Morphology of US Nanocrystals Synthesized in a Mixed Surfactant System, Nano Lett., 2:263–268 (2002). 137. Koran, A. R., Hull, R., Opila, R. L., Bawendi, M. G., Steigerwald, M. L., Carroll, P. J., and Brus, L. E., Nucleation and Growth of CdSe on ZnS Quantum Crystallite Seeds, and Vice Versa, In Inverse Micelle Media, J. Am. Chem. Soc., 112:1327–1332 (1990). 138. Zarur, A. J., and Ying, J. Y., Reverse Microemulsion Synthesis of Nanostructured Complex Oxides for Catalytic Combustion, Nature, 403:65–67 (2000). 139. Jang, J. K., and Yoon, H., Novel Fabrication of Size-tunable Silica Nanotubes Using a Reverse-microemulsion-mediated Sol-gel Method, Adv. Mater., 16:799–802 (2004). 140. Wang, W., and Asher, S. A., Photochemical Incorporation of Silver Quantum Dots in Monodisperse Silica Colloids for Photonic Crystal Applications, J. Am. Chem. Soc., 123:12528–12535 (2001). 141. Herrig, H., and Hempelmann, R., Microemulsion Mediated Synthesis of Ternary and Quaternary Nanoscale Mixed Oxide Ceramic Powders, Nanostruct. Mater., 9:241–244 (1997). 142. Mann, S., and Williams, R. J. P., Precipitation Within Unilamellar Vesicles, Part 1, Studies of Silver Oxide Formation, Dalton T., 311–316 (1983). 143. Mann, S., and Hannington, J. P., Formation of Iron Oxides in Unilamellar Vesicles, J. Colloid Interf. Sci., 122:326–335 (1988). 144. Bhandarkar, S., and Bose, A., Synthesis of Submicrometer Crystals of Aluminum Oxide by Aqueous Intravesicular Precipitation, J. Colloid Interf. Sci., 135:531–538 (1990). 145. Chow, G. M., Markowitz, M. A., and Singh, A., Synthesizing Submicrometer and Nanoscale Particles via Self-assembled Molecular Membranes, J. Miner., Met., and Mater. Soc., 45:62–65 (1993). 146. Markowitz, M. A., Chow, G. M., and Singh, A., Polymerized Phospholipid Membrane Mediated Synthesis of Metal Nanoparticles, Langmuir, 10:4905–4102 (1994). 147. Chow, G. M., Markowitz, M. A., Rayne, R., Dunn, D. N., and Singh, A., Phospholipid Mediated Synthesis and Characterization of Gold Nanoparticles, J. Colloid Interf. Sci., 183:135–142 (1996). 148. Froes, F. H., Suryanarayana, C., Russell, K. C., and Ward-Close, C. M., Far From Equilibrium Processing of Light Metals, in: Novel Techniques in Synthesis and Processing of Advanced Materials, (J. Singh, and S. M. Copley, eds.), The Minerals, Metals, and Materials Society, pp. 1–21 (1995). 149. Qi, L. M., Colfen, H., and Antonietti, M., Synthesis and Characterization of CdS Nanoparticles Stabilized by Double-hydrophilic Block Copolymers, Nano Lett., 1:61–65 (2001). 150. Zhang, M. F., Drechsler, M., and Muller, A. H. E., Template-Controlled Synthesis of Wire-like Cadmium Sulfide Nanoparticle Assemblies Within Core-shell Cylindrical Polymer Brushes, Chem. Mater., 16:537–543 (2004).

44

Processing

151. Ichinose, I., Kimizaka, N., and Kunitake, T., Formation of a Novel CdS Cluster in An Organic Multilayer Template: A Case of An Organic/Inorganic Superlattice, J. Phys. Chem., 99:3736–3742 (1995). 152. Zelenski, C. M., Hornyak, G. L., and Dorhout, P. K., Synthesis and Characterization of CdS Particles Within a Nanoporous Aluminum Oxide Template, Nanostruct. Mater., 9:173–176 (1997). 153. Korgel, B. A., and Monbouquette, H. G., Controlled Synthesis of Mixed Core and Layered (Zn,Cd)S and (Hg,Cd)S Nanocrystals Within Phosphatidylcholine Vesicles, Langmuir, 16:3588–3594 (2000). 154. Korgel, B. A., and Monboriquette, H. G., Synthesis of Size-Monodisperse CdS Nanocrystals Using Phoshatidylcholine Vesicles as True Reaction Compartments, J. Phys. Chem., 100:346–351 (1996). 155. Colvin, V. L., The Potential Environmental Impact of Engineered Nanomaterials, Nat. Biotechnol., 21:1166–1170 (2003). 156. Hoet, P. H. M., Nemmar, A., and Nemery, B., Health Impact of Nanomaterials? Nat. Biotechnol., 22:19 (2004). 157. Giles, J., Size Matters When It Comes to Safety, Report Warns. Nature, 430:599 (2004). 158. Oberdorster, G., Oberdorster, E., and Oberdorster, J., Nanotoxicology: An emerging discipline evolving from studies of ultrafine particles, Environ. Health Persp., 113:823–839 (2005). 159. Sayes, C. M., Fortner, J. D., Guo, W., Lyon, D., Boyd, A. M., Ausman, K. D., Tao, Y. J., Sitharaman, B., Wilson, L. J., Hughes, J. B., West, J. L., and Colvin, V. L., The Differential Cytotoxicity of Water-soluble Fullerenes, Nano Lett., 4:1881–1887 (2004). 160. Derfus, A. M., Chan, W. C. W., and Bhatia, S. N., Probing the Cytotoxicity of Semiconductor Quantum Dots, Nano Lett., 4:11–18 (2004). 161. Hilger, I., Fruhauf, S., Linss, W., Hiergeist, R., Andra, W., Hergt, R., and Kaiser, W. A., Cytotoxicity of Selected Magnetic Fluids on Human Adenocarcinoma Cells, J. Magn. Magn. Mater., 261:7–12 (2003). 162. Gupta, A. K., and Gupta, M., Cytotoxicity Suppression and Cellular Uptake Enhancement of Surface Modified Magnetic Nanoparticles, Biomaterials, 26:1565–1573 (2005). 163. Kirchner, C., Liedl, T., Kudera, S., Pellegrino, T., Javier, A. M., Gaub, H. E., Stolzle, S., Fertig, N., and Parak, W. J., Cytotoxicity of Colloidal CdSe and CdSe/ZnS Nanoparticles, Nano Lett., 5:331–338 (2005). 164. Connor, E. E., Mwamuka, J., Gole, A., Murphy, C. J., and Wyatt, M. D., Gold Nanoparticles Are Taken up by Human Cells But Do Not Cause Acute Cytotoxicity, Small, 1:325–327 (2005). 165. Yin, H., Too, H. P., and Chow, G. M., The Effects of Particle Size and Surface Coating on The Cytotoxicity of Nickel Ferrite, Biomaterials, 26:5818–5826 (2005). 166. Huang, J. F., and Sun, I. W., Fabrication and Surface Functionalization of Nanoporous Gold by Electrochemical Alloying/dealloying of Au—Zn in an Ionic Liquid, and the Self-assembly of L-cysteine Monolayers, Adv. Funct. Mater., 15:989–994 (2005). 167. Wang, H. C., Sun, S. G., Yan, J. W., Yang, H. Z., and Zhou, Z. Y., In situ STM Studies of Electrochemical Growth of Nanostructured Ni films

1: Chemical Synthesis, Shi et al.

168. 169. 170. 171.

172.

173.

174.

175.

176.

177. 178. 179. 180.

45

and Their Anomalous IR Properties, J. Phys. Chem. B, 109:4309–4316 (2005). Wang, F., Hosoiri, K., Doi, S., Okamoto, N., Kuzushima, T., Totsuka, T., and Watanabem, F., Nanostructured L10 Co—Pt Thin Films by an Electrodeposition Process, Electrochem. Commun., 6:1149–1152 (2004). Murthy, S. K., Vemagiri, J. K., Gunasekaran, R. A., Coane, P., and Varahramyan, K., Electroless Deposition of Soft Magnetic CoNiP Thin Films, J. Electrochem. Soc., 151:C1–C7 (2004). Chow, G. M., Pazirandeh, M., Baral, S., and Campbell, J. R., TEM & HRTEM Characterization of Metallized Nanotubules Derived From Bacteria, Nanostruct. Mater., 2:495–503 (1993). Krebs, J. J., Rubenstein, M., Lubitz, P., Harford, M. Z., Baral, S., Shashidhar, S., Ho, Y. S., Chow, G. M., and Qadri, S., Magnetic Properties of Permalloy-Coated Organic Tubules, J. Appl. Phys., 70:6404–6406 (1991). Chow, G. M., Stockton, W. B., Price, R., Baral, S., Ting, A. C., Ratna, B. R., Schoen, P. E., Schnur, J. M., Bergeron, G. L., Czarnaski, M. A., Hickman, J. J., and Kirkpatrick, D. A., Fabrication of Biologically Based Microstructure Composites for Vacuum Field Emission, Mat. Sci. Eng. AStruct., 158:1–6 (1992). Kirkpatrick, D. A., Bergeron, G. L., Czarnaski, M. A., Hickman, J. J., Chow, G. M., Price, R., Ratna, B. R., Schoen, P. E., Stockton, W. B., Baral, S., Ting, A. C., and Schnur, J. M., Demonstration of Vacuum Field Emission from a Self-assembling Bimolecular Microstructure Composite, Appl. Phys. Lett., 60:1556–1558 (1992). Unuma, H., Kanehama, T., Yamamoto, K., Watanabe, K., Ogata, T., and Sugawara, M., Preparation of Thin Films of MnO2 and CeO2 by a Modified Chemical Bath (oxidative-soak-coating) Method, J. Mater. Sci., 38:255–259 (2003). Chow, G. M., Kurihara, L. K., Feng, C. R., Schoen, P. E., and MartinezMiranda, L. J., Alternative Approach to Electroless Cu Metallization of AlN By a Nonaqueous Polyol Process, Appl. Phys. Lett., 70:2315–2317 (1997). Chow, G. M., Ding, J., Zhang, J., Lee, K. Y., Surani, D., and Lawrence, S. H., Magnetic and Hardness Properties of Nanostructured Ni—Co Films Deposited by a Nonaqueous Electroless Method, Appl. Phys. Lett., 74:1889–1891 (1999). Chow, G. M., Ding, J., and Zhang, J., Enhanced Magnetization of Nanostructured Granular Ni/[Cu(II)-C-O] Films, Appl. Phys. Lett., 80:1028–1030 (2002). Klein, L. C., (ed.), Sol Gel Technology for Thin Films, Fibers, Preforms, Electronics, and Specialty Shapes, Noyes Publications, NJ (1988). Wang, C. H., Hu, J. F., Chow, G. M., Hsu, P. C., and Hwu, Y. K., Structural and Optical Properties of Sol-Gel-Derived Au/BaTiO3 Nanocomposite Thin Films, J. Am. Ceram. Soc., 88:758–767 (2005). McDonogh, C., Sheridan, F., Butler, T. M., and MacCraith, B. D., Characterization of Sol-Gel Derived Silica Films, J. Non-Cryst. Solids, 194:72–77 (1996).

46

Processing

181. Doherty, S., and Fitmaurice, D., Preparation and Characterization of Transparent Nanocrystalline TiO2 Films Possessing Well-Defined Morphologies, J. Phys. Chem., 100:10732–19738 (1996). 182. Shacham, R., Avnir, D., and Mandler, D., Electrodeposition of Dye-Doped Titania Thin Films, J. Sol-Gel Sci. Techn., 31:329–334 (2004). 183. Barrow, D. A., Petroff, T. E., and Sayer, M., Thick Ceramic Coatings Using a Sol-Gel Based Ceramic-Ceramic 0–3 Composite, Surf. Coat. Tech., 76–77:113–118 (1995). 184. Moriguchi, I., Maeda, H., Teraoke, Y., and Kagwa, S., Preparation of a TiO2 Nanoparticulate Film Using a Two Dimensional Sol-Gel Process, Chem. Mater., 9:1050–1057 (1997). 185. Chen, W., Zhang, J. Y., Fang, Q., Li, S., Wu, J. X., Li, F. Q., and Jiang, K., Sol-gel Preparation of Thick Titania Coatings Aided by Organic Binder Materials, Sensor. Actuat. B-Chem., 100:195–199 (2004).

2 Synthesis of Nanostructured Materials by Inert-Gas Condensation Methods C. Suryanarayana and Balaji Prabhu Department of Mechanical, Materials and Aerospace Engineering, University of Central Florida, Orlando, Florida, USA

2.1 Introduction Nanostructured materials are single-phase or multiphase polycrystalline solids with a typical average grain size of a few nanometers (1 nm = 10−9 m), typically less than 100 nm. Such materials exhibit properties that are substantially different from and often superior to those of conventional coarse-grained materials, due to their unique microstructure. Since the grain sizes are so small, a significant volume fraction of the atoms resides in grain boundaries. Thus, the material is characterized by a large number of interfaces in which the atomic arrangements are different from those of the crystal lattice.[1] At such small grain sizes, the surfaces start to dominate the bulk behavior of the material and, consequently, nanostructured materials exhibit special and completely new and unexpected properties, often not observed in coarse-grained materials. Accordingly, nanostructured materials have been shown to exhibit very high strength, increased diffusivity, reduced sintering temperatures, useful catalytic properties, and attractive physical properties. Due to their novel and improved properties and varied potential applications in different fields of technology, these materials are attracting increasing attention from researchers all over the world. Some excellent reviews are available in the literature detailing the synthesis, characterization, properties, and potential applications of nanostructured materials.[1–11]

2.2 Classification Nanostructured materials can be classified into different categories depending on the number of dimensions in which the material has Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 47–90 © 2007 William Andrew, Inc.

47

48

Processing

nanometer modulations. Thus, they can be classified into (a) layered or lamellar structures, (b) filamentary structures, and (c) equiaxed nanostructured materials. A layered or lamellar structure is a one-dimensional (1D) nanostructure in which the magnitudes of length and width are much greater than the thickness, which is only a few nanometers in size. One can also visualize a two-dimensional (2D) rod-shaped nanostructure that can be termed filamentary; in this the length is substantially larger than the width or diameter, which are of nanometer dimensions. The most common of the nanostructures, however, is basically equiaxed (all the three dimensions are of nanometer size) and these are termed nanostructured crystallites (three-dimensional [3D] nanostructures).[12] The nanostructured materials may contain crystalline, quasicrystalline, or amorphous phases and can be metals, ceramics, polymers, or composites. If the grains are made up of crystals, the material is called nanocrystalline. On the other hand, if they are made up of quasicrystalline or amorphous (glassy) phases, they are termed nanoquasicrystals and nanoglasses, respectively. Gleiter[5] has further classified the nanostructured materials according to the composition, morphology, and distribution of the nanocrystalline component. Among the above, most research work is conducted on the synthesis, consolidation, and characterization of the 3D-nanostructured crystallites followed by the 1D-layered nanostructures. While the former are expected to find applications based on their high strength, improved formability, and a good combination of soft magnetic properties, the latter are targeted for electronic applications. Relatively few investigations have been carried out on the 2D-filamentary nanostructures.

2.3 Synthesis of Nanostructured Materials There are two basic approaches to the synthesis of nanostructured materials. The first is the “top-down” approach, which involves breaking down the bulk material into particles with nanometer-sized grains. The other approach is the “bottom-up” approach, in which individual atoms or molecules are put together to form the required nanoparticles. In the “topdown” methods we have mechanical processes such as mechanical alloying,[13,14] high-energy ball milling,[13,14] equal channel angular pressing,[15,16] high pressure torsion,[16,17] and accumulative roll bonding.[18] In contrast, in the “bottom-up” approach we have methods of gas-phase synthesis that include inert-gas condensation,[19] physical/chemical vapor condensation, electrodeposition,[20] spray pyrolysis, high-temperature

2: Synthesis, Suryanarayana and Prabhu

49

evaporation, flame synthesis, and plasma synthesis techniques. We also have the wet chemical processes including sol-gel, chemical reduction, and biological methods. A recent handbook provides descriptions of the different methods available to synthesize nanostructured materials.[21] Table 2.1 summarizes the different synthesis techniques based on the starting phase and the nature of the end product obtained.[6] It may be mentioned in passing that the resultant nanostructured materials can have significantly different properties, depending on the chosen fabrication route. The most important characteristics of nanostructured materials are (i) the size (grain size in case of a single-phase material and phase size in the case of multiphase materials), (ii) the composition and distribution of phases in case of materials with two or more phases, and (iii) the nature of the interfaces. There has been a growing interest in the synthesis of nanoparticles of well-defined size, shape, and composition in an effort to exploit these materials for niche industrial applications. Since the properties of materials are determined by their grain size, it is of interest to control the grain size to as narrow a range as possible. This assumes even greater significance in the case of nanostructured materials. Among all the techniques listed in Table 2.1, the inert-gas condensation (IGC) technique has been the most popular since it can produce particles of well-defined grain size Table 2.1 Methods for Synthesizing Nanostructured Materials[6]

Starting phase Vapor

Liquid

Solid

Technique Inert-gas condensation Physical vapor deposition Evaporation and sputtering Plasma processing Chemical vapor condensation Chemical reactions Rapid solidification Electrodeposition Chemical reactions Mechanical alloying/milling Devitrification of amorphous phases Spark erosion Sliding wear

Dimensionality of Product 3D 1D 3D 3D, 2D 3D 3D 1D, 3D 3D 3D 3D 3D 3D

50

Processing

and with a narrow size distribution. With this motivation in mind, this chapter will review the process of IGC starting from its inception to the latest developments. The technique is also referred to as gas-phase condensation (GPC) in the literature. We will start with a brief introduction to the inception and early research efforts in using IGC and this will be followed by the principle behind IGC. We will then discuss the mechanism of the technique, taking each stage of the process into consideration. After this, the various vaporization techniques will be explained along with the effect of the different process parameters on the particle size achieved. The advantages and drawbacks of the IGC technique will then be briefly discussed. We conclude the chapter with a brief description of the latest developments in this field.

2.4 Early Studies on Inert-Gas Condensation It has been suggested that primordial materials with phases on the order of nanometers condensed from our solar nebula.[3,22,23] It was only during the twentieth century that the technological synthesis of ultrafine particles in the laboratory began. The first recorded experiment, which formed the basis for later developments of the IGC method, was carried out in 1930 by Pfund[24] and this was immediately followed by the work of Burger and Van Cittert[25] and then later by Harris et al.[26] In this work, nucleation and growth were found to occur in a noble-gas atmosphere and this technique was referred to as the inert-gas evaporation method. Some of the disadvantages of the earlier methods (applicable to only a few selected materials) were overcome in this method. For example, the method was able to produce particles with (a) very small sizes, typically <10 nm, (b) narrow, and accurately known, distribution of particle sizes, and (c) electrically insulated individual particles. But, since the metal vapor source was not maintained at constant temperature, the evaporation rate and consequently the average particle size varied during evaporation. Additionally, the amount of sample produced was very small since the resistive filament could not carry a large charge. These difficulties were overcome in a method modified and developed by Granqvist and Buhrman,[27] who carried out the most comprehensive work on producing ultrafine particles by the inert-gas evaporation method. They evaporated metals (Mg, Zn, Sn, Cr, Fe, Co, Ni, Cu, Ga) from a crucible into a low-pressure atmosphere of an inert gas to produce nanoparticles and defined the various parameters (type of gas, gas pressure, and evaporation rate) which influence the particle size formed via this method.

2: Synthesis, Suryanarayana and Prabhu

51

The inert-gas evaporation method was extensively used by the Japanese school, starting from the 1960s (see, for example, refs. 28–31) and an excellent and exhaustive summary of the Japanese literature was presented by Uyeda in 1991.[32] They showed that a wide variety of metal ultrafine particles (10–100 nm in diameter) could be synthesized in a low-pressure inert-gas atmosphere and that their sizes could be controlled by varying the gas pressure (in the range of 1–30 Torr). Subsequent work by Thölén[33] from Denmark extended these investigations to other metals and studied the “physics” of the process (nucleation and coalescence of the clusters). However, the real breakthrough in the use of this technique came about when Gleiter[34] suggested that these ultrafine particles could be consolidated in situ to produce bulk materials with a large fraction of the atoms in the grain boundaries. Such materials were referred to as “microcrystalline materials” or “interfacial materials.” It was only in 1984 that the name “nanocrystalline materials” was coined[35] to denote the characteristics of materials with nanometer grain sizes when the volume fraction of the interfaces becomes comparable to the volume fraction of the crystalline regions. Several modifications have been made to the original technique and this technique is currently used to synthesize a variety of nanostructured metals and ceramics. Although the early work was mostly concerned with the synthesis and characterization of nanoparticles of pure metals, some studies were also reported on the synthesis of compounds, alloys, and mixtures of particles by IGC.[36–39] In the early stages oxides were investigated by the simple method of introducing oxygen into the vapor phase.[40,41] Nitrides and carbides of the metals could also be synthesized by filling the chamber with nitrogen or ammonia gases or by maintaining a carbonaceous atmosphere. The types of nanostructured materials prepared by the IGC method include metals (e.g., Cu, Fe, Ni, Pd, and W), ionic compounds (e.g., FeF2, CaF2, Fe2O3, and TiO2), and also covalent substances (e.g., Si).[1,7,42]

2.5 The Principle of Inert-Gas Condensation Inert-gas condensation (IGC) is a bottom-up approach to synthesizing nanostructured materials, which involves two basic steps. The first step is the evaporation of the material and the second step involves a rapid controlled condensation to produce the required particle size. The IGC set-up designed and used by Granqvist and Buhrman is shown in Fig 2.1.[27] In this unit, the chamber is evacuated to a pressure of about 2 × 10−6 Torr by an oil diffusion pump. A crucible containing the metal to be evaporated

Processing

52 Cooling water tubes Pressure gauge

Optical pyrometer Gas flow impedance

Glass cylinder Cu cooling plate

High pressure regulators

L-shaped rod Small shutter

Gas bombs

Big lid Crucible with metal

Ar

O2

Pt-PtRh thermocouple Graphite heating element To diffusion pump

Figure 2.1 Cross-sectional sketch of the experimental set-up used by Granqvist and Buhrman[27] to synthesize ultrafine metal particles by inert-gas evaporation.

is slowly heated via radiation from a graphite heater element. The temperature was set at a predetermined value. After evacuation, an inert gas (He, Xe, or Ar) is leaked into the chamber at a low pressure, usually 0.5–4 Torr, and the crucible is heated rapidly (at constant temperature and inertgas pressure). The ultrafine metal particles that nucleate and grow in the gas phase are collected on a water-cooled surface. A carbon-coated electron-microscope grid, attached to the center of the water-cooled surface, allows the powder particles to be collected and directly observed in the transmission electron microscope. With the use of a shutter, it was possible to sample less than a monolayer of particles on the grid. Figure 2.2 shows a schematic of a typical apparatus used nowadays.[43] The basic components of the apparatus are similar to what were used by Granqvist and Buhrman[27] except that the new units (i) contain a scraper to collect the powder particles into a container and (ii) allow in-situ compaction of the powders into bulk. The main unit for producing bulk nanocrystalline materials consists of an ultrahigh-vacuum (UHV) system fitted with one or more evaporation sources, a cluster collection device to collect the deposited powder, a scraper to remove the collected powder particles, and a compaction device to consolidate the collected nanophase particles. The powder pellets are usually about 8–9 mm in diameter and about 0.1–0.5 mm in thickness. If one is only interested in nanostructured powders, then the compaction

2: Synthesis, Suryanarayana and Prabhu

53

Liquid nitrogen

Scraper

Main vacuum chamber

A Evaporation sources

B Gas inlet Funnel

Vacuum pumps Bellows Fixed piston Anvil Low pressure compaction unit

Slide Sleeve

Piston

Piston

High pressure compaction unit

Figure 2.2 Schematic of the inert-gas condensation chamber for the synthesis of nanostructured materials.[43]

device is not required. The cluster collection device is frequently cooled with liquid nitrogen (although not necessary) to improve the heat transfer efficiency and provide a large temperature gradient. The process of IGC consists of evaporating a metal (by resistive heating, radiofrequency heating, sputtering, electron beam heating, laser/plasma heating, or ion sputtering) inside a chamber that is evacuated to a very high vacuum by a turbo-molecular pump (to below 10−5 Pa) and then back-filling it with a low-pressure high-purity inert gas, at pressures of typically a few hundred pascals. The evaporated metal atoms collide with the inert-gas atoms inside the chamber, lose their kinetic energy, and condense in the form of small, discrete crystals of loose powder, close to the crucible. This process leads to a large degree of supersaturation, leading to nucleation[2] and results in

54

Processing

the formation of particles. Convection currents, generated due to the heating of the inert gas by the evaporation source and cooling by the liquid nitrogen-filled collection device (cold finger), carry the condensed fine powders from the crucible region to the collector device, where they are collected via thermophoresis (as a result of a temperature gradient within the flow, which induces the particles to travel in the direction of decreasing temperature). The powder particles collected on the surface of the cold finger are very loosely packed in the form of fractals.[28,44] Therefore, these can be easily stripped off into a compaction device by moving an annular Teflon ring down the length of the tube and consolidated at room temperature and a pressure of about 1–2 GPa. These steps are also carried out under UHV conditions, following the removal of the inert gas, mainly to avoid any entrapment of the inert gas into the powder during compaction. Additionally, conducting these operations under UHV conditions, without exposing the materials to the atmosphere, ensures that the cleanliness of the powder particle surfaces is maintained when the compact is taken out of the chamber. The density of the samples after consolidation is typically about 90–95% of the theoretical value in the case of nanostructured metals, but is only about 75–85% in the case of the hard nanostructured ceramic particles. Assuming that cluster coalescence is not an issue, there are three fundamental rates that represent the core of the process and essentially control the formation of atom clusters: (1) The rate of supply of atoms to the supersaturation region where condensation takes place (2) The rate of energy removal from the hot atoms through a condensing medium such as the inert gas (3) The rate of cluster/particle removal from the supersaturation region Many systems that can be atomized into the gaseous phase by a variety of high-energy methods such as laser ablation, electric arc discharge, ion sputtering, etc. can recombine with each other to grow into large pieces again. The rate of recombination can be regulated by careful control of the atmosphere in which the atoms or molecules find themselves. A high concentration of inert-gas atoms can control the collision rate and hence the coalescence rate. The particle growth can therefore be limited to produce structures in the nanoscale range under the right conditions.

2: Synthesis, Suryanarayana and Prabhu

55

Evaporation of metal/alloy/nonmetallic precursor results in clusters/particles of the same material, but in reconstituted form. However, if alloys or compounds such as ceramic oxides need to be prepared, variations such as introduction of reactive gases or gas mixtures into the chamber can be employed. For example, Ti of 99.7% purity was evaporated from a resistanceheated tungsten boat at temperatures between 1550 and 1650°C into a 0.3–0.7 kPa He atmosphere over a period of 15–30 minutes. The Ti metal clusters that condensed in the He atmosphere were deposited on the cold finger in the chamber and subsequently oxidized to TiO2 by the rapid introduction of about 2 kPa of oxygen.[36] Similarly, α-Al2O3 was produced when Al clusters of 18 nm average grain size were annealed in air at 1000°C.[45] On the other hand, if the vapor pressure of a compound is sufficiently large, as in the cases of MgO and ZnO, these materials could be directly sublimed from the vapor phase. For example, MgO particles of about 5 nm size were produced by directly heating MgO to 1600°C in a vacuum chamber to which a few hundred pascals of He was added. Since the MgO particles produced were oxygen-deficient, they were fully converted to stoichiometric MgO by subsequent exposure to oxygen introduced into the chamber.[45] Production of nanostructured powders, which are agglomerated but do not form hard agglomerates and which can be compacted in the apparatus itself without exposing them to air, constitutes the unique feature of IGC. The major technical difficulties to overcome in developing a successful bottom-up approach are (i) controlling the growth of the particles to retain the nanostructures and (ii) then stopping the newly formed particles from agglomerating. It is essential here to differentiate between nanoparticles and clusters. The term nanoparticle is generally used to indicate particles with diameters smaller than 100 nm,[46] while the term cluster refers to smaller nanoparticles with fewer than 104 molecules or atoms, corresponding to a diameter of only a few nanometers. A number of different designs are available in the literature for the nanocluster generation and collector systems. Leung et al.[47] described the design of such a system based on the earlier work of Andres et al.[48] In this design, the nanoclusters are produced by vaporizing the sample and expanding the vapor into a low-pressure environment. The resulting clusters formed during nucleation are collected using thermophoresis. There are three main parts in the gas condensation process–generation, supersonic expansion, and collection. The material of interest is first vaporized using ceramic resistance heating in a generator consisting of an alumina tube closed at one end except for a small hole (1 mm in

56

Processing

diameter) which serves as a sonic nozzle. The metal vapor is carried by the inert gas through the sonic nozzle, at which point conditions are maintained in such a way that the pressure in the expansion section is below the critical pressure. Upon exiting the sonic nozzle, the flow accelerates and the pressure and temperature rapidly decrease. The collection chamber consists of a small tube cooled to liquid-nitrogen temperature. The nanoclusters are collected via thermophoresis. After the run is completed, the collection tube is removed and the particles are removed with a Teflon-coated plunger and placed in a compactor. The nanoclusters can also be accelerated into a sampling chamber for analysis with a mass spectrometer. Since an inert ceramic tube is used for evaporating the material, it is possible to use various carrier gases such as oxygen, nitrogen, etc.

2.6 Evaporation Techniques A number of different techniques have been employed to evaporate metals in the IGC chamber. These have varied from the traditional thermal methods used in the early 1930s to sputtering and plasma methods used currently.

2.6.1 Thermal Evaporation Thermal evaporation was the first technique used to evaporate the metals in the IGC process during the early 1930s by Pfund[24] and Burger and van Cittert.[25] Bulk pure metals can be resistively heated in ceramic crucibles at high heating rates to transform the metal from the solid to the vapor state. Granqvist and Buhrman[27] were able to produce a host of metal nanoparticles by thermal evaporation from a crucible, which was heated indirectly by a graphite heating element. The metal smoke consisted of only one zone as compared to three in the case of resistive heating where a boat was used instead of the crucible. In addition to Joule-heated evaporation sources, other vaporization techniques such as laser ablation, sputtering, electron beam heating, plasma methods, electric arc discharge, and radiofrequency heating are also used. These allow the use of refractory or reactive precursors and play a major role in synthesis of complex nanomaterials. However, in some of the vaporization methods such as electron beam heating and other plasma methods, the low-pressure limit is a disadvantage.

2: Synthesis, Suryanarayana and Prabhu

57

Other methods for producing metal nuclei in the gas phase include inductive heating, chemical vapor deposition (CVD) and chemical vapor condensation (CVC). The inductive heating process is used when temperatures above 2700 K are needed to evaporate the metal precursor.[49] It is essentially a variation of the Joule heating process in which an inductive heater is used to heat the precursor. Refractory metals[50] and other materials such as Pd[51] have been evaporated successfully. A disadvantage, however, is that undesirable heating of other parts in the reactor leads to outgassing. In the CVD and CVC processes, chemical precursors are converted to their gaseous phase and subsequently dissociated by thermal energy or plasma to generate nuclei.[52] In the CVD process, a chemical reaction between the substrate surface and the gaseous precursor is activated. Activation can be achieved either by means of temperature (thermal CVD) or with plasma (PECVD: plasma enhanced chemical vapor deposition). In the CVD process the metal nuclei are carried away from the plasma reaction zone and coagulate as in other gas condensation processes, while in the CVC process the precursors are transported into a bubbler and then passed into a heated furnace by He gas, where they are subsequently decomposed. Sometimes, powder precursors such as oxide powders have also been used in the CVD process, when the oxides are reduced to their respective metals. Particles with 5–50 nm size of Mn, Fe, Co, Zn, and Mo were produced by reducing their respective oxides.[53] Metal alloy nanoparticles such as Fe-Ni and Fe-Cu have also been produced in a 50% H2–50% Ar plasma.[54] Choi et al.[55] used Fe(CO)5 and Co2(CO)8 to produce nanoparticles of Fe and Co, respectively, through the CVC process. Thus metal oxides and nanoceramics are the most significant products of the CVC process.[56,57] Vaporization methods such as laser evaporation[58] and sputtering,[59] however, provide better control of the evaporation parameters and are discussed in detail in the following section.

2.6.2 Laser Vaporization The laser vaporization technique uses a laser to evaporate a sample target in an inert gas flow reactor.[60] The source material is locally heated to a high temperature, enabling vaporization. The vapor is then cooled by collisions with the inert gas molecules and the resulting supersaturation induces nanoparticle formation. Kato[61] produced nanoparticles of many complex refractory oxides such as Fe3O4, CaTiO3 and Mg2SiO4 using a 100 W continuous-wave CO2 laser. The particle size varied between 6 and

58

Processing

100 nm. Nanocomposite films were also produced by simultaneous evaporation of a metallic and a ceramic target using a CO2 laser.[58] Laser vaporization should be distinguished from laser ablation. In the case of laser ablation, apart from atoms and ions, fragments of solid or liquid material are also ablated from the substrate. Laser ablation is a technique in which a pulsed laser rapidly heats a very thin (<100 nm) layer of the substrate material resulting in the formation of energetic plasma above the substrate. The metal vapor is generated by the pulsed-laser ablation of a target. Laser ablation can vaporize materials that cannot readily be evaporated. The source is generally a pulsed source that is able to produce clusters of almost any metal, including refractory metals. The duration and energy of the pulse determine the relative amounts of the ablated atoms and particles. Short-pulse (10–50 ns) laser heating enables the synthesis of nanoparticles of materials that normally decompose when vaporized directly (e.g., most semiconductor and superconductor materials). Typically Nd:YAG lasers (with a wavelength of 532 nm) and excimer lasers (with wavelengths of 191, 248 and 308 nm) are used. Synthesis of nanoparticles appears to be a function of the pressure. Yamamoto and Mazumder[62] reported that while nanoparticles were not produced when NbAl3 was laser-ablated at a He pressure of 0.1 mbar, 6-nm nanoparticles could be produced when the operating pressure was increased to 1 mbar. Johnston et al.[63] carried out reactive laser ablation by ablating an Al target in an O2 (reactor gas) atmosphere, producing Al2O3 nanoparticles. Some other recent examples include preparation of nanoparticles of magnetic oxides,[64] titania,[65] and hydrogenatedsilicon.[66] Laser methods provide several advantages, such as production of high-density, high-speed, directional vapor of any metal in an extremely short time (10−8 s).

2.6.3 Sputtering Sputtering involves vaporizing materials from a solid surface by bombarding a target with high-velocity ions of an inert gas, which causes ejection of atoms and clusters. Since high pressures hinder transportation of the sputtered material, sputter sources such as an ion gun or a hollowcathode plasma sputter source are normally used under vacuum. Positively, negatively, as well as neutrally charged clusters are produced. The clusters produced are hot and cool down by evaporation in flight. Since the intensity distribution tails off exponentially with cluster size, this tech-

2: Synthesis, Suryanarayana and Prabhu

59

nique is only suited to producing clusters of small size. Despite this limitation Fayet et al.[67] produced platinum clusters of up to 10 atoms using a 30 keV xenon ion gun. When electrons from an electron gun, instead of ions, are used, the technique is called electron beam heating. Iwama et al.[68,69] produced TiN and AlN nanoparticles, smaller than 10 nm in size, by evaporating Ti and Al targets in a nitrogen or ammonia atmosphere with an electron gun at 10−5 mbar. Günther and Kumpmann[70] produced 5 nm Al2O3 and SiO2 particles by applying an electron beam in an inert gas atmosphere. The difficulty of soft-landing ions on a substrate, due to the large energy spread of the sputtered ions, is one of the problems associated with sputtering. The main advantage of sputtering is that there is no difference in the composition of the sputtered material and the target. Nanoparticles of refractory metals with vapor pressures too low for Joule heating and deposition of thin films[71] can be generated by plasma sputtering. High-energy ions of inert gas from direct current (DC) or radiofrequency (RF) power sources are accelerated toward the target and bombard it, causing the ejection of a small fraction of ions and predominantly neutral atoms from the target surface. Hahn and Averback[59] demonstrated that a DC or RF magnetron sputter source could be operated in the millibar range, ejecting nanoparticles of sizes between 5 and 20 nm. (When a static magnetic field is applied near the target of DC power sources to produce high-density plasma, it is known as magnetron sputtering.) DC magnetron sputtering was used to synthesize 15–60 nm Cu and 5–25 nm SnO2 particles.[72] Terauchi et al.[73] used RF magnetron sputtering to synthesize Au nanoparticles.

2.6.4 Electrical Arc Discharge Electrical arc discharge is based on laser vaporization, but an electric arc is used to vaporize the material directly, rather than indirectly through a laser pulse source.[74] About 10% of the clusters formed are ions, and this avoids the need for a separate cluster ionization stage. Generally, cluster sizes of up to around 50 atoms are deposited on the target. Uyeda[32] used the arc discharge heating to produce ultrafine powders of Fe, Si, SiC, and Al2O3. When the arc current is increased, spattering of the molten droplets takes place along with the formation of smoke, but large droplets separate from the smoke. Gas pressure and arc current are the critical parameters that need to be controlled during evaporation to obtain the desired particle size.

60

Processing

2.6.5 Plasma Heating Plasma heating is another means of providing the energy necessary to induce reactions that lead to supersaturation and particle nucleation.[75] The plasma generally decomposes the material fully into atoms, which can then react or condense to form particles when cooled by mixing with cool gas or by expansion through a nozzle. The substrate is heated by a plasma jet flame and nanoparticles are formed upon cooling while exiting the plasma region. The plasma temperatures are in the order of 104°C, decomposing the reactants into ions and dissociating atoms and radicals. The main types of thermal plasmas are DC plasma jets, DC arc plasmas, and RF induction plasmas.[76] It is important that the droplet sizes are small in order to obtain complete evaporation, as the residence time of the droplets in the plasma is very short. High cooling rates of the plasma gas in the condensation region of the reactor guarantee homogeneous nucleation. Complex materials such as multicomponent oxides can be obtained by using appropriate mixtures. Solid powder feeds can also be decomposed by the plasma. Heberlein et al.[77] have applied these methods to the production of nanoparticles of SiC and TiC for nanophase hard coatings. The plasma gas condensation technique (a combination of sputter vaporization and IGC techniques) was used to produce fine clusters of Nb and NbN in the size range 3–13 nm in diameter.[78] Yoshida et al.[79] developed the “hybrid” plasma concept, which is characterized by the superposition of RF plasma and a DC arc jet, and applied it to synthesize SiC powders.

2.7 Particle Transport Particle transport takes place either by convection current due to the temperature difference between the source and a cold surface or by a combination of a forced gas flow and convection current. Natural convection current flow was used for streaming an elemental reactant into the condensation chamber in the early IGC processes. Although an increase in heating drives natural convection faster, the flow rate is insufficient to overcome the concomitant increase in metal evaporation rate, since the vapor pressure increases exponentially with temperature. As a result, crystallites become larger and aggregate. Therefore, cooling of the powder collection device to liquid-nitrogen temperature may be desirable to achieve a small grain size. In addition to the limitation of natural convection, oxidation of metal crystallites into metal-oxide ceramics must

2: Synthesis, Suryanarayana and Prabhu

61

occur in a separate processing step. This type of oxidation process is not suited to commercial-scale production, and in fact, oxidation is often incomplete.[80,81] Andres et al.[82] and Siegel[83] proposed the utilization of forced gas flows in the IGC process for production of larger quantities of nanoparticles. Haas et al.[84] modified the classical IGC process by introducing a continuous inert-gas jet above the heated crucible. The particle transport by forced convection resulted in a reduction of number concentration and residence time of the nuclei so that smaller nanosized powders could be produced. Forced convection (i) allows control of the particle/gas stream by eliminating the ill-defined flow pattern of natural convection, (ii) speeds the particle transport from the particle growth region to increase metal vapor generation, and (iii) completes the reaction of metal crystallites to form oxides and nitrides. Thus, forced gas flow is used in sophisticated cluster synthesis methods to produce, even if with low yields, very narrow and even monosized size distributions.

2.8 Particle Collection Particles are collected by thermophoresis on the cold surface above the source. The thermophoretic driving force is the result of a temperature gradient, which makes the particles move in the direction of decreasing temperatures. After deposition, the powder is scraped off into the funnel, collected in a die, and finally compacted into a nanostructured solid. Most applications of the IGC method carry this approach to extremes by cooling the deposition substrate with liquid nitrogen to enhance the deposition efficiency. Rotation of the cold finger helps in achieving a better mixing of the vapor, especially when alloys or compounds need to be produced. Transmission electron microscopy (TEM) observations can be made by collecting the particles on a TEM grid located on or near the cold surface, and thus sample preparation for TEM is not required. In case of thicker deposits, small angle x-ray scattering (SAXS) can be used for the investigation. The compaction unit is mounted on a metallic stand and kept at the bottom of the chamber. The die for making the pellet is loaded on a rackand-pinion set-up. In the compaction unit, the die moves into three positions to pellet the powder: (i) powder collection position, (ii) powder compaction position, and (iii) pellet ejection position. Typically, two handoperated hydraulic presses are mounted, one on the head of the powdercompaction unit and the other on the head of the pellet-ejection unit. These movements of die and the plunger unit can be viewed through quartz

Processing

62

view-ports while the hydraulic presses are in operation. The compaction unit is evacuated separately using a turbo-pump and isolated from the UHV chamber with a gate-valve. The chamber has provisions for making multilayers using RF/DC sputtering.

2.9 Nucleation and Growth During vapor-phase synthesis of nanoparticles, conditions are created in which the vapor phase is thermodynamically unstable relative to the formation of the solid material to be prepared in nanoparticulate form. Thus, particles will nucleate homogeneously if the degree of supersaturation is sufficiently high and the reaction/condensation kinetics permit. Once nucleation occurs, the remaining supersaturation can be relieved by condensation or reaction of the vapor-phase molecules on the resulting particles, resulting in particle growth rather than further nucleation. Thölén[44] conducted extensive investigations into the study of nucleation and coalescence of clusters. The particle formation occurring directly from vapor is, however, assumed to take place via homogeneous nucleation. Homogeneous nucleation occurs in the absence of any foreign particles when the vapor molecules condense to form droplets. The free energy for nucleation is the sum of two terms: a positive contribution from the surface free energy and a negative term from the bulk free energy difference between the supersaturated vapor and the solid. The thermodynamics of nucleation are often modeled with the bulk free energy change of the system:[85] ΔG = 4 πr 2 σ −

4 3 πr ΔGv 3

Eq. (2-1)

where ΔG is the Gibbs free energy of the system, ΔGv is the free energy change per unit volume on going from the vapor to the solid phase, and σ is the surface free energy of a nucleus per unit area. The nucleation and growth of particles is, however, determined primarily by the local cooling rate, the residence time distribution, and the number density in the nucleation and growth zone.[86] The local cooling rate (defined as the rate at the location of the nucleation burst) influences significantly the homogeneous nucleation. According to the Kelvin equation r∗ =

4γΩ kT ln S

Eq. (2-2)

2: Synthesis, Suryanarayana and Prabhu

63

where r* is the radius of the critical nucleus, γ is the surface tension, Ω is the atomic volume, k the Boltzmann constant, T is the absolute temperature, and S is the supersaturation ratio. A high supersaturation causes a decrease in the critical nucleus size, resulting in smaller embryos in the prenucleation size distribution becoming stable. The growth time is determined by the effective cooling rate in the growth region of the reactor. Short growth times are achieved by rapid cooling. However, the steadystate nucleation theory is not applicable[87–89] in cases of extremely high supersaturation since the nucleus may contain only a few atoms and it is unreasonable to treat the nucleus as a macroscopic entity with microscopic properties such as surface tension. Nanoparticles usually continue to grow after nucleating by acquiring more atoms and growth occurs by coalescence when the particles collide with each other where metal atoms lose their kinetic energy. Coalescence takes place provided that the temperature is high enough and the surfaces of the particles are clean. If the density of the particles is relatively low and the collection time is short, then the particle agglomerates are small. However, after the particles are formed, they either coalesce or coagulate. While coagulation is a stepwise process wherein two nuclei meet and are joined, causing sintering due to processes such as Brownian motion, coalescence refers to the sintering and diffusion of species within particles where contact is made.[90] Granqvist and Burhman[27] assumed that growth of nanoparticles is a stepwise process and that the primary particle size is a result of a number of coalescence steps. Their model was based on the statistical coalescence of nuclei. They found that coalescence is the dominant mechanism and indicated how coalescence leads to a log-normal distribution for the particles (Fig. 2.3). Nakaso et al.[91] carried out a detailed experimental and theoretical study of the in-flight sintering (coalescence) of gold nanoparticles after evaporation/condensation synthesis. Once particles form in the gaseous phase, they coagulate at a rate that is proportional to the square of their number concentration and that is only weakly dependent on particle size. At sufficiently high temperatures, particles coalesce (sinter) faster than they coagulate, and spherical particles are produced. At lower temperatures, where coalescence is negligibly slow, loose agglomerates with quite open structures are formed. At intermediate conditions, partially sintered nonspherical particles are produced. The rate of coalescence of the small crystallites formed in the inert gas just above the evaporation source controls particle growth in the gaseous phase. The coalescence rate depends on the particle number concentration and the residence time in the hot zone. Coalescence as well as grain growth can be restricted by covering the surface of the crystallites with a second element.

Processing

64 150

Number of particles

LNDF’s x = 4.1 nm s = 1.60 s = 1.55 Al data

100

50

0 0

5

15 10 Particles diameter in nm

20

Figure 2.3 Particle size distribution for aluminum cluster samples. The dots represent the mid-interval values obtained from the size histograms. The curves represent the log-normal distribution functions for two different values of the geometric standard deviations. The arrow indicates the median particle diameter x¯.[27]

Haubold[92] succeeded in preparing nanostructured Cu doped with Bi. The Cu-Bi binary alloys are known to be immiscible. When a melt of Cu and Bi is rapidly quenched, Bi is found to segregate to grain boundaries of the Cu crystallites. The crystallites produced by this method are subunits of larger agglomerate particles, often with substantial neck formation that interferes with consolidation. Neck formation in those agglomerate particles that do form is diminished by starting the growth process at a high initial temperature. Examination of the theory of particle growth under such conditions reveals that once agglomerate particles begin to form, particle growth rapidly accelerates. The crystallite size is, therefore, determined by growth prior to the onset of agglomeration.

2.10 Limitations of the Classical Nucleation Theory Despite the success of the classical nucleation theory, questions remain over its applicability to clusters of only a few atoms. It is questionable whether surface tension is really an appropriate way of modeling on an atomic scale and whether the bulk value of surface tension is the correct

2: Synthesis, Suryanarayana and Prabhu

65

value to use for small clusters. Kappes et al.[93,94] pointed out that that the theory ignores the rotational degrees of freedom. Another concern is the cluster size. Clusters smaller than the critical size, according to the theory, should not be stable and should evaporate spontaneously. For example, even though the critical size of silver clusters in the nucleation zone is around 100 atoms, beams containing much smaller clusters than this are routinely generated by this technique. The reason is probably that the classical nucleation theory does not allow for temperature changes in the system. A small cluster leaves the growth region before it reaches the critical size but does not evaporate if it contains insufficient internal energy to do so. In the IGC method, cooling by the inert gas removes much of the initial thermal energy of the vapor. Also, most real systems have timevarying properties that cannot be accommodated easily into the classical nucleation theory and this leads to the questionability of the classical theory. A method to overcome such drawbacks is the use of Monte Carlo simulation methods in which many growth processes are incorporated into one model.[95–97] A well-known example of this method is the simulation of the Brownian movement by stochastic processes. When sufficient number of particles is taken, collisions will result from the random Brownian motions and a change in the particle size distribution can be recorded. Coagulation rates can be calculated with the help of such trajectory methods based on the Langevin equation, Brownian coagulation of droplets,[98] and agglomerate formation by Brownian coagulation.[99] But such methods are time-consuming and are not suited for simulating a large number of particles.

2.11 Crystal Structure and Morphology Granqvist and Burhman[27] noted that the size and crystal habit of the nanoparticles depended on the inert-gas pressure and species, and defined a particle growth region above the surface of the evaporation source. The amount of pressure in the growth region and the time the nuclei will spend in the growth region essentially determine their size and morphology. While a high pressure in the growth region results in larger particles due to agglomeration and less sintering, low pressures result in smaller particle sizes. Essentially the structure and size of the particles varies depending upon the zone where they are collected. There are three zones, which have different temperatures and metal vapor concentrations. At sufficiently high temperatures, particles coalesce (sinter) faster than they coagulate, and spherical particles are produced. At low temperatures, where

Processing

66

coalescence is negligibly slow, loose agglomerates with quite open structures are formed. 150 The grains in the nanophase compacts are rather equiaxed and retain their narrow, log-normal size distributions. N 100 Figure 2.4 shows a typical grain size distribution for a nanosized rutile sample compacted at room temperature and a 50 pressure of 1.4 GPa.[36] Although an equiaxed grain structure is expected to 0 minimize the surface area, in some 0 5 10 15 20 25 30 cases,[37] such as MoO3, the particles Grain diameter, d in nm were in the form of plates and needles. Figure 2.4 Grain size Kaito et al.[37,41] demonstrated how the distribution for a nanostructured nature of these particles could be altered TiO2 (rutile) sample compacted by heating the plume above the source. at room temperature and 1.4 They reported that, depending upon the GPa pressure. The grain sizes position relative to the center of the were determined from dark-field smoke plume, a range of particle sizes electron microscopy experiments.[36] and structures were seen above a certain height, but the sizes on any streamline did not appear to change appreciably. However, if a second heater was placed above the source wire so that the plume passed between heated grids, the crystallite size increased. This was attributed to the coalescence of the crystallites that make up the complex aggregate particles present in the plume. These aggregate particles are useful in some applications, such as reinforcing rubber and other polymers. However, the hard agglomerates severely hinder consolidation of the metal; this is especially true for ceramic powders. In most cases the particles have the shape of polyhedra as a result of the kinetics of growth and also to minimize the surface energy. The large surface area of these nanoparticles makes them highly reactive and factors such as temperature, composition, impurities, and surface energy result in some unusual crystal structures, shapes, and sizes. Krauss and Birringer[100] synthesized metastable β-W and β-Ta nanoparticles by varying the level of undercooling and cooling rate of the metal nuclei. They also synthesized either the cubic or the monoclinic phase of Y2O3 by changing the process conditions, where pure Y was evaporated and then oxidized in a separate step. Saito et al.[101] found that the W nanoparticles had an A15 structure and that Cr and Mo had either A15 or BCC structure. Stappert et al.[102] 200

TiO2

2: Synthesis, Suryanarayana and Prabhu

67

Table 2.2 Comparison of the Crystal Structures of Different Materials in the Coarse-grained and Nanostructured Conditions

Material

Crystal Structure Coarse-grained Condition

Chromium Tantalum Tungsten Y2O3

Reference

Nanostructured Condition

BCC BCC BCC

β-W type structure β-W type structure β-W type structure

[101] [100] [100, 101]

Monoclinic

Cubic

[100]

observed that the morphology of the FePt nanoparticles changed from irregular particle shapes, when sintered at room temperature, to increasingly spherically shaped particles with increasing sintering temperature. Bose et al.[103] showed the existence of a high-pressure orthorhombic phase when SnO2 was annealed. Michels et al.[104] observed that with decreasing crystallite size Tc shifted to lower temperature and the Curie transition took on an increasingly broadened and rounded appearance. Wang et al.[105] observed transformation from austenite to martensite when a nanograined Fe-12 at% Ni was cooled down to 100ºC. Dong et al.[106] produced Co powders with long striplike or spherical morphologies, with the coexistence of FCC and HCP type structures. Table 2.2 compares the crystal structures and/or microstructural changes when the material is in the coarse-grained or in the nanocrystalline state.

2.12 Influence of Process Variables on Particle Size Reasonably strict control is required over the different process parameters such the inert-gas type, pressure, flow rate, evaporation rate, and temperature, which determine the formation and size of nanoparticles. The fact that clusters are formed at all shows that the incoming helium gas is significantly cooler than the metal. In practice, inert-gas temperatures from 77 K up to room temperature have been used, room temperature being adequate for the formation of small to medium-size clusters. Inert-

68

Processing

Table 2.3 Trend of Average Particle Size Change with an Increase in the Process Parameters

Parameter (increasing) Inert-gas pressure Inert-gas temperature Inert-gas molecular weight Inert-gas flow rate Crucible temperature Growth distance Evaporation rate

Average particle size Increases Decreases Increases Decreases Increases Increases Increases

gas pressures range upward of a few tens of millibars. The dependence of the mean particle size on condensation parameters such as the temperature of the metal,[27] type of the inert gas,[27] pressure of the inert gas,[27,30,107] and geometry of the condensation chamber[30,107] have been investigated. Table 2.3 summarizes the dependence of average particle size on these different parameters, and Table 2.4 lists the different types of materials produced via IGC in the nanostructured condition. A variety of materials have been synthesized in the nanostructured condition using the IGC method. These include metals, alloys, compounds, semiconductors, and ceramic materials. While it has been relatively easy to synthesize metals, especially those of relatively high melting points, in the nanostructured condition, compounds and alloys have also been synthesized. It is in the synthesis of these materials that the process parameters need to be carefully controlled to achieve the desired size and shape of the powder particles. Let us now discuss the effect of the different process parameters on the particle size obtained.

2.12.1 Inert-gas pressure The size of the particles varies with changes in the gas pressure.[29,30,144] It was suggested that it is the partial vapor pressure of the precursor, rather than the overall system pressure, that determines the powder particle size.[90] The total pressure, however, serves to regulate the diffusion

2: Synthesis, Suryanarayana and Prabhu

69

Table 2.4 Different Types of Materials Produced by IGC and Their Particle Sizes

Substance

Pure Metals Ag

Average Particle Size (nm)

Reference

Substance

5–20 [108] Mg 10–180 [109] Mn Al 5 [27, 100] Nb Au 1 [73, 110] Ni Bi 50 [111] Pd Co 5–50 [27, 112] Sb Cr 7.6–13 [27, 113] Si Cu 15–30 [27, 114] Sn 15–60 [72] Ta Fe 4–20 [27, 115] Te 15 [116] Ti Ga <20 [27] W Gd 12 [104] Zn Ge 10 [117] Compounds (Oxides, Nitrides, Fluorides, etc.) AλN <10 [68, 69] Mn(O) α-Al2O3 5 [70] NbN α-Al2O3 18 [43] NiFe2O4 α-Al2O3 30 [126] PbF2 CaF2 8(?) [127] SiOx CaTiO3 – [61] SiO2 CdS 16 [128] SnO2 CeO2 10 [129] CoO 50 [130] TiN Fe3O4 2 [61, 131] TiO2 Fe2O3 4 [132] Y2O3 GeOx 5–20 [117] ZnS MgO 5 [46] ZrO2 Mg2SiO4 – [61]

Average Particle Size (nm)

Reference

40 5–50 3–13 10–100 10 2–10 15 30 5–20 13 10 10 <20

[27, 118] [119] [78] [27, 120] [121] [122] [117] [27, 123] [100] [124] [43] [125] [27]

27–43 3–13 90 21–43 5–20 5 5–25 6–10 <10 12 <20 7–9 <10

[119] [78] [130] [133] [117] [70] [72] [103] [68, 69] [36] [134] [135] [136]

Processing

70

Table 2.4 Different Types of Materials Produced by IGC and Their Particle Sizes (cont’d)

Substance

Average Particle Size (nm)

Intermetallics and Alloys AgAl 200 Fe3Al 7 FePt 7–9 Ni3Al 6 NiAl 70–120 TiAl 15 TiAl 10–20

Composites Ag/Fe2O3 Ag/PbS CuOx/CeO2

2–3 10

Reference

Substance

Average Particle Size (nm)

Reference

[130] [137] [138] [137] [130] [137] [139]

Ag—Al Al—Zr Co—Fe—Ni Cu—Sn Fe—Al Fe—Co Fe—Cu Fe—Gd Fe—Ni Ni—Co Ni—Cu Pb—Sn

20 10 30–90 100 200 20–100 40–120 100 27 20–100 40–100 100

[130] [140] [130] [130] [130] [130] [130] [130] [105] [130] [130] [130]

[141]

Fe/Fe2O3 W/Ga

13 ± 2 10

[143] 125

[142]

of vapor from the growth source. Diffusion increases when the total pressure is lowered due to a wide dispersion of the particles, thus restricting the growth by coagulation. In general, an increase in the inert-gas pressure leads to an increase in particle size.[27] This is because an increase in the inert-gas pressure slows the diffusion of metal vapor away from the crucible so that the vapor is enriched. Hence, growth predominates over nucleation and larger clusters form. High pressures shorten the mean free path of atoms, effectively increasing the quenching rate, which leads to enhanced nucleation. If the inert-gas pressure is reduced, the vapor diffuses away from the crucible faster and the nucleation zone becomes more spread out. The lack of vapor prevents clusters from growing very large. As the pressure is reduced, the mean particle diameter increases and the distribution becomes log-normal, suggesting that a

2: Synthesis, Suryanarayana and Prabhu

71

combination of reduced nucleation rate and particle-particle coalescence is responsible for the observed behavior. In case of composites, both the crystal size and the secondary particle size of the composite were controlled by the inert-gas pressure during the IGC synthesis, with larger sizes associated with higher pressures.[142] The helium flow pressure facilitates the nucleation and growth of the clusters. At the same time, the helium flow also helps to direct the cluster beam through the nozzle and then the skimmer to the deposition chamber. The average particle size of the powder was 12 nm at 1.0 Torr and 26 nm at 10 Torr.[141] The nucleation and growth of Fe and Ag particles was closely related to the concentration of inert helium gas, and readily took place as a result of the higher frequency of collision between the vapor atoms at a higher helium gas pressure.[141] It was also reported that control of the inert-gas pressure affects not only the particle size but also the constitution of the resulting material.[145] It was possible to obtain evidence of a phase change by observing the color of the powder before and after exposure to oxygen. If the He pressure was maintained at ≥500 MPa, the ultrafine Ti particles produced always appeared black on the cold finger prior to oxidation. When this powder was exposed to oxygen, a noticeable “flash” was observed as the material spontaneously oxidized; the resulting TiO2 (rutile) always had a characteristic bluish-white color. In contrast, if the He pressure was <500 MPa, but greater than about 10 MPa, small particles of Ti were still produced, but these particles were gray-colored both before and after exposure to oxygen. Further, the Ti particles were large in size and did not transform to TiO2 when exposed to oxygen. Instead, an unexpected amorphous phase formed. No noticeable effect of He pressure was noted in the case of Pd powders produced using a wide range of He pressures.[145]

2.12.2 Inert-Gas Temperature A rise in the inert-gas temperature leads to a decrease in the temperature gradient near the crucible and the nucleation zone moves away from the crucible to a region of lower vapor density, resulting in smaller clusters. When the inert gas is very cold, most of the clusters are nucleated in a region of high vapor density, leading to rapid growth and large ultimate sizes since the temperature gradient close to the crucible is steepest. However, since nucleation does not appear to require a particularly large

72

Processing

drop in temperature, one would not expect the temperature of the inert gas to have a particularly dramatic effect.

2.12.3 Inert-Gas Type The choice of inert gas along with the pressure and temperature gradient is an important parameter to control the diffusion rate. An inert gas is generally used because the frequent collisions of the metal vapor atoms with the gas atoms decrease the diffusion rate of atoms away from the source region. These collisions cool the metal atoms; consequently the diffusion rate is reduced and this allows achievement of supersaturation. The inert gases (Ar, He, Ne, and Xe) limit diffusion by shortening the mean free path. Heavier gas atoms are most effective in limiting the mean free path and confining the metal vapor. Studies have shown that while all inert gases provide similar effects, the use of helium gas produced much smaller particle sizes than the use of argon or nitrogen.[29] It was proposed that, due to the lower mass of helium, collisions between the gas atoms and the metal atoms did not provide the same supercooling as the larger inert-gas species and did not confine the particles to the growth region as effectively as the more massive species. Thus, helium allowed more metal to be removed from the growth region, resulting in smaller particle sizes. The metal particles are therefore able to escape from the evaporation source more easily, thus condensing in a region of fewer metal nuclei, giving smaller sizes.[29] Granqvist and Buhrman[27] had clearly demonstrated the effect of increasing the mass and pressure of the carrier gas on the nanoparticle size. They showed that the particle sizes of Al and Cu increased when the mass of the inert gas increased on going from He to Xe (Fig. 2.5). In the case of production of Co particles also it was reported that sputtering with He resulted in smaller nanoparticles than when sputtering with the heavier gas argon.[146] Dong et al.[106] studied the effect of the carrier gas (argon and helium) on the size and constitution of Co particles obtained by chemical vapor condensation. They noted that an oxide layer covered the metallic core of Co in both the cases, but the cobalt powders produced in the Ar atmosphere possessed spherical or long-string morphologies with an inner metallic core and an outer cobalt oxide. Both the FCC and HCP-type structures of Co were found to coexist in this case. On the other hand, the Co nanoparticles synthesized in the He carrier gas consisted of finer cores of metallic Co homogeneously distributed in a mass of cobalt oxide, and the inner Co metal exhibited only the FCC-type structure. Further, the size

2: Synthesis, Suryanarayana and Prabhu

73

MEDIAN PARTICLE DIAMETER (nm)

200 Al : Open symbols Cu : Filled symbols

Xe

100

Ar 50

He 20 He gas Ar gas Xe gas

10

1

2

5

10

20

50

100

200

INERT GAS PRESSURE (torr)

Figure 2.5 Influence of type of gas and the inert-gas pressure on the particle size produced in the inert-gas condensation method.[27] Note that the particle size increases with increasing inert-gas pressure and also with the atomic weight of the inert gas. The straight lines serve only as a guide to the reader.

of metallic Co particles produced in Ar was larger than that in He due to the higher atomic weight of Ar.

2.12.4 Inert-Gas Flow Rate Increasing the inert-gas flow rate reduces the length of time that the clusters spend in the growth region of high vapor density and consequently leads to a decrease in the particle size. They have less time to grow and so are smaller. It also increases the effectiveness of cooling, so that the temperature gradient is larger near the crucible and more clusters are nucleated. This parameter, coupled with the growth distance, probably has the most dramatic effect on the mean size of the clusters. The increase of mass flow rate with the inert-gas pressure is due to the efficiency of the inert gas in entraining the clusters, removing them from the crucible region and transporting them out of the condensation cell. At low pressures many clusters may be deposited on the walls of the condensation cell; but at higher pressures they are swept along in viscous flow with the inert gas. Brock et al.[147] found that if the inert-gas jet was turbulent, the particle sizes were smaller and had a higher number concentration than for laminar jet flows. Yamamuro et al.[113] observed a wider particle size

Processing

74

Figure 2.6 Bright-field TEM images and electron diffrection patterns (insets) of as-deposited (a) Fe51Pt49 and (b) Fe49Pt51 alloy clusters with the mean cluster size d ∼ 7 and 9 nm, corresponding to Ar gas flow rate RAr = 300 and 400 sccm, respectively.[138]

distribution for increasing inert-gas flow rates in a forced-flow IGC system, and therefore there was no decrease in the average particle size. They attributed this behavior to the gas flow disturbing the growth and thermalization process.[113] Peng et al.[138] observed decreasing particle size with increased Ar gas flow rate (Fig. 2.6).

2.12.5 Evaporation Rate Evaporation rate is the mass evaporated per unit area in unit time. The production rate is determined mostly by the evaporation rate. High evaporation rates have resulted in larger particles.[148] The evaporation rate (Wg) in a gas atmosphere is given by M ⎞ Wg = ( ps − p)⎛ ⎝ 2 πRT ⎠

12

Eq. (2-3)

where ps = saturation pressure of metal at temperature T, p = partial pressure of metal vapor, M = molecular weight of evaporated metal, R = gas constant, and T = evaporation temperature. Huh et al.[149] showed that the evaporation rate and also the size of the synthesized particles increased with increasing evaporation temperature, as shown in Table 2.5. The population of the atoms in a given space in

2: Synthesis, Suryanarayana and Prabhu

75

Table 2.5 Effect of Evaporation Source Temperature on Evaporation Rate and Size of Synthesized Tin Particles[149]

Temperature (°C)

Evaporation Rate (g/s)

Particle Size (nm)

800

8.65 × 10−5

1000 1100

4.75 × 10−4 7.32 × 10−3

6.6 17.6 27.3

which particles form can be attributed to the dependence of the size of the nanoparticles on the evaporation rate.

2.12.6 Nozzle Diameter The size of the atomic cluster is expected to increase on decreasing the size of the nozzle because of adiabatic expansion and subsequent cooling of the vapor. Raviprasad et al.[122] showed that when the nozzle diameter and helium pressure were maintained at 3 mm and 5 mbar, respectively, they were able to get Sb deposition with a steady deposition rate of 0.1 Å/s, as detected by the thickness monitor. The Sb vapor was deposited mostly on the inner side of the tantalum heat shield, due to the back flow caused by the increased helium pressure.

2.12.7 Chamber Size and Growth Distance IGC operation in a large chamber leads to the deposition of nanostructured particles on a large area of the cold finger. The deposition of assynthesized particles on a large area of the cold finger is preferable for easy collection of particles. Sufficient free space is required not only for convection in the IGC chamber but also for the collection of nanophase particles. However, if the condensation chamber is very large, then growth may stop when the vapor is exhausted even before the exit is reached. A UHV stainless-steel chamber of about 0.6 m diameter and 0.6 m in height was more recently used to produce nanopowders.[103] However, this is not a fixed parameter and varies based on requirements. The chamber can be made to lift about 1 m in height from the base using a hydraulic lift, which is operated using a power-pack system. Growth distance is the distance

76

Processing

from the crucible to the exit of the condensation. Huh et al.[149] studied the convection currents simulated with different chamber heights.

2.12.8 Effect of the Position of the Evaporation Source The effect of the position of the evaporation source was determined by varying the location of the resistive boat. The evaporation rate was virtually unaffected by change in the position of the evaporation source. However, the deposition behavior of the nanophase particles on the cold finger and chamber wall was influenced by the position of the evaporation source. When the resistive boat was placed at a position close to the cold finger (6 cm from the center of the chamber), concentrated deposition of particles on a small area of the cold finger occurred.[149] In contrast, a considerable part of the deposition took place on the chamber wall during IGC operation when the resistive boat was kept far away from the cold finger.

2.12.9 Effect of Reactive Gases Reactive gases such as oxygen or nitrogen are used to form nano oxides, nitrides and other compounds. The particles are introduced into oxygen, nitrogen, or carbonaceous atmospheres, when they form the respective oxides, nitrides, carbides, etc. Metal oxides are formed in a two-step process: first the metal is formed and this is followed by subsequent oxidation by the addition of oxygen.[43] For example, to produce nanostructured TiO2, Ti metal clusters were first collected on the cold finger and subsequently oxidized by introduction of oxygen into the chamber.[36] A similar method was used to produce α-Al2O3 by oxidizing Al clusters in air at 1000°C.[150] The addition of oxygen results in particle coating and thereby reduces surface diffusion and sinter neck growth rate. This is believed to suppress the sintering and growth of powders during processing, which also changes the surface tension of the particles and results in the formation of fine spherical particles.[109] It was also noted[109] that the addition of oxygen to He gas had a marked effect on the size and morphology of the Ag nanopowders produced. The size of particles produced in pure He ranged from 15 to 180 nm, with a mean particle size of about 75 nm (Fig. 2.7(a)), while the powders produced in the He atmosphere containing some (1–20 cm3/min) oxygen had finer particles compared to

2: Synthesis, Suryanarayana and Prabhu

77

90

Number of Particles

80 70 60 50 40 30 20 10 0 0

15 30 45 60 75 90 105 120 135 150 165 180 200

Particle size (nm)

a

Number of Particles

35 30 25 20 15 10 5 0 1

10

20

30

40

50

60

70

80

90 100 110 120

Particle size (nm)

b Figure 2.7 Size distribution of Ag nanopowders produced in (a) 20 mbar He atmosphere, (b) He with 10 cm3/min flowing O2 atmosphere.[109]

those produced in pure He atmosphere. The particle size of the powders produced in 10 cm3/min O2 ranged from 10 to 120 nm with a mean particle size of about 44 nm (Fig. 2.7(b)). It was observed that with increasing additional oxygen in the carrier gas, the mean size of primary particles strongly decreased and resulted in an increase in the specific surface areas from 6.5 to 24 m2/g for oxygen addition of 0–20 cm3/min. Figures 2.8(a) and (b) show scanning electron micrographs corresponding to the abovementioned conditions. Residual oxygen and water vapor act as impurities in the gaseous atmosphere throughout the process at concentrations in the ppm range. This may lead to the formation of a surface oxide layer. If the oxygen level is too high, this oxide sublimes and is introduced into the powder product as an integral component that cannot subsequently be removed.

Processing

78

Figure 2.8 SEM micrographs of Ag nanopowders produced in (a) 20 mbar He atmosphere, with no additional O2, showing some fine and coarse particles with sinter necks, and (b) 20 mbar He + 10 cm3/min oxygen.[109]

It is believed that the residual oxides adversely affect the properties of nanopowders. Baker et al.[150] synthesized Fe nanoparticles with an ironoxide passivated layer by introducing a controlled amount of air into the chamber after synthesis. Sometimes sputtering in the presence of a reactive gas can lead to oxidative or nitride poisoning of the sputter target, which reduces yields.[151] However, post-sputter oxidation can be carried out in a controlled manner, where the particles are oxidized in the growth chamber after synthesis, though grain growth and agglomeration have been observed.[59]

2.13 Advantages of IGC The IGC method of synthesizing and assembling nanomaterial clusters has the following unique advantages in producing bulk nanostructured materials: (1) A wide range of materials including metals, alloys, intermetallic compounds, ceramics, semiconductors, and composites can be synthesized by this technique. It is possible to produce virtually any material that can be vaporized. (2) IGC is a very flexible technique in terms of the range of cluster sizes that can be made. It is possible to control the size and size distribution of the clusters/nanoparticles over a large range by altering process parameters such as temperature and pressure. Thus, particles of well-defined or

2: Synthesis, Suryanarayana and Prabhu

(3)

(4)

(5)

(6)

(7) (8)

79

predetermined size can be synthesized with enhanced properties. The ultrafine sizes of the particles and their surface cleanliness allow conventional restrictions of phase equilibria and kinetics to be overcome during synthesis by combining short diffusion distances, high driving forces, and uncontaminated surfaces. The possibilities for reacting, mixing, and coating various types, sizes, and morphologies of clusters create significant potential for synthesis of new, more complex, and versatile materials. Particle synthesis is usually a continuous process, while liquid-based/mechanical synthesis processes are often performed in a batch form and can result in product characteristics that vary from one batch to another. Scraping and compaction are carried out under UHV conditions, as a result of which high levels of purity can be maintained. Even though high-purity gases contain some oxygen and water vapor, contamination is much less than in the liquid-based processes since even the most ultrapure water contains traces of minerals, which are detrimental for electronics-grade semiconductors. The ability to incorporate in-situ diagnostics in the system offers more versatility to the system. In-situ analyses using TEM, SAXS, photocorrelation spectroscopy, etc., makes the technique highly flexible.

2.14 Drawbacks of IGC Notwithstanding the above-mentioned specific advantages of the IGC process, the method suffers from the following drawbacks: (1) Agglomeration of particles is a problem in consolidated nanopowders. The van der Waals forces caused by a temporally varying charge distribution in each individual nanopowder particle can cause rapid agglomeration into branched bodies. These entities are difficult to break up on compaction and sintering, and thus lead to interagglomerate voids and residual porosity in the sample. (2) Cost becomes a factor because the entire process is carried out under UHV conditions. In order for the substrate

Processing

80

(3) (4)

(5)

(6) (7) (8)

chamber to be kept under high vacuum during deposition, substantial pumping of a large flow of an inert gas is necessary. The pumps make up most of the cost and bulk of the apparatus. Maintaining a clean vacuum may not always be easy. Ensuring repeatability and stability of operating conditions can also be a problem, since cluster generation is very sensitive to source conditions. IGC is not always suitable for synthesizing compound semiconductors such as GaAs and InP because these easily decompose to the individual components. Neutral clusters produced require ionization, which is usually an inefficient process. It may be difficult to adjust conditions to obtain high fluxes for very small clusters of just a few atoms. It has not been easy to scale-up the process. Consequently, the amount of powder produced per batch is limited.

2.15 Recent Developments in IGC The basic operational principles of gas condensation have not changed significantly since their inception, but major improvements have been achieved in the cluster size ranges obtainable by this method. These have been possible by modifying the evaporation methods, optimizing the geometry of both the condensation chamber and the collection device, and optimizing the different process parameters. Novel and improved materials have also been synthesized. Gleiter and co-workers[152] made significant modifications of the equipment. These modifications include the introduction of a scraper, a funnel for feeding the powder into the die, and the in situ compaction unit for pressing the powder into bulk shapes about 8–9 mm in diameter and 0.1–0.5 mm in thickness. Sanders et al.[153] modified the IGC system designed by Hahn and co workers.[154] The modified chamber had nearly 30% larger diameter than the old synthesis chamber and also had heat shields on both walls and the collector side of the boat. The large Viton seal flange is five times farther away from the source in the modified chamber. Reduction in processing defects such as porosity and impurities was also noted in the powder processed in the modified chamber, which produced large samples with improved density and smaller grain size. However, most developments are focused on the different evaporation and

2: Synthesis, Suryanarayana and Prabhu

81

particle collection techniques. Baker et al.[155] have designed a versatile gas condensation system for nanoparticle synthesis using “reactive gas condensation” to produce ferromagnetic iron oxides and AlN. Chang et al.[156] modified the inert-gas condensation apparatus by replacing the evaporation source with a heated tubular reactor and using a metal organic precursor such as hexamethyl-dilazane (HMDS) instead of a pure metal. On heating, the metal organic precursor decomposes to form a continuous stream of clusters or nanoparticles and these are entrained in a carrier gas. The nanoparticles condense out on a rotating liquid nitrogencooled substrate from which the particles are scraped off and collected. Oda et al.[157] used the gas deposition method to produce fine powders that are deposited at high velocity on the substrate to form a nanostructured specimen without the necessity for a compaction process. Using this technique, the writing of micron-sized patterns was demonstrated. Low-energy cluster beam deposition, used for the preparation of nanostructured films, has received serious attention recently because it can be used to prepare films of high quality, while preserving the novel properties of the clusters.[158] Vapor of material evaporated in a heated BN crucible condenses in a flowing inert gas (helium or argon) cooled with water or liquid nitrogen. The typical pressure of the inert gas is 1–20 Torr. An effusive jet containing a mixture of the inert gas and the clusters is formed through the first skimmer of the condensation chamber, and isoentropic expansion occurs. Then an aligned cluster beam is formed through the second and the third skimmers, and the films are prepared when the beam clusters deposit onto the substrate. In the early IGC processes, metal particles were removed from the hot source by convective flow due to the large temperature gradient from the source to the powder collection device. Convective flow of the particles is generally a low-yield process, and to produce larger amounts of nanoparticles, forced-flow gas condensation techniques were developed.[159,160] Haas et al.[84] developed a process in which a nozzle was used to pass the inert gas in a jet stream over the evaporation source. This process increased production and, by removing the particles more quickly from the growth zone, reduced their size. Wagener and Günther[161] used magnetron sputtering to deposit nanoparticles into low-vapor-pressure liquids such as silicone oil in a process known as vacuum evaporation on running liquids. Surfactants were added to the oil to prevent agglomeration of the particles. Techniques have been developed to coat the nanoparticles[92,162] with other metals as they grow. Konrad et al.[162] prepared nanostructured Cu coated with Bi by the co-evaporation process. In this method, an inert gas

Processing

82

is injected into the synthesis chamber through a nozzle in order to establish a convective flow that carries crystallites of one element above an evaporation source of a second element so that atoms of the second source condense on the surface of the crystallites in the form of a thin coating. In this manner, Cu crystallites were coated with Bi, resulting in Bi-doped grain boundaries after consolidation of the Bi-coated Cu crystallites. The presence of two sources (Fig. 2.9) allows for the production of coated nanoparticles with close control over core diameter and shell thickness. Nakayama et al.[141] and Yamamoto et al.[163] obtained iron oxide–silver nanocomposites using the co-evaporation process. IGC can be used to make clusters of virtually any material that can be vaporized by thermal evaporation, e-beam evaporation or sputtering, including ordered magnetic alloys.[164] Maisels et al.[165] prepared composite nanoparticles of PbS with Ag by separate evaporation and condensation. Ohno[166] prepared Si/In, Ge/In, Al/In, and Al/Pb composite nanoparticles by condensation of In or Pb onto Si, Ge, or Al particles prepared by inert-gas condensation and brought directly into a second condensation reactor. Gold nanoclusters of ∼1 nm diameter were prepared by combining laser vaporization and an inert-gas condensation source.[167]

Cold finger (77 K) Ga W 111In

Thermal evaporation

Ga+ 111In

Sputtering Inert gas (Ar)

Rotating Cold Finger (liquid nitrogen) Scraper

Vacuum pump

Nozzle

Forepump

He

WGa: 111In Compaction unit

Inert Gas (e.g., He) Funnel

(a)

(b)

Figure 2.9 Schematic drawing of a gas-condensation chamber with two evaporation sources, and nozzle and forepump to establish a continuous gas flow.[162]

2: Synthesis, Suryanarayana and Prabhu

83

Very high cooling rates of up to about 1010 K/s, which essentially govern the formation of original nanoscale systems in nonequilibrium conditions, was the unique feature of this type of cluster source compared with the others currently used (thermal, sputtering, etc.). Using a pulsed highpressure inert-gas condensation method, cooling rates as high as 108– 1010 K/s have been achieved, which allow formation of functionalized pure or mixed clusters that cannot be obtained using conventional techniques. Nanostructured samples of Cr were obtained by using the plasma gas condensation-type cluster beam deposition apparatus,[113] which combines both plasma-glow-discharge vaporization and inert-gas condensation techniques. Goldby et al.[168] constructed an IGC system in which massselected metal clusters with a controlled landing energy could be deposited onto substrates to form either clustered materials or thin films. Magnetic nanofluids have also been produced by replacing the cold finger by a drum rotating into a trough containing a fluid–surfactant mixture, which eliminates agglomeration and produces a well-dispersed magnetic fluid in a single step.[169] Laliberte and Boisjoli[170] synthesized unusual complex cobweb-like pattern monomers on a nanometer-scale by laser evaporation. Wolf et al.[125] synthesized W-Ga composites that cannot otherwise be formed by melting because W and Ga do not form an alloy; the composite had completely new properties, different from those of the constituents.

2.16 Conclusions IGC is a simple method involving evaporation of material from a source in a flowing gas, followed by condensation into nanoparticles in the cooler parts of the system. Synthesis of nanostructured materials by IGC yields ultrafine-grained materials with narrow grain size distributions, clean grain boundaries, and excellent resistance to grain growth owing to the high degree of physical and chemical control over various parameters, resulting in narrow particle/grain size distribution. Among the available techniques to synthesize nanostructured materials, IGC has the best means of control over the processing parameters. This chapter essentially brings out the critical dependence of the size of the nanomaterials on the different process parameters. Although the techniques for vaporizing the material differ, the basic idea is to bring the material into the vapor state and then cool it. Despite drawbacks such as cost and small volume production, the technique serves as a useful tool for the synthesis of extremely pure nanostructured materials.

84

Processing

New cluster generation sources are required in the future for generation of larger quantities of material with controlled grain sizes. This will also lead us to the synthesis of more complex multicomponent nanophase materials with properties not achieved so far. The question arises whether huge volume production at viable prices will be feasible, since many current techniques cannot be scaled up sufficiently to target the required volume markets. Aspects such as interface characteristics and in-situ characterization can be expected to develop with progress in the science and technology of nanostructured materials. The fact that these materials can be made at this scale gives them the potential to exhibit some very interesting properties. The current and potential applications for nanoparticles are growing and cover an extremely broad range of markets.

Acknowledgments One of the authors (Balaji Prabhu) would like to thank Dr. P. C. Angelo, Director, Metals Testing and Research Center India, for his continued support and help. The authors also thank Mr. Oliver Friedrichs for a critical reading of the manuscript and for constructive suggestions, Dr. Ismat Shah, University of Delaware for providing valuable information in the writing of this article, Mr. Seelam Umamaheswara Rao for help in drawing the figures used, and Professor Mehmet Türker for providing some of the figures.

References 1. Suryanarayana, C., Internat. Mater. Rev., 40:41 (1995). 2. Gleiter, H., Prog. Mater. Sci., 33:223 (1989). 3. Siegel, R.W., in: Materials Science and Technology–A Comprehensive Treatment: Processing of Metals and Alloys, (R.W. Cahn, ed.), 15:p. 583, VCH, Weinheim, Germany (1991). 4. Mishra, R.S., Semiatin, S.L., Suryanarayana, C., Thadhani, N.N., and Lowe, T.C. (eds.), Ultrafine Grained Materials, TMS, Warrendale, PA (2000). 5. Gleiter, H., Acta Mater., 48:1 (2000). 6. Suryanarayana, C., and Koch, C.C., Hyperfine Interactions, 130:5 (2000). 7. Koch, C.C. (ed.), Nanostructured Materials: Processing, Properties and Applications, William Andrew Publishing, New York, NY (2002). 8. Suryanarayana, C., JOM., 54(9):27 (2002). 9. Han, B.Q., and Lavernia, E.J., Adv. Eng. Mater., 7:453 (2005). 10. Kelsall, R.W., Hamley, I.W., and Geoghegan, M. (eds.), Nanoscale Science and Technology, p. 237, Wiley, New York, NY (2005). 11. Suryanarayana, C., Adv. Eng. Mater., 7:983 (2005). 12. Suryanarayana, C., and Froes, F.H., Metall. Trans., A23:1071 (1992). 13. Suryanarayana, C., Prog. Mater. Sci., 46:1 (2001).

2: Synthesis, Suryanarayana and Prabhu

85

14. Suryanarayana, C., Mechanical Alloying and Milling, Marcel Dekker, New York, NY (2004). 15. Segal, V.M., Mater. Sci. Eng., A197:157 (1995). 16. Valiev, R.Z., Islamgaliev, R.K., and Alexndrov, I.V., Prog. Mater. Sci., 45:103 (2000). 17. Smirnova, N.A., Levit, V.I., Pilyugin, V.I., Kuznetsov, R.I., Davydova, L.S., and Sazonova, V.A., Fiz. Met. Metalloved., 68:1170 (1986). 18. Saito, Y., Tsuji, N., Utsunomiya, H., Sakai, T., and Hong, R.G., Scripta Mater., 39:1221 (1998). 19. Granqvist, C., Kish, L., and Marlow, W. (eds.), Gas Phase Nanoparticle Synthesis, Kluwer Academic, Dordrecht, The Netherlands (2004). 20. Karimpoor, A.A., Erb, U., Aust, K.T., and Palumbo, G., Scripta Mater., 49:651 (2003). 21. Nalwa, H.S. (ed.), Handbook of Nanostructured Materials and Nanotechnology, Vol. I: Synthesis and Processing, Academic Press, San Diego, CA (2000). 22. Blander, M., and Katz, J.L., Geochim. Cosmochim. Acta, 31:1025 (1967). 23. Grossman, L., Geochim. Cosmochim. Acta, 36:597 (1972). 24. Pfund, A.H., Phys. Rev., 35:1434 (1930). 25. Burger, H.C., and Van Cittert, P. H., Z. Physik, 66:210 (1930). 26. Harris, L., Jeffries, D., and Siegel, B.M., J. Appl. Phys., 19:791 (1948). 27. Granqvist, C.G., and Buhrman, R.A., J. Appl. Phys., 47:2200 (1976). 28. Kimoto, K., Kamiya, Y., Nanoyama, M., and Uyeda, R., Jpn. J. Appl. Phys., 2:702 (1963). 29. Wada, N., Jpn. J. Appl. Phys., 6:553 (1967). 30. Yatsuya, S., Kasukabe, S., and Uyeda, R., Jpn. J. Appl. Phys., 12:1675 (1973). 31. Uyeda, R., J. Cryst. Growth, 45:485 (1978). 32. Uyeda, R., Prog. Mater. Sci., 35:1 (1991). 33. Thölén, A.R., in: Nanophase Materials: Synthesis-Properties-Applications, (G.C. Hadjipanayis and R.W. Siegel, eds.), p. 57, Kluwer Academic, Dordrecht, The Netherlands (1994). 34. Gleiter, H., in: Deformation of Polycrystals: Mechanisms and Microstructures, (N. Hansen, A. Horsewell, T. Leffers, and H. Lilholt, eds.), p. 15, Risø National Laboratory, Roskilde Denmark (1981). 35. Gleiter, H., and Marquardt, P., Z. Metallkde., 75:263 (1984). 36. Siegel, R.W., Ramaswamy, S., Hahn, H., Zongquan, L., Ting, L., and Gronsky, R., J. Mater. Res., 3:1367 (1988). 37. Kaito, C., Fujita, K., Shibahara, H., and Shiojiri, M., Jpn. J. Appl. Phys., 16:697 (1977). 38. Yukawa, N., Hida, M., Imura, T., Kawamura, M., and Mizuno, Y., Metall. Trans., 3:887 (1972). 39. Kashu, S., Fuchita, E., Manabe, T., and Hayashi, C., Jpn. J. Appl. Phys., 23(L):910 (1984). 40. Chaudhari, P., and Matthews, J.W., Appl. Phys. Lett., 17:115 (1970). 41. Kaito, C., Jpn. J. Appl. Phys., 17:601 (1978). 42. Gleiter, H., Adv. Mater., 4:474 (1992). 43. Seigel, R.W., and Eastman, J.A., in: Multicomponent Ultrafine Microstructures, (L.E. McCandlish, D.E. Polk, R.W. Siegel, and B.H. Kear, eds.), 132:p. 3, Mater. Res. Soc., Pittsburgh, PA (1989).

86

Processing

44. Thölén, A.R., Acta Metall., 27:1765 (1979). 45. Eastman, J.A., Liao, Y.X., Narayanasamy, A., and Siegel, R.W., in Mater. Res. Soc. Symp. Proc., 155:255 (1989). 46. El-Shall, M.E., and Edelstein, A.S., in: Nanomaterials: Synthesis, Properties and Applications, (A.S. Edelstein, and R.C. Cammarata, eds.), p. 13, Institute of Physics Publishing, Bristol, UK (1996). 47. Leung, A., Martian, P., Crowe, C.T., Chung, J.N., and Hoagland, R.G., in: Nanophases and Nanocrystalline Structures (R.D. Shull, and J.M. Sanchez, eds.), p. 1, TMS, Warrendale, PA (1993). 48. Andres, R.P., Bowles, R.S., Kolstad, J.J., and Calo, J.M., Surface Sci., 106:117 (1981). 49. Milani, P., and Ianotta, S., in: Cluster Beam Synthesis of Nanostructured Materials, Springer-Verlag, Berlin (1999). 50. Bulgin, D., Dyke, J., Goodfellow, J., Jonathan, N., Lee, E., and Morris, A.J., Electron Spectrosc., 12:67 (1977). 51. Tsantilis, S., Pratsinis, S.E., and Haas, V., J. Aerosol Sci., 30:785 (1999). 52. Gonsalves, K.E., Li, H., Perez, R., Santiago, P., and Yacaman, M.J., Coord. Chem. Rev. 206:607 (2000). 53. Vissokov, G.P., Stefanov, B.I., Gerasimov, N.T., Oliver, D.H., Enikov, R.Z., Vrantchev, A.I., Balabanova, E.G., and Pirgov, P.S., J. Mater. Sci., 23:2415 (1988). 54. Ohno, S., and Uda, M., J. Jpn. Inst. Metals, 53:946 (1989). 55. Choi, C.J., Dong, X.L., and Kim, B.K., Scripta Mater., 44:2225 (2001). 56. Chen, Y., Glumac, N., Kear, B.H., and Skandan, G., Nanostruct. Mater., 9:101 (1997). 57. Chang, W., Skandan, G., Danforth, S. C., Rose, M., Balogh, A. G., Hahn, H., and Kear, B., Nanostruct. Mater., 6:321 (1995). 58. Chow, G.M., Klemens, P.G., and Strutt, P.R., J. Appl. Phys., 66:3304 (1989). 59. Hahn, H., and Averback, R.S., Appl. Phys. Lett., 67:1113 (1990). 60. Powers, D.E., Hansen, S.G., Guesic, M.E., Pulu, A.C., Hopkins, J.B., Dietz, J., Duncan, M.A., Smith, P.R.R., and Smiley, R.E., J. Phys. Chem., 86:2556 (1982). 61. Kato, M., Jpn. J. Appl. Phys., 15:757 (1976). 62. Yamamoto, T., and Mazumder, J., Nanostruct. Mater., 7:305 (1996). 63. Johnston, G.P., Muenchausen, R., Smith, D.M., Fahrenheit, W., and Foltyn, S., J. Amer. Ceram. Soc., 75:3293 (1992). 64. Shinde, S.R., Kulkarni, S.D., Banpurkar, A.G., Dixit, R.N., Date, S.K., and Ogale, S.B., J. Appl Phys., 88:1566 (2000). 65. Harano, A., Shimada, K., Okubo, T., and Sadakata, M., J. Nanoparticle Res., 4:215 (2002). 66. Makimura, T., Mizuta, T., and Murakami, K., Jpn. J. Appl. Phys., 41(L):144 (2002). 67. Fayet, P., Patthey, F., Roy, H.V., Detzel, T.H., and Schneider, W.D., Surface Sci., 269/270:1101 (1992). 68. Iwama, S., Hayakawa, K., and Arizumi, T., J. Cryst. Growth, 56:265 (1982). 69. Iwama, S., Hayakawa, K., and Arizumi, T., J. Cryst. Growth, 66:189 (1984). 70. Günther, B., and Kumpmann, A., Nanostruct. Mater., 1:27 (1992). 71. Haberland, H., Karrais, M., Mall, M., and Thurner, Y., J. Vac. Sci. Technol., A10:3266 (1992).

2: Synthesis, Suryanarayana and Prabhu

87

72. Suh, T.G., Umarjee, D.M., Prakash, S., Doerr, H.J., Deshpandey, C.V., and Bunshah, R.F., Surface Coat. Technol., 49:304 (1991). 73. Terauchi, S., Koshizaki, N., and Umehara, H., Nanostruct. Mater., 5:71 (1995). 74. Siekmann, H.R., Luder, C.H., Faehrmann, J., Lutz, H.O., and Broer, K.H., Z. Physik, D20:417 (1991). 75. Wada, N., Jpn. J. Appl. Phys., 8:551 (1969). 76. Young, R.M., and Pfender, E., Plasma Chem. Plasma Process, 5:1 (1985). 77. Heberlein, J., Postel, O., and Girshick, S.L., Surface Coat. Technol., 142–144:265 (2001). 78. Hihara, T., Yamada, Y., Katoh, M., Peng, D.L., and Sumiyama, K., J. Appl. Phys., 94:7594 (2003). 79. Yoshida, T., Endo, H., and Akashi, K., in: The Preparation of Ultrafine Si3N4 Particles in a Hybrid Plasma, p. 225, ISPC-6, Montreal, Canada. (1983). 80. Parker, J.C., and Seigel, R.W., J. Mater. Res., 5:1246 (1990). 81. Parker, J.C., and Siegel, R.W., Appl. Phys. Lett., 57:943 (1990). 82. Andres, R.P., Bielefeld, J.D., Henderson, J.I., Janes, D.B., Kolagunta, V.R., Kubiak, C.P., Mahoney, W.J., and Osifchin, R.G., Science, 273:1690 (1996). 83. Siegel, R.W., Ann. Rev. Mater. Sci., 21:559 (1991). 84. Haas, V., Gleiter, H., and Birringer, R., Scripta Metall. Mater., 28:721 (1993). 85. Randolph, A.D., and Larson, M.A., Theory of Particulate Processes, 2nd ed., Academic Press, New York, NY (1988). 86. Hinds, W.C., Aerosol Technology Properties: Behavior and Measurement of Airborne Particles, Wiley, New York, NY (1982). 87. Abraham, F.F., J. Chem. Phys., 51:1632 (1969). 88. Abraham, F.F., Kim, S.S., and Stein, G.D., J. Chem. Phys., 75:402 (1981). 89. Beylich, A.E., and Robben, F., Z. Angew. (Math). Phys., 25:443 (1974). 90. Flagan, M.M., and Lunden, R.C., Mater. Sci. Eng., A204:113 (1995). 91. Nakaso, K., Shimada, M., Okuyama, K., and Deppert, K., J. Aerosol Sci., 33:1061 (2002). 92. Haubold, T., Acta Met., 41:1769 (1993). 93. Kappes, M., Kunz, R.W., and Schumacker, E., Chem. Phys. Lett., 91:413 (1982). 94. Kappes, M., and Leutwyler, S., Atomic and Molecular Beam Methods, (G. Scoles, ed.), chapter 15, Oxford University Press, New York (1988). 95. Olynick, D.L., Gibson, J.M., and Averback, R.S., Phil. Mag., A77:1205 (1998). 96. Efendiev, Y., and Zachariah, M.R., J. Colloid Interface Sci., 249:30 (2002). 97. Maisels, A., Kruis, F.E., and Fissan, H., J. Aerosol Sci., 33:35 (2002). 98. Pearson, H. J., J. Fluid Mech., 143:367 (1984). 99. Gutsch, A., Pratsinis, S.E., and Löffler, F., J. Aerosol Sci., 26:187 (1995). 100. Krauss, W., and Birringer, R., Nanostruct. Mater., 9:109 (1997). 101. Saito, Y., Mihama, K., and Uyeda, R., J. Appl. Phys., 19:1603 (1980). 102. Stappert, S., Rellinghaus, B., Acet, M., and Wassermann, F., J. Cryst. Growth, 252:440 (2003). 103. Bose, C., Thangadurai, P., Ramasamy, S., and Purniah, B., Vacuum, 77:293 (2005).

88

Processing

104. Michels, D., Krill, C.E., and Birringer, R., J. Magn. Magn. Mater., 250:203 (2002). 105. Wang, H., Zhang, J., and Hsu, T.Y., Mater. Sci. Eng. A380:408 (2004). 106. Dong, X.L., Choi, C. J., and Kim, B.K., Scripta Mater., 47:857 (2002). 107. Kaito, C., Fujita, K., and Shiojiri, M., J. Appl. Phys., 47:5161 (1976). 108. Yeadon, M., Yang, J.C., Ghaly, M., Averback, R.S., and Gibson, J.M., Mater. Sci. Eng., B67:76 (1999). 109. Türker, M., Mater. Sci. Eng., A367:74 (2004). 110. Lee, K.M., Park, S.T., and Lee, D.J., J. Alloys and Compounds, 390:297 (2005). 111. Wegner, K., Walker, B., Tsantilis, S., and Pratsinis, S. E., Chem. Eng. Sci., 57:1753 (2002). 112. Yeadon, M., Zimmermann, C. G., Averback, R. S., and Gibson, J. M., J. Electron Microsc., 50:545 (2002). 113. Yamamuro, S., Sumiyama, K., and Suzuki, K., J. Appl. Phys., 85:483 (1999). 114. Ungar, T., Ott, S., Sanders, P.G., Borbely, A., and Weertman, J.R., Acta. Mater., 46:3693 (1998). 115. Fougere, G.E., Riester, L., Ferber, M., Weertman, J.R., and Siegel, R.W., Mater. Sci. Eng., A204:1 (1995). 116. Sasaki, Y., Hyakkai, M., Kita, E., Tasaki, A., Tanimoto, H., and Iwamoto, Y., J. Appl. Phys., 81:4736 (1997). 117. Oku, T., Nakayama, T., Kuno, M., Nozue, Y., Wallenberg, L.R., Niihara. K., and Suganuma, K., Mater. Sci. Eng., B74:242 (2000). 118. Friedrichs, O., Sánchez-López, J.C., López-Cartes, C., Dornheim, M., Klassen, T., Bormann, R., and Fernández, A., Appl. Surface Sci., 252:2334 (2006). 119. Dimesso, L., Heider, L., and Hahn, H., Solid State Ionics, 123:39 (1999). 120. Löffler, J.F., Braun, H.B., Wagner, W., Kostorz, G., and Wiedenmann, A., Mater. Sci. Eng., A304–306:1050 (2001). 121. Li, D.X., Ping, D.H., Huang, Y., Yu, Y.D., and Ye, H.Q., Micron, 31:581 (2000). 122. Raviprasad, G. V., Patel, K., Chakravarty, S., Mythili, R., Vijayalakshmi, M., Sekhar, B. R., Sahu, S. N., Ramamurthy, V. S., and Behera, S. N., Current Sci., 80:280 (2001). 123. Huh, M.Y., Kim, S.H., Ahn, J.P., Park, J.K., and Kim, B.K., Nanostruct. Mater., 11:211 (1999). 124. Chen, P.P., Wang, Z.Y., Yu, S.W., Hong, J.M., Poulsen, P.R., Ji, Y.L., Miao, B.Y., Han, M., and Wang, G.H., Phys. Lett., A244:407 (1998). 125. Wolf, H., Zimmer, H.G., Filz, T., Lauer, S., Wichert, T., and Kraub, W., Mater. Sci. Eng., A204:43 (1995). 126. Sanchez, J.C., Fernandez, A., Conde, C.F., Conde, A., Morant, C., and Sanchez, J.M., Nanostruct. Mater., 7:813 (1996). 127. Karch, J., Birringer, R., and Gleiter, H., Nature, 330:556 (1987). 128. Fernandez, A., Reddy, E. P., Rojas, T.C., and Sanchez, J.C., Vacuum, 52:83 (1999). 129. Chiang, Y.M., Lavik, E.B., Kosacki, I., Tuller, H.L., and Ying, J.Y., Appl. Phys. Lett., 69:185 (1996). 130. Jigatch, A.N., Leipunskii, I.O., Kuskov, M.L., Stoenko, N.I., and Storozhev, V.B., Instrum. Exp. Techniques, 43:839 (2000).

2: Synthesis, Suryanarayana and Prabhu

89

131. Bonetti, E., Savini, L., Deriu, A., Albanese, G., and Moya, J., J. Magn. Magn. Mater., 262:132 (2003). 132. Fung, K.K., Qin, B., and Zhang, X.X., Mater. Sci. Eng., A286:135 (2000). 133. Thangadurai, P., Ramasamy, S., and Kesavamoorthy, R., J. Phys. C: Condens. Matter, 17:863 (2005). 134. Betz, U., Sturm, A., Löffler, J.F., Wagner, W., Wiedenmann, A., and Hahn, H., Mater. Sci. Eng., A281:68 (2000). 135. Sanchez, J.C., and Fernandez, A., Thin Solid Films, 317:497 (1998). 136. Nagashima, M., Nakayama, T., Yamanaka, S., Fujikane, M., Hayashi, Y., Sekino, T., Kusunose, T., and Niihara, K., Mater. Lett., 57:4023 (2003). 137. Chang, H., Averback, R.S., and Altstetter, C.J., in: Nanophases and Nanocrystalline Structures, (R.D. Shull, and J.M. Sanchez, eds.), p. 41, TMS, Warrendale, PA (1993). 138. Peng, D. L., Hihara, T., and Sumiyama, K., J. Magn. Magn. Mater., 277:201 (2004). 139. Chang, H., Höfler, J., Altstetter, C., and Averback, R.S., Mater. Sci. Eng., A153:676 (1992). 140. Rittner, M. N., Weertman, J.R., Eastman, J.A., Yoder, K.B., and Stone, D.S., Mater. Sci. Eng., A237:185 (1997). 141. Nakayama, T., Yamamoto, T.A., Choa, Y.H., and Niihara, K., J. Mater. Sci., 35:3857 (2000). 142. Skårman, B., Grandjean, D., Benfield, R.E., Hinz, A., Andersson, A., and Wallenberg, L.R., J. Catalysis., 211:119 (2002). 143. Baker, C., Shah, S., and Hasanain, S.K., J. Magn. Magn. Mater., 280:412 (2004). 144. Harris, L., Jeffries, D., and Siegel, B.M., J. Chem. Phys., 18:261 (1950). 145. Eastman, J.A., in Multicomponent Ultrafine Microstructures, (L.E. McCandlish, D.E. Polk, R.W. Siegel, and B.H. Kear, eds.), 132:p. 27, Mater. Res. Soc., Pittsburgh, PA (1989). 146. Chandra, R., Taneja, P., John, J., Ayyub, P., Dey, G.K., and Kulshreshtha, S.K., Nanostruct. Mater., 11:1171 (1999). 147. Brock, J.R., Kuhn, P.J., and Zehavi, D., J. Aerosol Sci., 6–17:11 (1986). 148. Yatsuya, S., Kasukabe, S., and Uyeda, R., J. Cryst. Growth, 24/25:319 (1974). 149. Huh, M.-Y., Kim, J.-S., and Ahn, J.-P., Powder Metall., 45:154 (2002). 150. Baker, C., Pradhan, A., Pakstis, L., Pochan, J., and Shah, S., J. Nanosci. Nanotech., 5:244 (2005). 151. Baker, C., Hasanain, S.K., and Shah, S., J. Appl. Phys., 96:6657 (2004). 152. Birringer, R., Herr, U., and Gleiter, H., Trans. Jpn. Inst. Met. Suppl., 27:43 (1986). 153. Sanders, P.G., Fougere, G.E., Thompson, L.J., Eastman, J.A., and Weertman, J.R., Nanostruct. Mater., 8:243 (1997). 154. Hahn, H., Eastman, J.A., and Siegel, R.W., Ceramic Trans., 1(B):1115 (1988). 155. Baker, C., Shah, S., Hasanain, S.K., Ali, B., Shah, L., Ekiert, T., and Unruh, K.M., in: Mater. Res. Soc. Symp. Proc., 746:201 (2003). 156. Chang, W., Skandan, G., Danforth, S.C., Kear, B.H., and Hahn, H., Nanostruct. Mater., 4:507 (1994).

90

Processing

157. Oda, M., Fuchita, E., Tsuneizumi, M., Kashu, S., and Hayashi, C., Nanostruct. Mater., 1:203 (1992). 158. Jensen, P., Melinon, P., Treilleux, M., Hoareau, A., Hu, J.X., and Cabaud, B., Appl. Phys. Lett., 59:142 (1991). 159. Abe, H., Schulze, W., and Tesche, B., Chem. Phys., 47:95 (1980). 160. Bowles, R.S., Kolstad, J.J., Calo, J.M., and Andres, R.P., Surface Sci., 106:117 (1981). 161. Wagener, M., and Günther, B.J., J. Magn. Magn. Mater., 201:41 (1999). 162. Konrad, H., Haubold, T., Birringer, R., and Gleiter, H., Nanostruct. Mater., 7:605 (1996). 163. Yamamoto, T.A., Shull, R.D., and Hahn, H., Nanostruct. Mater., 9:539 (1997). 164. Stoyanov, S., Skumryev, V., and Zhang, Y., J. Appl. Phys., 93:7592 (2003). 165. Maisels, A., Kruis, F.E., Fissan, H., Rellinghaus, B., and Zahres, H., Appl. Phys. Lett., 77:4431 (2000). 166. Ohno, T., J. Nanoparticle Res., 4:255 (2002). 167. Perez, A., Bardotti, L., Prevel, B., Jensen, P., Treilleux, M., Mélinon, P., Gierak, J., Faini, G., and Mailly, D., New J. Phys., 4:76 (2002). 168. Goldby, M., Kuipers, L., Issendorf, B., and Palmer, R.E., Appl. Phys. Lett., 69:2819 (1996). 169. Hai, N.H., Lemoine, R., Remboldt, S., Strand, M., Shield, J.E., Schmitter, D., Kraus, Jr., R.H., Espy, M., and Leslie-Pelecky, D.L., J. Magn. Magn. Mater., 293:75 (2005). 170. Laliberte, M., and Boisjoli, A., Chaos, Solitons, and Fractals, 22:935 (2004).

3 Thermal Sprayed Nanostructured Coatings: Applications and Developments George E. Kim Perpetual Technologies, Inc., Ile des Soeurs, Quebec, Canada

3.1 Introduction This chapter provides an overview of the development and application of nanostructured oxide coatings. More specifically, the progression from their inception to their successful application onto challenging military and industrial service is described. Between 1997 and 2002, Dr. Lawrence T. Kabacoff (Program Officer) at the United States Office of Naval Research (ONR) headed the first effort on thermal spraying of nanostructured coatings through a program entitled, “Thermal Spray Processing of Nanostructured Coatings.”[1] The work was based on the notion that properties of existing materials change drastically when physical features (i.e., grain size, fiber diameter, layer thickness, particle diameter) of a material are reduced to and kept below 100 nm. ONR’s objective was to save the Navy money by extending the service life of ship components by incorporating any enhanced properties of nanostructured materials in coating form. Since the inception of the ONR program, there have been numerous successes in the use of nanostructured coatings for military and industrial applications. Among the numerous successes derived from the program, the work carried out by Gell et al.[2,3,4] at the University of Connecticut (UCONN) on nanostructured alumina–titania coating has stood out. The group at UCONN developed a method of applying a commonly used wear-resistant coating material, alumina–titania, in nanostructured form. The new n-Al2O3—13TiO2 coating possessed far superior properties over its microstructured commercial counterpart of the same composition. Some of the properties included: Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 91–118 © 2007 William Andrew, Inc.

91

Processing

92 • Enhanced bond strength (up to 2×) • Superior wear resistance (2–4×) • Remarkable toughness

An important aspect of the ONR approach that contributed to its widespread use within a short period by military, research, and industrial communities relates to its mandate of using off-the-shelf thermal spray equipment to pursue its objectives. The success of the program has led to direct cost savings of millions of dollars by extending the service life of Navy machinery and other assets. The nanostructured alumina–titania feedstock powder, qualified for use on U.S. Navy ships, is commercially available through Inframat Corporation. In May of 2000, three industrial companies, Perpetual Technologies, Mogas Industries, and FW Gartner Thermal Spraying Company, combined their efforts to incorporate the knowledge gained from the results of ONR’s program toward a severe-service industrial application associated with the extraction of nickel and cobalt from low-grade ore. This effort set several new milestones at the time, including: • Development and application of a patented TiO2-based nanostructured coating • Qualification and application of a nanostructured coating for an industrial application • Further enhancement in properties of nanostructured ceramic coating using numerous types of thermal spray processes • Evaluation and identification of unique erosion-resistant characteristics of nanostructured ceramic coatings

3.2 Thermal Spray Technology Thermal spray technology involves the projection of molten or semimolten particles of metals, ceramics, or their composites from powder or wire feedstock (Fig. 3.1).[5] Generally, any material that has a stable molten phase and can be processed into the appropriate feed specifications can be thermal sprayed. The melting is achieved chemically via oxygen-fuel combustion or electrically via an arc. The hot particles are accelerated by the combustion flame or the plasma jet onto a surface, forming a lamellar structure. Multiple passes result in a build-up of lamellae to the desired thickness, usually beyond 50 μm. A typical thermal-sprayed single-component coating consists of very fine grains, having properties associated with such a microstructure, as well as

3: Applications and Developments, Kim

93

nonhomogeneous features such as splat boundaries, pores, oxide inclusions, and unmelted particles. Although these features may not provide a picture of a clean and reliable structure, when applied correctly as a coating they often lead to substantial enhancements in protection against wear. The nonhomogeneous structure, however, makes as-sprayed thermal-sprayed coatings less ideal for nonsacrificial corrosion applications. In most cases, nonsacrificial coatings will require some form of sealant to inhibit penetration of liquid or gas to the substrate.

3.2.1 Types of Processes Thermal spray processes are differentiated from one another through the type of feedstock (e.g., powder or wire) and the type of heat source (e.g., combustion or electrical) used. Among the numerous types, two major process groups used to deposit protective coatings onto ball valves include combustion and plasma spraying, primarily for carbide metal-matrix composite and ceramic coatings, respectively. Combustion spraying utilizes compressed air or oxygen with one of a variety of fuels (e.g., acetylene, propane, hydrogen, natural gas, propylene, and methylacetylene–propadiene) to melt and propel the particles. With the exception of high-velocity oxyfuel, most combustion spray processes provide rather low flame velocities (about 50 m/s) and temperatures (below 3000°C) which result in a porous coating with limited bond strength. The newer combustion spray process of high-velocity oxyfuel (HVOF), has made a significant impact in the field of wear protection. While fuel types used are similar to those in conventional combustion spray processes, the HVOF process uses a torch design with a nozzle that allows for dramatic gas acceleration. The gas velocities have been measured in Figure 3.1 Schematic of the thermal spray process.

94

Processing

the range 1500 to 2000 m/s while maintaining a gas temperature in the order of 2900°C. Although the limited temperature range excludes this process from depositing monolithic ceramics, metals and their composites can be densely applied with good bond strength. Carbide composites such as WC-Co and Cr3C2-NiCr are typically deposited using the HVOF process since the gas temperature is sufficiently low to inhibit decarburization of the carbide phases. Figure 3.2[6] shows a working HVOF torch. Plasma spraying makes use of a gun, which uses a DC arc struck between a tungsten alloy cathode and an annular copper anode to heat a flowing inert gas stream (usually argon with a small percentage of enthalpy-enhancing gas such as hydrogen), forming a plasma jet. The gas temperature just outside the gun exit or the core of the plasma jet is approximately 17 000°C.[7] The average particle velocity for plasma spray processes (300–400 m/s) is approximately half that of HVOF particle velocities and can vary depending on the power capacity of the plasma gun. Due to the high gas temperatures, this process can be used to apply metals, ceramics, and their composites. Most plasma spray applications are carried out in atmospheric conditions (Fig. 3.3).[8] The vacuum or low-pressure plasma spray (VPS or LPPS) process applies the coating within a reduced pressure inert-gas chamber. The advantages of this process are mainly evident in coatings of metals, where the inert atmosphere inhibits oxidation, reduces porosity, and provides

Figure 3.2 High-velocity oxyfuel process.

3: Applications and Developments, Kim

95

Figure 3.3 Atmospheric plasma spray process.

superior bond strengths. The VPS process, however, has higher associated maintenance and production costs. Size limitations related to the dimensions of the chamber must also be considered.

3.2.2 Applications Thermal spray coatings are widely used in applications that require wear protection, corrosion protection, a thermal barrier, dimension control, and component restoration, to name a few examples. Industries such as aerospace, biomedical, chemical processing, mining, electrical utilities, automotive, pulp and paper, among others, rely on thermal spray coatings to enhance product performance, to extend component life, and/or to reduce maintenance costs. Wear and corrosion account for a large part of thermal spray applications and are key issues for ball valves used in the nickel high-pressure acid leach (NiHPAL) process. Coating materials such as oxides, carbide composites, and hard metal alloys are extensively used to resist abrasive, erosive, and adhesive wear. Since several factors including intrinsic surface properties (hardness, ductility, strength, etc.), surface finish, lubrication, temperature, and corrosion contribute to the wear resistance of a component, each application should be considered in detail prior to selecting a coating material and thermal spray process. Chromia, alumina–titania, tungsten carbide–cobalt, chrome carbide–nickel chrome, stellite, and molybdenum are some of the

96

Processing

commonly thermal-sprayed coating materials used extensively to resist wear. The nonhomogeneous features of as-sprayed thermal spray coatings, with the exception of spray and fused coatings, has limited most of their corrosion applications to those where the coating provides a sacrificial protection. Zinc and aluminum, as well as their alloys, are used to provide protection in this manner for infrastructures, ships, and other large surfaces. In cases where a more noble coating material is applied, it is important to address the issue of through-porosity, which inhibits the coating’s functioning as a physical barrier. In such instances, one may consider organic/inorganic sealants, heat treatment, or additional layer(s) to provide the barrier to the substrate. It is imperative that all the microcracks and through-pores that provide penetrable access for the corrodent be sealed; any unsealed coating anomalies would reduce the surface area of the anodic substrate and lead to a greater localized corrosion rate (e.g., pitting or crevice corrosion). In addition to these concerns, there may also be selective attack of the coating intersplat zones, grain boundaries, or anodic phases in the coating. Therefore, in order to qualify a coating for this service, one must carry out detailed chemical and physical characterization of the coatings prior to putting it into a corrosive service environment.

3.2.3 Thermal Spray Processes As mentioned in section 3.2.1, a typical as-sprayed thermal spray coating consists of a nonhomogeneous, lamellar structure with numerous other physical features (pores, unmelted particles, etc.). Depending on the coating material and the thermal spray process used, the structure of the coating, as well as the detailed features, vary in shape, size, and/or number. For metallic coatings, cooling rates of approximately 106°C/s lead to very rapid solidification, resulting in a metastable structure with fine grain sizes. Hence, thermal sprayed coatings possess properties that differ from those of bulk materials, e.g., higher hardness and lower ductility. Depending on the requirements of the application, thermal spray coatings may be deposited in various structural forms. Figure 3.4 shows micrographs of several coatings. These include single-layer monolithic, multilayered, composite, and functionally graded coatings.

3: Applications and Developments, Kim

a

b

c

d

97

Figure 3.4 Micrographs of various thermal spray structures.

Figure 3.4(a) illustrates the simplest thermal-sprayed structure, consisting of a single-layer monolithic coating. Many of the pure and alloyed metal coatings are applied in this form. For monolithic ceramic coatings, it is quite common to incorporate a metallic bond coat layer between the substrate and the ceramic top coat (Fig. 3.4(b)). The bond coat generally serves two main functions: (i) it can reduce the residual stress level at the interface region by providing an intermediate coefficient of thermal expansion (CTE) between the metal and the ceramic layers, and/or (ii) it can provide a barrier against corrosive liquids or gases. Composite coatings such as carbide cermets (e.g., WC—Co and Cr3C2—NiCr) provide the wear resistance of hard carbide particles with the toughness of the metallic matrix in a single layer (Fig. 3.4(c)). These coatings have been used extensively in many of the applications where resistance to sliding and erosive wear are required. The carbide phases also provide a low coefficient of friction. The final structure is referred to

98

Processing

as a functionally graded coating, which gradually moves from one monolithic layer to another by increasing the ratio of the second phase material (Fig. 3.4(d)). The graded layers provide a gradual change in the CTE and hence reduce the residual stress at the interphase (the region around the coating–substrate interface). It is important to note that the interlayers between the two monolithic layers may be subjected to increased deterioration (corrosion or oxidation) due to higher surface area exposure if one or more phases is prone to degradation.

3.3 Thermal-Sprayed Nanostructured Alumina–Titania Coating and United States Navy Applications 3.3.1 Nanostructured Alumina–Titania Coating As mentioned in the introduction, one of the successes of the ONR program on “Thermal Spray Processing of Nanostructured Coatings” originated from the work carried out at the University of Connecticut by Gell et al.[2,3,4]. Their general approach consisted of mixing nanoparticles of alumina and titania (87 : 13 wt% ratio) in a slurry and spray-drying this slurry to form spherical composite agglomerates in the size range of 15–150 μm with an average particle size of 50 μm. This powder was further thermally processed to remove the binder and to instill robustness to the agglomerates. A “modified” version of the nanostructured agglomerate powder was also developed and studied. This version consisted of a small amount of additional oxides dispersed throughout the spray-dried agglomerate and a final plasma densification process. The unmodified agglomerate powder had relatively high porosity, up to 45 %, whereas the modified version was denser and consisted of a finer microstructure. A conventional plasma spray torch, Metco 9MB was used to deposit coatings with thicknesses of several hundred micrometers. Gell et al. have published numerous findings[2,3,4] relating to the analyses of the different oxide phases found in these nanostructured alumina– titania feedstock powders, as well as the relationship of microstructure and coating properties to an empirical parameter that they have defined as the critical plasma spray parameter (CPSP). There is good information to be found in these detailed results. Some of the key property enhancements found in the coatings derived from the nanostructured alumina–titania coatings, especially the modified version, were as follows:

3: Applications and Developments, Kim

99

• Increased adhesion strength • Spallation resistance in bend and cup-tests • Resistance to abrasive and sliding wear. These properties are universally pertinent to the performance of coatings for most applications. Gell et al. attributed these remarkable properties to the bimodal microstructure of the coatings deposited from the nanostructured alumina–titania agglomerate feedstock. The bimodal microstructure has been identified as regions of fully melted and splat-quenched structure adjacent to regions of partially melted structure where some microstructural features of the original particles were retained. It was also observed that the cracks formed during indentation tests traveled predominantly along the splat boundaries for coatings derived from both conventional and nanostructured alumina–titania feedstock. The difference, however, was that the cracks propagating through the splat boundaries in the nanostructured coating would be arrested and/or deflected when they encountered a partially melted region of the bimodal microstructure. It was this crack resistant or toughness feature that seemed to provide the enhanced wear resistance, especially with regard to abrasive wear, even with the nanostructured coating’s lower microhardness.[2] Although the abrasive wear resistance is often superior for harder materials/coatings, the superior toughness of the nanostructured alumina–titania coating seemed to more than compensate for the reduced strength or hardness.

3.3.2 United States Navy Applications Based on specific empirical data generated from the US Navy’s investigation of nanostructured materials for thermal spray applications through the ONR (Kabacoff), there are several in-service applications currently under review to validate the benefit derived from the unique properties of employing nanomaterials. A very good example of a novel application that has clearly benefited from the unique qualities of the nanostructure alumina–titania coating is on a critical component of a minesweeper. This challenging application involves depositing a protective coating onto the main propulsion shafting of a minesweeper. Figure 3.5 shows a dry-docked minesweeper and a view of the main propulsion shaft.[9] The Ni—Al bronze shafts suffered from excessive wear in the stern tube and strut bearing areas. The consequence of this issue was frequent and costly repair, not to mention detriment to operational availability. This wear was

100

Processing

Figure 3.5 Avenger Class Mine Countermeasures Ship-USS Patriot (MCM-7) (top) and view of the main propulsion shaft (bottom).

attributed to bio-growth, originating between the bearing staves that scored the shafting when the shafting made turns prior to the water bearing being established. The nanostructured alumina–titania coating was deposited on the shaft, extending 3 inches either side of the bearing area. It is important to note that a conventional brittle ceramic coating would never have been considered for this application, since the coating and its substrate are subjected to significant torque, bending, and fatigue levels. Figure 3.6[9] provides images of the worn shaft and the steps taken in its repair via the thermal spraying of the nanostructured alumina–titania coating.

3: Applications and Developments, Kim

101

Figure 3.6 Worn shaft (top) and thermal spray repair incorporating nanostructured alumina-titania coating (bottom).

After 3 months of operation, a preliminary inspection indicated that the nanostructured coating showed no visual signs of wear. After 4 years in service, an underwater inspection revealed no signs of scoring, spalling, or degradation of the coating. Although the underwater inspection is not as definitive in determining the state of the coating as removing the shafting and conducting a more comprehensive inspection, it does provide a strong indication that the problem has been mitigated. This, in essence, has extended the life cycle of the shafting, thereby reducing the life-cycle maintenance costs. Based on the success of this coating performance, the Navy has plans to apply the same nanostructured coating onto other mine countermeasures ships (MCMs). The estimated cost to repair a ship set of two shafts using the nanostructured coating is $75 000; this is very attractive when compared to the previous approach of Original Equipment Manufacturer (OEM) weld repair, which cost $280 000 per ship set. The estimated return on investment cost avoidance, based on an improvement of 4 years of service life versus 18 months, is $34 000 000 over the remaining life of

102

Processing

these components. Based on the latest inspection after 4 years service, it is apparent this service life, and hence cost avoidance, will be further improved upon.

3.4 Development and Application of Nanostructured Titania-Based Coating for Industrial Application The nickel/cobalt high pressure acid leach (Ni/Co HPAL) process utilizes vessels, piping, valves, and other equipment to contain and flow-control a severe service environment. These components suffer significant damage due to factors such as corrosion, erosion, and abrasion and the profitability of the operation turns on their durability and dependability.

3.4.1 Background The metal-seated ball valve’s angular rather than volumetric displacement capability; its large, low stress, precision-lapped seal; and its ability to continue to operate in an environment with crushed, abrasive solids; makes it the best choice for isolation in severe service application. Metalseated ball valves used in Ni/Co HPAL service incorporate a protective thermal sprayed ceramic top coat that is lap-finished to minimize wear and maximize valve sealing life in severe service applications. The severe service consists typically of operating at around 260°C in Ni/Co HPAL autoclaves, with process designers targeting further increase in the process temperature. These temperatures prevent the use of soft sealing materials such as Teflon or Peek. Additionally, the elevated temperatures magnify the corrosiveness of the lading. Temperature also exerts thermal stresses at the coating–base metal interface. Crushed solids transported through the system abrade and damage the precision sealing surfaces of the valve seats. Solids tend to pack into clearances within the valve internals and prevent proper float of the sealing members, resulting in seat leakage and rapid erosive wear. Corrosion due to the environment can attack and deteriorate precision sealing surfaces, which develop corrosion products at the valve coating–substrate interface, resulting in spalling of the coating. Corrosion products can also expand to fill tight clearances in seats, again preventing proper float, resulting in seat leakage and rapid erosive wear. Corrosion can cause deterioration of the coating hardness and wear resistance.

3: Applications and Developments, Kim

103

Thermal spray application of single-layer and multilayer coatings has been predominantly used for autoclave ball valves. These coatings fared well under the less severe conditions found in gold autoclaves; however, there had been more coatings issues in the more severe conditions of Ni/Co autoclaves. Prior to the introduction and acceptance of the nanostructured titania coating, coatings of chromia-blend and pure titania applied via plasma spray in atmosphere (APS) or vacuum (VPS) onto titanium balls and seats were specified by the mines. The single-layer approach was considered due to the corrosion-resistant nature of the titanium base material. The typical chromia-blend coating consisted of chromia with the addition of silica (5 wt%) and titania (3 wt%) to provide some toughness. It was well known that the chromia-blend coating provides superior abrasion and erosion wear resistance over the titania coating; however, due to the better CTE match between the coating and the substrate, the microstructure and bond strength for the titania coating on titanium components was anticipated to be theoretically superior to those of the chromia coating. Dual-layer coatings of a ceramic top coat (e.g., chromia or titania) and a metallic bond coat (e.g., titanium and tantalum) were also specified and applied for different HPAL autoclaves. The metallic bond coat is typically used to provide one or both of the features of a corrosion barrier to protect the substrate and/or enhanced bond strength between the bond and top coats. Since the base material is relatively inert to the slurry environment, the key attribute required from the bond coat was its ability to enhance bond strength. As mentioned by Kim et al.,[10] the most notable differences in quality between coatings applied via VPS and APS were found when spraying metals. The inert, reduced pressure environment of the VPS process allowed for the application of dense, oxide-free titanium and tantalum coatings. These superior features were considered critical for most instances where the metallic bond coat was relied upon as a corrosion barrier. However, coatings with VPS bond coats did not perform well in field service.

3.4.2 Failure Analyses In May of 2000, an extensive product analysis was carried out by Mogas, FW Gartner, and Perpetual Technologies to better understand the performance of thermal-sprayed coatings in different HPAL environments. The goal was to gather as much data as possible relating to the wear and corrosion performance of coatings from actual field applications.

104

Processing

At the time, this was the only reasonable approach to making an educated decision relating to any future development work. The following are examples of revealing data from different autoclave services.

3.4.2.1 Vacuum plasma spray (VPS) applied chromium oxide blend top coat and tantalum bond coat on titanium components: 6″ ID Macraes (gold HPAL autoclave) seat in service for 6 months; and, 3″ ID Ball slurry discharge valve in Canada Different regions of the failed components were sectioned and polished to view their cross-sections. See Fig. 3.7 for as-sprayed samples, as well as for the two used components. It is quite evident from the cross-sectional views that similar characteristics were observed between the Macraes and Canadian autoclave components. The most obvious difference between the as-sprayed and failed coatings is that the top coat of the failed component was very porous. This differed greatly when compared to the assprayed structure of the VPS top coat (Fig. 3.7, left) and could be attributed to structural change due to chemical attack. A close analysis of the remaining top coat of the failed component revealed several key features: • Bright regions of the top coat were remnants of the coating with low aspect ratios • Dark regions or voids were uniformly distributed throughout the coating cross-section Upon further review of the results and characteristics of the two plasma spray processes, it was determined that the low aspect ratio remnants (bright regions) were particles that had been only partially-molten or unmolten during flight. These partially-molten or unmolten particles are inherent to thermal spray technology and can be attributed to the different particle feedstock sizes and to their variation in thermal exposure within the plasma jet. The resemblances in size and shape between the bright regions of the top coat and that of the starting powder provided clear evidence of their relation. The dark regions or voids were fairly well distributed throughout the cross-section of the top coat and were formed due to the preferential attack of the lamellar splats. These lamellar splat structures are formed by fully-molten particle impingement. Within the

3: Applications and Developments, Kim

105

Figure 3.7 Cross-sectional view of as-sprayed VPS chromia blend and tantalum coatings (left); two-layer coating flake from Macraes component (center); and intact coating region from Canadian Autoclave component (right).

reduced-pressure argon environment of the VPS process, the fully-molten particles underwent in-flight changes in stoichiometry. Takeuchi et al.[11] have shown that chromium oxide (Cr2O3) applied via the VPS process forms a coating composed of different stoichiometries, e.g., Cr2O3, CrO, Cr3O4, and CrO2. It was evident from the observed failures that the Cr2O3 stoichiometry, maintained by the partially-melted or unmelted deposits, was resistant to corrosive attack, while the reduced stoichiometries of the fully-melted deposits did not resist corrosive attack. Past experience has shown that reduced stoichiometries often oxidize to its stable forms to provide protection; however, the extremely severe corrosive nature of the Ni/Co HPAL service seemed not to allow for this transition. Figure 3.7 (center and right) of the tantalum bond coat layers exposed to the Macraes and Canadian autoclave conditions reveal apparent signs of chemical attack or etching at the splat boundaries. Because of the inert, low-pressure environment in the VPS process, as-sprayed tantalum coatings did not provide visual evidence of splat boundaries resulting from oxidation (Fig. 3.7, left). In fact, the splat boundaries were clearly evident upon exposure to the autoclave conditions leading to their preferential chemical attack. This splat boundary corrosion, in turn, led to the permeability of the corrosive media to the substrate, creating a galvanic reaction between the tantalum bond coat and titanium substrate. The damage resulting from the galvanic corrosion between the tantalum bond coat and titanium substrate is clearly evident in the lower portion of Fig. 3.7, right. One can clearly see the cracks and flaws formed in both the bond coat and the substrate. The cracks were caused by hydrogen

106

Processing

embrittlement, with hydrogen forming at the cathode region (tantalum layer) as a result of the electrochemical reaction within the galvanic cell. Both tantalum and titanium are susceptible to hydrogen attack. The degradation within the metallic bond coat was also evident through microhardness measurements, where the average microhardness of the tantalum layer from the Canadian autoclave ball was measured to be 213 HV0.1 as compared to 163 HV0.1 on VPS as-sprayed tantalum coating.[10] The increased hardness of the tantalum layer, after exposure, is indicative of a chemical deterioration within the region, including hydrogen embrittlement leading to the crack formation.

3.4.2.2 Atmospheric plasma spray (APS) applied chromium oxide blend top coat with APS applied titanium bond coat from 10″ discharge valve ball used at Cawse Nickel for 7 months Flakes of chromium oxide blend top coat were removed from a 10″ discharge valve ball and characterized to determine their structure and composition. The bond coat remained on the component. Figure 3.8 shows a cross-sectional view of the top coat and the elemental maps of key elements in the corresponding region of the coating. The SEM micrograph of the flake shows the presence of unmelted particles and microcracks in a matrix of dense chromium oxide blend (Cr2O3-blend), typical of an APS applied ceramic coating. When compared to the as-sprayed microstructure of the same coating, the flake did not show any obvious signs of degradation. Elemental mapping of the cross-section did not reveal any significant segregation within the flake. The presence of a distinct siliconrich region was likely a result of contamination originating from the powder or the polishing media; in any case, the silicon-rich regions were too few to play any role in the coating’s performance. The SiC at the surface was part of the silicates found in the scale. Based on these results, there was no evidence of corrosive attack within the APS top coat.

3.4.3 Case Study–Nanostructured Titania-base Coating of Ball Valve Components for High-pressure Acid Leach Autoclaves In May of 2000, Mogas Industries partnered with FW Gartner Thermal Spraying Co. (FWGTS) and acquired the consulting services of Perpetual Technologies to develop a customized, nanostructured ceramic coating

3: Applications and Developments, Kim

107

Figure 3.8 SEM cross-sectional view of the APS top coat (top left); elemental mapping of the corresponding region (right); and EDS spectra of local regions of coating (bottom left).

specifically for Au and Ni/Co HPAL ball valves. The overall objective was to develop a coating that would outperform all existing coatings for this specific application. Upon a detailed analysis of the failed components, it was evident that none of the existing coatings/processes would adequately serve the HPAL application. Hence, a decision was made to proceed with the development of a nanostructured titanium dioxide (n-TiO2) based top coat with no bond coat. The main reasons for this decision were as follows: • The successful performance of titanium in HPAL autoclaves indicated that titanium oxide would stand up to the corrosive element of the harsh environment. • Titanium oxide’s coefficient of thermal expansion, similar to that of titanium, would aid in reducing the residual stress between the coating and substrate, leading to enhanced bond strengths.

108

Processing • The presence of a metallic bond coat was proven to accelerate corrosion and embrittlement.

Development activities were systematically pursued and by October 1, 2001, Mogas Industries and FW Gartner had qualified and initiated commercial application of the first-generation n-TiO2 coatings onto HPAL ball valve components. Numerous oral presentations and papers have been given and published on Mogas’ progressive n-TiO2 coatings effort.[12–18] Since this group’s initial introduction of n-TiO2 coatings to the public, at least one other group has focused their effort in developing a similar coating.[19] On February 26, 2003 an Australian Innovation Patent was issued to Mogas Industries and FW Gartner for the application of n-TiO2 coatings;[20] patents for other countries were filed in 2002. A U.S. patent was granted in December of 2004.[21]

3.4.3.1 Coatings Development (in chronological order) The general approach to the processing of nanostructured thermal-spray ceramic coatings is illustrated in Fig. 3.9. The main aspects of the approach included: • Synthesis of starting nanosized or ultrafine particles • Agglomeration of nanostructured or ultrafine particles • Development and optimization of thermal spray parameters and process selection • Characterization/testing of novel coatings Once an adequate powder feedstock was processed, thermal spraying was initiated using the APS process. Spray parameters were developed to optimize bond strength, microstructure, microhardness, and crack propagation characteristics. Comparisons were made between APS TiO2 coatings derived from conventional microstructured and customized nanostructured powders. Figure 3.10 compares the two coating crosssections. The commercial coating (Fig. 3.10, left) consists of lamellar structures with vertical microcracks and dense unmelted particles (outlined by dashed box) distributed throughout the coating. The different shades of gray reflect the different phases of TiO2 in the coating. The microstructure of the APS-applied nanostructured TiO2 coating (Fig. 3.10, right) had characteristics similar to the APS commercial coating. However, upon closer evaluation, differences were found in the nanostructured coating that included the presence of finer microcracks and a

3: Applications and Developments, Kim

109

Figure 3.9 Approach to thermal spraying of nanostructured ceramic coatings.

Figure 3.10 Cross-sectional view of APS TiO2 coatings: conventional (left) and nanostructured (right).

uniform distribution of unmelted agglomerates with fine pores (outlined by dashed box). A very interesting feature was noted when viewing the crack patterns formed by micro-indentations. Figure 3.11 provides views of the indentation regions (at 500 g load) for both APS coatings. It was apparent that the commercial coating had a higher number of wider and longer cracks, as compared to the nanostructured coating. A closer look at the region around the cracks in the APS-applied nanostructured coating (Fig. 3.12) revealed an interesting connection between the presence of nano-porous

110

Processing

Figure 3.11 Micro-indentation crack pattern on commercial (left) and nanostructured (right) APS-applied TiO2 coatings.

Figure 3.12 Field emission SEM micrograph of cracks around porous nanostructured unmelted particles.

unmelted or partially-melted particles and crack propagation. These finepored agglomerates seemed to deflect the crack and/or blunt the crack tip, thus hindering its propagation. The same effect was not observed on dense unmelted particles found in the commercial coating (arrow in Fig. 3.11, left). Once a connection between coating toughness and porous unmelted particles was realized, the focus shifted from the development and application of the novel nanostructured titanium dioxide coating using a conventional thermal spray process (APS) to the evaluation and development of other thermal spray processes to further optimize the microstructure and performance of the coating.

3: Applications and Developments, Kim

111

Figure 3.13 Cross-sectional view of n-TiO2 coating deposited using a proprietary process: field emission SEM of coating (left) and porous particles (center); micro-indentation crack pattern (right).

The technical objective was to determine whether the toughness of the coatings could be further improved without compromising their hardness or strength. A proprietary process was developed that met the challenge of introducing more of the porous unmelted particles without compromising the mechanical, cohesive integrity of the coating. Figure 3.13 provides a cross-sectional view and crack pattern of the n-TiO2 coating deposited by the proprietary process. At first glance, it was obvious that this structure differed greatly from previous APS-applied coatings. The structure was denser (<1% porosity) and uniform (nonlamellar and single-phase), with no signs of microcracks at these magnifications. The high-magnification view of the dark spots within the structure (Fig. 3.13, center) revealed the presence of the familiar porous unmelted particles, distributed uniformly throughout the coating. The crack pattern around the Vickers indent (Fig. 3.13, right) showed very low numbers of fine and short cracks, as compared to the APS coatings. This apparent increase in toughness was in addition to a notable increase in average microhardness from 783 HV0.3 to 1150 HV0.3. The ASTM G65 (procedure E) dry sand rubber wheel abrasion test results for the coatings are presented in Fig. 3.14. It is well understood that abrasive wear resistance is a reflection of strength or hardness and toughness. From left to right in Fig. 3.14, the coating gets not only harder but also tougher. It is this dramatic enhancement in toughness that provides such a drastic increase in abrasive wear resistance. Unlike conventional materials or coatings, where an increase in hardness is often accompanied by a decrease in toughness, these nanostructured coatings seem to defy this trend. The other wear property with direct relevance to the HPAL environment is slurry erosion. The following test procedures were used to evaluate the coatings:

112

Processing

Figure 3.14 Abrasive wear volume loss and microhardness of TiO2 coatings.

• A minimum of three tests were carried out on each coating with the scar volume loss measured using laser profilometry. • Alumina erosive particle Number 100 (150 μm) was mixed with neutral pH de-ionized water to form a slurry concentration of 0.66 wt%. • Continuous slurry jet had a constant flow rate of 1 liter/min and a constant velocity of 20 m/s. • The spray nozzle had a 1 mm inside diameter and 5 cm long barrel, placed 7.6 cm from the test-piece, which provided a straight cylindrical shaped jet. • Impingement angles were 30° and 90°. • Exposure durations were 5 min for 90° and 20 min for 30°. The results from the slurry erosion test (Fig. 3.15) showed that the nTiO2 coating had improved resistance, in the same order as the observed abrasion resistance. The hypothesis derived from the abrasion test results of enhanced toughness without compromising the hardness was substantiated when the n-TiO2 showed improved erosion resistance at both lowand high-angle slurry impingements. Conventional materials (including coatings) typically have a stronger resistance against low-angle erosion when the material is harder and more brittle (e.g., ceramics) and a stronger resistance against high-angle erosion when the material is softer and more ductile (e.g., metals). Changing the characteristics of a material to better resist one angle of impingement is typically compromised by a reduced resistance at the opposite angle of impingement. The fact that the n-TiO2 coating showed improved erosion resistance at both low and high angle

3: Applications and Developments, Kim

113

Figure 3.15 Slurry erosion volume loss at low and high impingement angles.

Figure 3.16 Photographs of n-TiO2 coating on Ti Gr12 coupons before (left) and after (right) exposure to 270°C sulfuric acid solution, at pH 0.09.

of impingement is a clear indication that the coating possessed higher hardness and toughness. The severe environment of the Ni/Co autoclaves made it difficult to determine the coating’s performance outside of placing it in an actual operating autoclave. The next best option was to carry out corrosion tests in environments approaching those of actual autoclaves. Test coupons were sprayed by FW Gartner and corrosion tested by Bacon Donaldson Consultants in British Columbia, Canada. Figure 3.16 provides photographs of the n-TiO2 coating on Ti Gr12 test samples before (left) and after (right) exposure to 270°C sulfuric acid solution, at pH 0.09. Although the corrosion test conditions were not severe enough to inflict any noticeable macroscopic coating degradation, changes in the bond strengths were evident and may be indicative of the degree of chemical attack within the coating and/or substrate. Details on the test conditions

Processing

114

Table 3.1 Corrosion Test Conditions and Their Effects on Bond Strengths

Sample Coating/Substrate n-TiO2/Ti Gr12 TiO2/Ti Gr12 n-TiO2/SAF 2507 n-TiO2(+)a/SAF 2507 TiO2(Au)b/SAF 2507 Cr2O3(Ta)c/A20 Cr2O3/A20 Cr2O3(+)a/A20 Cr2O3(Au)b/A20

Condition 270°C @ pH 0.09 270°C @ pH 0.09 190°C @ pH 2.0 190°C @ pH 2.0 190°C @ pH 2.0 50°C in H2SO4 50°C in H2SO4 50°C in H2SO4 50°C in H2SO4

Bond Strength (psi) As-Sprayed

After Exposure

5100 5820 2790 6660 7320 2760 2042 1260 1200

6600 3960 3120 7320 1800 1800 1350 2160 1050

a

(+) were coatings with the enhanced processing. (Au) were coatings with gold bond coat; all coatings in this category were sprayed by a thermal spray provider other than FW Gartner. c (Ta) were coatings with tantalum bond coat. b

and bond strengths for coatings sprayed by both FW Gartner and another thermal spray provider are provided in Table 3.1. Although the bond strength data after corrosion testing were derived from a single sampling per coating type, values that show changes greater than the typical data scatter of below 15% are likely indicative of real changes. As expected, the as-sprayed bond strengths were generally related to the degree of mismatch in CTE between the top coat and substrate materials. In most cases, greater CTE mismatches resulted in lower bond strength. The increase in bond strength found in some samples exposed to the relatively elevated temperatures is likely due to the stress relief effect within the coating. Characterization work on the samples is ongoing and will continue to provide new insights into the unique properties of this coating. At present, it is hypothesized that any significant decrease in bond strengths is likely attributable to corrosive attack at the metallic substrate surface. A critical component to the performance and longevity of any coating is its bond strength with the underlying substrate. The program described above invested considerable effort into enhancing this critical feature and in addition incorporates a post-spray enhancement process to further

3: Applications and Developments, Kim

115

increase bond strength, hardness, and the density of the n-TiO2 coating. This process enhancement has substantially increased the bond strength in present production n-TiO2-2004 coatings (11 832 psi bond strength), as compared to the previous production n-TiO2-2002 coatings (4960 psi bond strength). The other positive attributes of the process enhancement included a reduction of at least 35% in porosity and an increase of 7% or more in microhardness.

3.4.3.2 Field Result Due to numerous variables in the service conditions that each valve endures, most performance comparisons between coated valves are made in an indirect manner with, in some cases, uncertain critical assumptions. In 2003, Lihir Gold (a mine in Papua New Guinea) installed two 10″ ID Ti Gr5 Mogas ball valves (with n-TiO2 and Cr2O3-blend coatings) into the same service exposures and conditions for the same duration. Figure 3.17 shows photographs of the ball surface upon inspection after 10 months service. It was clearly evident that the n-TiO2 coating was in a far superior state compared to the nonnanostructured Cr2O3-blend coating. The surface of the Cr2O3-blend-coated ball had large regions without coating. In contrast, the n-TiO2-coated ball had a few isolated regions without coating. Two additional valves have recently been removed for viewing during a scheduled shutdown at Lihir Gold. The n-TiO2 coating on both the autoclave feed (8″ ID) and discharge (10″ ID) valves were in very good shape following a 12-month service period (Fig. 3.18). Apart from some minor

Figure 3.17 Photographs of the ball surfaces with Cr2O3-blend coating (left) and with n-TiO2 coating (right) after same service exposure and conditions.

116

Processing

Figure 3.18 Photographs of ball surfaces with n-TiO2 coating after 12 months of service in Lihir’s gold autoclaves: 8″ ID feed valve (left); 10″ ID discharge valve (right).

scratches and localized chips, the coating was in very good shape and was returned into service.

3.5 Conclusions In the field of surface engineering there is constant and widespread research seeking to develop new, advantageous processes and materials. Among these research initiatives, thermal-sprayed nanostructured oxide coatings emerge as one of the very few that have matured and proven themselves to be practically advantageous for both military and industrial applications. Although many commercial, nonnanostructured thermal spray coatings provide adequate performance for certain applications, nanostructured coatings will provide solutions to improve on existing applications or to allow for new applications that require the unique and added benefits that can only be attained through this approach.

Acknowledgments I would like to thank Professor Enrique Lavernia for his direction, assistance, and guidance. I also wish to thank Dr. Lawrence Kabacoff, Professor Julie Schoenung, and Ken Scandell for their constructive contributions toward the preparation and the review of the manuscript. Special thanks should be given to Jimmy Walker and FW Gartner Thermal Spraying Co. for their contribution of industrial thermal-spray processing photographs. Last but not least, I wish to thank John Williams of Scientific Valve + Seal, formerly with Mogas Industries, for his leadership towards the successful development of the nanostructured titania coating.

3: Applications and Developments, Kim

117

References 1. Lawrence T. Kabacoff, “Nanoceramic Coatings Exhibit Much Higher Toughness and Wear Resistance than Conventional Coatings”, The AMPTIAC Newsletter, vol. 6, no. 1 (Spring 2002). 2. M. Gell with E.H. Jordan et al., “Fabrication and Evaluation of Plasma Sprayed Nanostructured Alumina-Titania Coatings with Superior Properties,” Mater. Sci. Eng., A301:80–89 (2001). 3. M. Gell with L. Shaw et al., “Development and Implementation of Plasma Sprayed Nanostructured Ceramic Coatings,” Surface and Coatings Technology, vol. 146–147:48–54 (2001). 4. D. Goberman, Y. Sohn, L. Shaw, E. Jordan, and M. Gell, “Microstructure Development of Al2O-13 wt%TiO2 Plasma Sprayed Coatings Derived from Nanocrystalline Powders,” Acta Materialia, 50:1141–1152 (2002). 5. T. Bernecki, “Surface Science,” in: Handbook of Thermal Spray Technology, (J.R. Davis, ed.), p. 14–35, ASM International, Materials Park, OH (2004). 6. F.W. Gartner Thermal Spraying Co. 7. R. Unger, “Comparison of Thermal Spray Bond Coats,” Proceedings of the National Thermal Spray Conference, pp. 365–370, ASM International, Materials Park, OH (1987). 8. F.W. Gartner Thermal Spraying Co. 9. K. Scandell, “From Science to Service . . . What Are The Issues As A View From The Fleet?” Nano Materials for Defense Applications Symposium, Honolulu, HI (Island of Oahu), February 17–18 (2005). 10. G.E. Kim, T.A. Brzezinski, L. Leblanc, and E. Kharlanova, “Thermal Spray Coatings for Ball Valves used in Nickel/Cobalt Pressure Acid Leaching,” Proceedings of the ITSC 2000 Conference, pp. 1149–1153, Montreal, Quebec, Canada (2000). 11. Takeuchi J., Nakahira H., and Nagai J., “Physical Properties of Some Oxide Coatings by Low Pressure Plasma Spraying,” 2nd Plasma Technik Symposium, vol. 2:141–151 (1991). 12. J. Williams, G.E. Kim, and J. Walker, “Surface Coatings for Ball Valves in NiHPAL Service,” Proceedings of the ALTA 2001 Ni/Co Conference, PAL Plant Materials and Equipment, Tuesday, 15th of May, Presentation 9, pp. 1–23, Perth, Australia (2001). 13. G.E. Kim, “Advances in Nanostructured Thermal Spray Coatings,” Invited Speaker at ASM International’s Houston Chapter Meeting, Houston, TX (2004). 14. G.E. Kim, “Nanostructured Coatings and Bulk Materials,” Invited Speaker at ASM International’s Montreal Chapter Meeting, Montreal, Quebec, Canada (2001). 15. G.E. Kim, J. Williams, and J. Walker, “Nanostructured Coating Application in High-Pressure Acid-Leach Process,” Proceedings of the Nano2002 Conference, Orlando, FL (2002). 16. G.E. Kim, “Modern Surface Coatings for Metal Seated Ball Valves in Severe Service Environments,” Oral Presentation at Valve World Expo 2002, Maastricht, Holland (2002).

118

Processing

17. G.E. Kim, “Nanostructured Coatings Application in High Pressure Acid Leaching Process,” Invited Speaker at the Surface Technology (SURFTEC) Group Meeting, National Research Council of Canada, Montreal, Quebec, Canada (2002). 18. J. Williams, G.E. Kim, and J. Walker, “Ball Valves with Nanostructured Titanium Oxide Coatings for High-Pressure Acid-LeachService: Development to Application,” Proceedings of Pressure Hydrometallury 2004, pp. 663–681, Banff, Alberta, Canada, October 23–27 (2004). 19. R.S. Lima, L. Leblanc, and B.R. Marple, “abrasion behaviour of nanostructured and conventional titania coatings thermally sprayed via APS, VPS and HVOF,” Proceedings of the ITSC 2004 Conference, Osaka, Japan (2004). 20. G.E. Kim, J. Williams, and J. Walker, “Nanostructured Titania Coated Titanium,” Australian Innovation Patent, No. 2002100618, February 26 (2003). 21. G.E. Kim, J. Williams, and J. Walker, “Nanostructured Titania Coated Titanium,” United States Patent No. 6,835,449, December 28 (2004).

4 Nanostructured Materials and Composites Prepared by Solid State Processing H.J. Fecht and Yu. Ivanisenko University of Ulm, Department of Materials, Ulm, Germany

4.1 Introduction and Background Mechanical deformation under shear conditions and high strain rates (∼101–104 s−1) leads to the formation of nanostructures within powder particles and thin foils or at the surface of metals and alloys exposed to friction-induced wear conditions. For example, mechanical attrition and mechanical alloying of powder particles has been developed as a versatile alternative to other processing routes in preparing nanoscaled materials with a broad range of chemical composition and atomic structure.[1,2,3] In this process, lattice defects are produced by “pumping” energy into initially single-crystalline powder particles of typically 50 μm particle diameter. This internal refining process with a reduction of the average grain size by a factor of 103–104 results from the creation and self-organization of small-angle and high-angle grain boundaries within the powder particles during the mechanical deformation process. As a consequence, a change of the thermodynamic, mechanical, and chemical properties of these materials has been observed with the properties of nanophase materials becoming controlled by the grain size distribution and the specific atomic structure and cohesive energy of the grain or interphase boundaries. In the 1970s, the method of mechanical attrition (MA) of powder particles followed by high-temperature sintering was developed as an industrial process to successfully produce new alloys and phase mixtures. For example, this powder metallurgical process allows the preparation of alloys and composites that cannot be synthesized via conventional casting routes. This method can yield[4] Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 119–172 © 2007 William Andrew, Inc.

119

Processing

120

(1) Uniform dispersions of ceramic particles in a metallic matrix (superalloys) for use in gas turbines (2) Alloys with different compositions than alloys processed from the liquid (3) Alloys of metals with quite different melting points with the goal of improved strength and corrosion resistance In the 1980s, the method of high-energy milling gained a lot of attention as a nonequilibrium solid state process resulting in materials with nanoscale microstructures. The formation of nanocrystals within initially singlecrystalline powder samples was first studied systematically in pure metals and intermetallic compounds.[5] Moreover, solid state (mechanical) alloying beyond the thermodynamic equilibrium solubility limit can lead further to the formation of amorphous metallic materials as observed for a broad range of alloys.[6,7,8] The formation of the amorphous phase occurs by intermixing of the atomic species on the atomic scale, thus softening and destabilizing the crystalline lattice[9] and driving the crystalline solid solution outside of its range of stability against “melting” or amorphization.[10] This process is considered as a result of both mechanical alloying[11,12] and the incorporation of lattice defects into the crystal lattice.[13] More recent investigations demonstrate that the formation of nanostructure can also occur in several unexpected cases, such as brittle ceramics, ceramic phase mixtures, polymer blends and metal/ceramic nanocomposites.

4.2 Phenomenology of Nanostructure Formation The milling of materials is of prime interest in the mineral, ceramic processing, and powder metallurgy industries.[14] Typical objectives of the milling process include particle size reduction (comminution), solid-state alloying, mixing or blending, and particle shape changes. These industrial processes are mostly restricted to relatively hard, brittle materials that fracture, deform, and cold weld during the milling operation. While oxide dispersion strengthened superalloys have been the primary application of mechanical alloying, the technique has been extended to produce a variety of nonequilibrium structures including nanocrystalline,[15] amorphous,[6] and quasicrystalline[16] materials (for a review see Ref. 17). A variety of ball mills has been developed for different purposes including tumbler mills, attrition mills, shaker mills, vibratory mills, planetary mills, etc.[18] The basic process of mechanical attrition is illustrated in

4: Solid State Processing, Fecht and Ivanisenko

121

Figure 4.1 Schematic sketch of the process of mechanical attrition of metal powders. Vector V indicates the velocity directions of colliding balls.

Fig. 4.1. Powder particles with typical particle diameters of about 50 μm are placed together with a number of hardened steel or Tungsten carbide (WC) coated balls in a sealed container that is shaken or violently agitated. High-energy milling forces can be obtained by using high frequencies and small amplitudes of vibration. Ball mills (e.g.. SPEX model 8000), which are preferable for small batches of powder (∼10 cm3, sufficient for research purposes), are highly energetic and reactions can take place one order of magnitude faster compared with other types of mills. Since the kinetic energy of the balls is a function of their mass and velocity, dense materials (steel or tungsten carbide) are preferable to ceramic balls. During the continuous severe plastic deformation, a continuous refinement of the internal structure of the powder particles to nanometer scales occurs during high-energy mechanical attrition. The temperature rise during this process is modest and is generally estimated to be ≤100°C to 200°C. The collision time generally corresponds to about 2 μs. For all nanocrystalline materials, surface and interface contamination constitutes a major problem. During mechanical attrition, contamination by the milling tools (Fe) and atmosphere (trace elements of O2, N2 in rare gas) can be problematic. By minimizing the milling time and using the purest, most ductile metal powders available, a thin coating of the milling tools by the respective powder material can be obtained, which reduces Fe-contamination tremendously. Atmospheric contamination can be minimized or eliminated by sealing the vial with a flexible “O”-ring after the powder has been loaded in an inert-gas glove box. Small experimental ball mills can also be enclosed completely in an inert-gas glove box. As a consequence, the contamination with Fe-based wear debris can generally be reduced to less than 1–2 at% and oxygen and nitrogen

122

Processing

contamination to less than 300 ppm. In this respect, the nanoscaled powder material obtained has often a higher purity than materials synthesized by alternative methods, such as chemical processes or inert-gas evaporation and condensation. In addition, the structural contributions of small pores occurring during compaction of small clusters can be safely neglected.[19] Milling of refractory metals (e.g., tungsten) in a shaker or planetary mill for extended periods of time (>30 hours) can result in levels of Fecontamination of more than 10 at% if high vibrational or rotational frequencies are employed. On the other hand, contamination from the milling atmosphere can have a positive impact on the milling conditions if one intends to prepare metal/ceramic nanocomposites with one of the metallic elements being chemically highly reactive with the gas (or fluid) environment. The deformation processes within the powder samples are important for fundamental studies of extreme mechanical deformation and the development of nanostructured states of matter with particular physical and chemical properties; similar processes control the deformation of technologically relevant surfaces. For example, the effects of work hardening, material transfer, and erosion during wear situations result in microstructures of wear surfaces comparable to those observed during mechanical attrition.[20] In particular, during sliding wear, large plastic strains and strain-gradients are created near the surface.[21] Similarly to mechanical attrition of powder particles, this is the consequence of the formation of dislocation cell networks, subgrains, and grain boundaries, with the subgrains becoming smaller near the surface.

4.3 High-Energy Ball Milling and Mechanical Attrition 4.3.1 Examples 4.3.1.1 Metallic Elements and Intermetallics During mechanical attrition, the metal powder particles are subjected to severe plastic deformation from collisions with the milling tools. Consequently, plastic deformation at high strain rates (∼103–104 s−1) occurs within the particles and the average grain size can be reduced to a few nanometers after extended milling. As such, the metal particle is plastically deformed, with most of the mechanical energy expended in the

4: Solid State Processing, Fecht and Ivanisenko

123

deformation process being converted into heat, but the remainder being stored in the metal, thereby raising its internal energy.[22,23] This was first investigated in detail for a number of high-melting metals with bcc and hcp crystal structures.[5,15] Metals with fcc structure are inherently more ductile and often exhibit a stronger tendency to adhere to the container walls and to sinter to larger particles, often several millimeters in diameter, during the milling process. In a detailed study, the successful preparation of nanocrystalline fcc metals has been described.[24] Since in general contamination from the milling devices (hardened steel) can occur, the experimental results discussed in the following are mainly concentrated on iron powder as a model system for mechanical attrition. The microstructural changes as a result of mechanical attrition can be followed by X-ray diffraction (XRD) methods averaged over the sample volume. The XRD patterns exhibit an increasing broadening of the crystalline peaks as a function of milling time. The peak broadening is caused by size as well as internal strain effects.[25,26] The average coherently diffracting domain size (grain or crystal size) and the microstrain as a function of milling time are obtained from the integral peak widths assuming Gaussian peak shapes (see Fig. 4.2).[27] After corrections for Kα and instrumental broadening, the line broadening due to the small crystal size is constant in K-space and is given by ΔK = 0.9 (2π/d), where d is the average domain or grain diameter. The strain broadening corresponds to ΔK = A〈e2〉1/2 K, with A being a constant

Figure 4.2 The average grain size and microstrains as determined from X-ray line broadening as function of milling time for iron powder.

124

Processing

depending on the strain distribution (A ≈ 1 for a random distribution of dislocations[28]) and〈e2〉1/2 being the rms strain. Additional defects that might contribute to the peak broadening, such as stacking faults, can be safely neglected in all cases discussed here. However, for some metals with very small stacking fault energies (e.g., Co), the contribution of stacking faults to the peak broadening is considerable.[29] In the very beginning, mechanical attrition leads to a rapid decrease of the average grain size to 40–50 nm. Further refinement occurs slowly to about 15 nm after extended milling. The average atomic-level strain reaches values up to 0.7%. Depending on the method of X-ray analysis applied (Scherrer formula,[30] Williamson and Hall method at full width at half-maximum, integral peak width at half-maximum,[25] Warren–Averbach analysis,[31] etc.), the weighting of the grain size distribution is different and therefore the average grain size can vary by a factor of 2. Essentially all metals and compounds investigated so far have exhibited similar behavior in terms of grain size reduction and an increase in atomic-level strains. Typical values for the average grain sizes of fcc, bcc, and hcp metals vary between 10 and 20 nm, whereas the atomic-level strains can increase up to about 1%.[32] It is interesting to note that plastic deformation can be introduced in nominally brittle materials by mechanical attrition. The minimum grain or domain sizes for intermetallic compounds with CsCl structure have been found to vary between 12 nm for CuEr[33] and 2 nm (amorphous) for NiTi.[34] Furthermore, in ordered intermetallic compounds anti-site disorder is introduced during mechanical attrition, leading to atomic-level strains up to 3%. Whereas for the CsCl compounds the reduction of the long-range chemical order parameter saturates at about 0.7, other intermetallic compounds exhibit complete disordering together with the formation of a nanocrystalline solid solution. For example, the A-15 type compounds Nb3Al,[35] V3Ga,[36] and Nb3Au[37] transform after extended milling to a bcc solid solution with nanometer-sized grains.

4.3.1.2 Nonequilibrium Crystalline and Amorphous Solid Solutions The mechanical alloying of powder mixtures generally results in the formation of solid solutions extended in composition far beyond their equilibrium solubility limit (up to a factor 10) and this is caused by mechanically driven enhanced interdiffusion. The phenomenon of stressinduced diffusion is typical if large potential gradients prevail, which lead

4: Solid State Processing, Fecht and Ivanisenko

125

to high-rate diffusion processes in the vicinity of a dislocation even at temperatures at which self-diffusion is not possible.[38] At large deformations, layered structures or precipitates can therefore be dissolved. The action of deformation (i.e. “driven systems”) and especially shearing processes in causing atomic-scale mixing have been further clarified recently by computer simulation.[39,40] The mechanical alloying process is considered as an athermal process that yields a high level of homogenization of the component atoms provided that sufficient deformation is applied. In fact, Monte Carlo simulations indicate that deformation can yield a solid solution even in alloys with a positive enthalpy of mixing. In this case, the driven system action, characterized by the forcing parameter (the frequency ratio between forced and thermally activated jumps), results in a behavior with enhanced diffusivity characteristic of a hightemperature high-entropy state with extended solubilities.[41–43]

4.3.1.2.1 Positive Enthalpy of Mixing Surprisingly, mechanical alloying is achieved also for powder mixtures having a positive enthalpy of mixing. Although in some cases, such as Ag-Fe, an intimate phase mixture of nanostructured Ag and Fe particles is produced,[44] in other cases real miscibility on an atomic level can be obtained, e.g., for Cu—Fe,[45] Cu—W,[46] Cu—Ta,[47] and Cu—V.[48] This apparent violation of the rules of equilibrium thermodynamics is a vital example of the potential of mechanical attrition in synthesizing new materials under nonequilibrium conditions. For example, mechanical alloying can lead to the formation of nanocrystalline single-phase solid solutions of up to 60 at% Fe in Cu and 20 at% Cu in Fe.[49,50] The steady-state grain sizes range from 20 nm for Cu to 8–10 nm for Fe-rich alloys as determined by X-ray line broadening (see Fig. 4.3). The enhanced solubility of alloys exhibiting spinodal behavior in coarse-grained systems has been attributed to the contribution of the capillary pressure to the free energy of nanosized grains due to their small radii of curvature. It has been found that during mechanical attrition of Fe-Cu powder mixtures, agglomerates of multilayers are formed, leading to microstructures very similar to those obtained by cold-rolling.[51] Furthermore, mechanical attrition can also produce ultrafine-scaled phase mixtures if a brittle material is milled together with a more ductile material. For example, 10 nm sized Ge particles can be embedded in a ductile matrix of Sn or Pb.[52] Similarly, very fine dispersions at the nanometer scale have been found, for example, in TiNi-C[53] and Ag—

126

Processing Fe.[54] In additional Mössbauer studies, mutual solubility Ag in Fe domains (and Fe in Ag domains, respectively) could be shown. As such, the resulting microstructure and chemical arrangement are very similar to nanocrystalline phase mixtures prepared by gas condensation methods.

4.3.1.2.2 Negative Enthalpy of Mixing/Glass Formation Extended solid solutions far beyond the thermodynamic equilibrium have also been noted in the course of mechanical milling of alloys with negative enthalpies of [55] The formation of an amorphous structure has often been mixing. observed for phase mixtures with large (>15%) differences of atomic radii.[56] During this process long-range solute diffusion and therefore solute partitioning are suppressed. For example, during mechanical alloying of 75a/o Zr and 25a/o Al the formation of a supersaturated hcp (α-Zr) solid solution was observed prior to the solid-state amorphization reaction.[57] However, in all cases of binary alloys it remained unclear whether indeed a metallic glass has been formed or just an X-ray amorphous structure. More recently, a similar phase transformation sequence was found in a mechanically alloyed multicomponent elemental Zr60Al10Ni9Cu18Co3 powder mixture with bulk glass-forming composition as shown in Fig. 4.4.[58] The X-ray spectra at different stages of the milling process are characterized by the successive disappearance of the elemental Al, Co, Cu, and Ni peaks and a simultaneous shift of the Zr peaks to higher scattering angle, corresponding to a decrease in the lattice constant of the hcp-Zr as a result of the rapid dissolution of the smaller atoms such as Cu, Ni, Co, and Al in the (α-Zr) matrix.[59] Calorimetric investigations further revealed that a glass transition indeed occurred when the amorphous material was heated to the undercooled liquid state without prior crystallization. Figure 4.5 shows the corresponding increase of the specific heat capacity when the metallic glass sample prepared in the solid state was heated through the glass transition. Figure 4.3 Average grain size for FexCu100−x powders after 24 hours of milling versus Fe content (after Ref. 50).

4: Solid State Processing, Fecht and Ivanisenko

127

Figure 4.4 X-ray spectra for mechanically alloyed Zr60Al10Ni9Cu18Co3 powder samples exhibiting a transition from the initially crystalline powder mixture to an amorphous glasslike structure.

Figure 4.5 Thermal analysis showing the difference in heat capacity Cp of the sample material Zr60Al10Ni9Cu18Co3 in comparison with the thermodynamically stable crystalline configuration (at 0) for amorphous mechanically alloyed powder (round black symbol), bulk metallic glass (open square), and foil stacks (closed square).

128

Processing

Similar observations concerning the crystal-to-glass transition by mechanical attrition have also been observed for the equilibrium intermetallic phase mixture.[60] These rather surprising results confirm that a metallic glass can really be formed in the solid state (without cooling from the liquid state) by destabilization of a crystal due to the incorporation of static disorder causing high elastic stresses.[61]

4.3.1.3 Ceramics Whereas ductile materials can be deformed as described above, it is surprising that nominally brittle materials, such as ceramics, also show a refined microstructure after extended mechanical attrition. For example, ZrO2 has been milled for up to 40 hours, resulting in a grain size reduction to several nanometers. At a grain size of less than about 12 nm, a transition of the most stable monoclinic structure to a metastable orthorhombic modification has been observed.[62] A simple estimate shows that below a critical grain size of about 30 nm the high-temperature phase becomes more stable than the low-temperature phase due to the incorporation of defects. Mechanical alloying occurs for ceramic powder mixtures as well. For example, (Fe,Cr)2O3 solid solutions can be obtained from a Fe2O3/Cr2O3 powder mixture with an average grain size of about 20 nm.[63] Similar observations have been reported for ZrO2/Y2O3 powder mixtures.[64] Chemical processes can also be induced by milling of ceramic materials. For example, extended periods of mechanical milling can lead to the deoxidation of oxides when appropriate materials are added. Mechanochemical reactions have been reported for α-Fe2O3/Ti, Ag2O/C[65] and CuO/Ca[66] mixtures.

4.3.1.4 Polymer Blends Mechanical alloying of polymeric materials has also been developed during the past decade. Similarly to metallic materials, mechanical attrition leads to an increase of the internal energy. For example, polyamide (PA), polyethylene (PE), acrylonitrile-butadiene-styrene (ABS), polypropylene (PP) and polystyrene (PS) have been investigated in detail.[67] In order to fracture the polymer chains, the milling process is conducted below the respective glass transition temperatures. As a result, the crystallinity of the powder material can be decreased considerably by mechanical milling. The corresponding storage of energy allows consol-

4: Solid State Processing, Fecht and Ivanisenko

129

idation of the powder to bulk samples at lower temperatures than for conventionally processed material. In addition, the mechanical properties achieved are considerably improved. The milling process also allows mechanical alloying of polymer mixtures (PA/PE, PA/ABS),[68] as well as mechanical alloying of polymers with ceramic (PP/SiC) and metal powder (PS/Sn, PP/Al, PE/Cu[69]). By mechanical milling, the normal compatibility restrictions for polymer formation derived from regular polymer science principles have been removed. This unique opportunity allows the synthesis of new materials and materials combinations with enhanced properties that cannot be achieved by any other method.

4.3.1.5 Nanocomposites Mechanical attrition is also a very versatile process for preparing nanocomposites. Since this process is sensitive to contamination resulting from the milling environment, atmospheric control can be used purposely to induce chemical reactions between the milled powders and their environment. By a proper choice of a reactive gas atmosphere (O2, N2, etc.) or a milling fluid (organic fluids), the metal powder can be intentionally modified by reactive milling to a nanocrystalline metal–ceramic composite[70] or fully reacted to a nanocrystalline ceramic, for example a metal nitride.[71] Metal powders of Ti, Fe, V, Zr, W, Hf, Ta and Mo[72–74]) are transformed to a nanocrystalline nitride by high-energy ball milling under nitrogen gas flow. This solid-state interdiffusion reaction during reactive ball milling is triggered by fragmentation of the starting powder, thus creating new surfaces. These freshly created surfaces react with the flowing nitrogen gas to form a nitride surface layer over the unreacted core particle. With further milling, this reaction continues and a homogeneous nitride phase is formed and the unreacted core of metal disappears, resulting in a nanostructured (often metastable) metal nitride with a grain size of typically 5 nm. By ball milling in organic fluids as surfactants, which are sometimes used to prevent contamination by the milling tools, chemical reactions can be induced leading to the formation of fine carbides. For example, by milling Al (-Ti, -Zr or -Hf) alloys in hexane, an average grain size of 9 nm can be achieved, with carbon being dissolved in the matrix.[75] During dynamic compaction at about 1300 K, grain growth occurs up to about 44 nm together with precipitation of ZrC particles, 7 nm in size.

130

Processing

Figure 4.6 Scanning electron microscope image of metal/ceramic composite of 10 vol% SiC particles in an amorphous matrix (Zr65Al7.5Cu17.5Ni10) prepared by mechanical alloying.

Such ultrafine-grained composites are expected to exhibit considerably improved strength and ductility.[76] More recently, metallic glass/ceramic composites were obtained by mechanical alloying of multicomponent Zr-based elemental metallic powders together with SiC particles.[77,78] A scanning electron microscopy image of such a Zr65Al7.5Cu17.5Ni10 + 10 vol.% SiC metallic glass/ceramic composite after a milling time of 30 hours (Fig. 4.6) reveals a uniform distribution of fine SiC particles in the metallic glass powder matrix, as demonstrated by further XRD and EDX analysis. The size distribution of the SiC particles ranges from 1 μm down to values below 50 nm. It is further interesting to note that the SiC particles do not act as potent heterogeneous nucleation sites when the composite is heated to the crystallization temperature above the glass transition temperature. As such, mechanical alloying represents a convenient method to achieve dispersion-strengthened amorphous alloys with considerably improved strength and wear resistance by a powder metallurgical pathway.

4.3.2 Mechanism of Grain Size Reduction From wide angle X-ray spectra, information about lattice defects (grain boundaries, dislocations, etc.) is obtained via their disturbing influence on the coherent superposition of radiation diffracted at the atomic lattice sites, which causes the broadening of Bragg peaks. In small-angle neutron scattering experiments (SANS) the lattice defects themselves give rise to a

4: Solid State Processing, Fecht and Ivanisenko

131

scattering contrast because of the (scattering length) density fluctuations associated with them. Ball-milled Fe powder samples were measured in quartz cuvettes at neutron wavelengths l–0.60 nm. The cuvettes were filled with D2O, resulting in a 79% reduction of scattering contrast at the wetted surfaces of the powder particles. Different positions between the sample and the area-sensitive detector were chosen covering a range of momentum transfer q = (4p/l) sin(q) from 0.045 nm−1 to 0.85 nm−1 (2q = scattering angle). During the measurements, a homogeneous magnetic field of 0.7 T was applied to the sample in horizontal direction perpendicular to the incoming neutron beam. For a magnetically saturated sample, the scattered intensity as a function of the vector of momentum transfer q can be written as I(q) = IN(q) + IM(q) sin2α, where IN(q) and IM(q) represent the structure function of the nuclear and magnetic scattering, respectively,[79] and α is the azimuthal angle between q and the magnetic field, projected on the area perpendicular to the incoming beam. The anisotropic intensity distributions were analyzed by radial averaging over angular sectors parallel and perpendicular to the direction of the applied magnetic field. While in the first case, α = 0°, the spectra thereby obtained represent the nuclear scattering contribution, perpendicular averaging, α = 90°, yields a linear combination of nuclear and magnetic scattering. Radial distribution functions (RDF) were calculated from the SANS spectra by the indirect Fouriertransformation method.[80] Figure 4.7 shows the SANS spectra of Fe powder samples before milling and after milling for 0.5 hours and 30 hours averaged parallel to the applied magnetic field. Compared to the spectrum of the unmilled sample, an increase of scattering intensity occurs after 0.5 hours milling over the entire q-range covered by the measurements. This increase may be explained by the refinement of microstructure in the early stages of the milling process (i.e., the scattering contribution of grain boundaries, dislocations, and triple junctions), which is demonstrated by the X-ray peak broadening. It should be noted that the increase of scattering intensity extends to the high-q range, implying that structural inhomogeneities on a small length scale of a few nanometers are present after short milling times. This observation underlines that the structural refinement is strongly inhomogeneous and might occur in shear bands of high dislocation density surrounded by less-deformed sample regions. After 30 hours milling time, a drop of the scattering intensity is observed over the entire q-range compared to the 0.5-hour sample. The drop at high q values is especially surprising since, from X-ray

132

Processing

Figure 4.7 SANS spectra of ball-milled Fe powder after different milling times (0 h, 0.5 h, 30 h).

Figure 4.8 Radial distribution functions (volume-weighted) calculated from spectra in Figure 7 (milling times 0.5 h, 30 h).

measurements, it is known that the average grain size is further reduced to a volume average of about 16 nm, which is in good agreement with the average grain size derived from the RDF of about 15 nm after 30 hours milling. Furthermore, the RDF also shows a decrease in magnitude without any significant shift of their maxima (Fig. 4.8). Obviously, the observed change of the SANS intensity cannot be explained solely by a shift of crystallite size distribution to smaller dis-

4: Solid State Processing, Fecht and Ivanisenko

133

tances in real space during the milling process. Instead it is believed that the scattering contrast due to dislocations plays an important role for the interpretation of the measured data. Since the volume dilatation in the vicinity of dislocations is small, their nuclear scattering contrast is small and disappears for pure screw dislocations.[81] Their magnetic scattering contrast due to orientation fluctuations of the magnetic moments of Fe atoms may exceed by factors up to 10–100 their contribution to nuclear scattering.[82] These fluctuations are caused by magnetoelastic coupling between the magnetic moments and the dislocation strain field. The SANS data may be interpreted with respect to the magnetic scattering contribution of dislocations as a result of a changing distribution of dislocations in the deformed material. According to this model, the increase of scattering intensity at high q after 0.5 hour milling is caused by an increasing dislocation density with an average dislocation distance of a few nanometers, which is in the range of the final grain size after long milling times. The subsequent decrease of scattering intensity after 30 hours milling time may be caused by the rearrangement of dislocations to form grain boundaries and the absorption of dislocations as secondary grain boundary dislocations. The grain refinement, which can be followed only in an integral way by X-ray or neutron scattering methods, has been investigated in more detail by TEM on intermetallic AlRu powder samples.[83] Figure 4.9 shows a series of TEM micrographs of an AlRu particle after mechanical attrition. It can be seen that the crystal is heavily strained and the deformation occurs in a rather inhomogeneous way. The arrows in Fig. 4.9(a) indicate a highly deformed region of a width of about 1 μm that extends throughout the entire particle. These shear bands have been observed in rolled metals and are typical for deformation mechanisms that occur at high strain rates in contrast to slip and twinning mechanism at low and moderate strain rates.[84] The observed shear bands are separated by areas of similar lateral dimensions in the micrometer range having low defect densities. Highresolution imaging of areas in the shear bands reveals a microstructure consisting of individual grains with a diameter of ∼20 nm which are slightly rotated with respect to each other at a rotation angle of less than 20°, as shown in Fig. 4.9(b). With longer durations of mechanical attrition, the shear bands grow over larger areas (see Fig. 4.9(c)) and eventually the entire sample disintegrates into subgrains with a final grain size of 5–7 nm for AlRu after 64 hours (see Fig. 4.9(d)), thus ductilizing the originally brittle intermetallic compound. The elemental processes leading to grain size refinement include three basic stages:

134

Processing

Figure 4.9 (a) TEM bright-field images at relatively low magnification of an AlRu powder particle after 10 minutes of mechanical attrition. The arrows point to highly deformed regions (shear bands). The inset shows the corresponding diffraction pattern, demonstrating the gradual smearing out of the initially sharp diffraction spots. (b) TEM high-resolution bright-field image with its corresponding diffraction patterns of AlRu after 10 minutes. (c) After 2 hours and (d) after 64 hours of mechanical attrition.

(1) Initially, the deformation is localized in shear bands consisting of an array of dislocations with high density. (2) At a certain strain level, these dislocations annihilate and recombine to small-angle grain boundaries separating the individual grains. The subgrains formed via this route are already in the nanometer size range (∼20–30 nm). (3) The orientations of the single-crystalline grains with respect to their neighboring grains become completely random. This can be understood in the following way. The yield stress σ required to deform a polycrystalline material by dislocation movement is related to the average grain size d by σ = σ0 + kd −1/2, where σ0 and k are constants (Hall–Petch relationship[85,86]). An extrapolation to nanocrystalline dimen-

4: Solid State Processing, Fecht and Ivanisenko

135

sions shows that very high stresses are required to maintain plastic deformation. Experimental values for k and σ0 are typically k = 0.5 MNm−3/2 and σ0 = 50 MPa.[87] For a grain size of 10 nm, the minimum yield stress is of the order of 5 GPa, corresponding to 15% of the theoretical shear stress, which sets a limit to the grain size reduction achievable by plastic deformation during ball milling. Therefore, the reduction of grain size to a few nanometers is limited by the stresses applied during ball milling as long as no dramatic elastic softening of the crystal lattice occurs. Further energy storage by mechanical deformation is only possible by an alternative mechanism. Grain boundary sliding has been observed in many cases at high temperatures, leading to superplastic behavior. Alternatively, grain boundary sliding can also be achieved at very small grain size and low temperature by diffusional flow of atoms along the intercrystalline interfaces, which allows the synthesis of ductile ceramics.[88] This provides a mechanism for the self-organization and rotation of the grains observed here, thus increasing the energy of the grain boundaries proportionally to their misorientation angle and excess volume. This behavior is typical for deformation processes of bcc metals and intermetallic compounds at high strain rates. However, it is surprising that nominally brittle materials, such as intermetallics, develop considerable ductility under shear conditions.

4.3.3 Property–Microstructure Relationships Decreasing the grain size of a material to the nanometer range leads to a drastic increase of the number of grain boundaries, reaching typical densities of 1019 interfaces per cm3. The large concentration of atoms located in the grain boundaries in comparison with the crystalline part scales roughly as a reciprocal grain size dependence, 1/d. Consequently, due to their excess free volume, the grain boundaries in nanocrystalline materials can cause considerable differences in the physical properties of the materials compared with the conventional polycrystalline counterpart.

4.3.3.1 Thermal Properties As a result of the cold work, energy is stored in the powder particles. During heating in the process of differential scanning calorimetry (DSC), a broad exothermic reaction is generally observed. Integrating the exothermal signals gives the energy release ΔH during heating of the sample.

136

Processing

For example, the stored enthalpy reaches values of 7.4 kJ/mol (after 24 hours) and 10 kJ/mol (after 32 hours) for Ru, which corresponds to 30–40% of the heat of fusion ΔHf. One would expect the recovery rates during the milling process to correlate with the melting point of the specific metal. With the exceptions of Co (due to a large number of stacking faults) and Hf, Nb, and W (possibly due to an increased level of Feimpurities from the milling tools stabilizing the nanostructure), such a relationship is indeed observed. Similar results have been obtained for metals with fcc structure as well.[89] Consequently, the most effective energy storage occurs for metals with melting points above 1500 K, resulting in average grain sizes between 6 nm (Ir) and 13 nm (Zr). For the compound phases, similar high values for the stored energies are found ranging from 5 to 10 kJ/mol and corresponding to values between 18% and 39% of the heat of fusion for grain sizes between 5 and 12 nm. The final energies stored during mechanical attrition largely exceed those resulting from conventional cold working of metals and alloys (cold rolling, extrusion, etc.). During conventional deformation, the excess energy is rarely found to exceed 1–2 kJ/mole and, therefore, is never more than a small fraction of the heat of fusion.[90] In the case of mechanical attrition, however, the energy determined can reach values typical for crystallization enthalpies of metallic glasses corresponding to about 40% of ΔHf. A simple estimate demonstrates that these energy levels cannot be achieved by the incorporation of defects that are found during conventional processing. In the case of pure metals, the contribution of point defects (vacancies, interstitials) can be safely neglected because of the high recovery rate at the actual processing temperature. Even taking into account nonequilibrium vacancies, which can form as a consequence of dislocation annihilation up to concentrations of 10−3,[91] such contributions are energetically negligible in comparison. For intermetallics, on the other hand, point defects are relevant in order to describe the stability of the material. The maximum dislocation densities that can be reached in heavily deformed metals are less than 1016 m−2, which would correspond to an energy of less than 1 kJ/mol. Therefore, it is assumed that the major energy contribution is stored in the form of grain boundaries and related strains within the nanocrystalline grains that are induced through grain boundary stresses. Large differences generally also arise in the specific heat cp at constant pressure. The specific heat of the heavily deformed powder particles was measured in the range from 130 K to 300 K, i.e., at low enough tempera-

4: Solid State Processing, Fecht and Ivanisenko

137

tures to prevent the recovery processes from taking place. For all samples, a considerable increase in cp has been found experimentally after 24 hours of milling, reaching values of 15% for Ru. For pure metals, a linear correlation is observed between the increase of the heat capacity Δcp at 300 K and the stored enthalpy ΔH given as a proportion of the heat of fusion (ΔH/ΔHf) after extended mechanical attrition (see Fig. 4.10). Such a relationship is also predicted by the free volume model for grain boundaries.[92] This energetic microstructure–property relationship is further emphasized by Fig. 4.11. Here the stored enthalpy ΔH in attrited Fe powder is shown as a function of average reciprocal grain size 1/d, since 1/d scales with the volume density of grain boundaries in the nanocrystalline material (∼3δ/d where δ is the thickness of the grain boundary[93]). Two different regimes can be clearly distinguished: for small grain size reductions at the early stages of mechanical attrition, i.e., stage (i), the stored enthalpy shows only a weak grain size dependence typical for dislocation controlled deformation processes. After the average domain (grain) size is reduced below d * = 30–40 nm, energy storage becomes more efficient. The critical grain size d * corresponds to the size of nanograins that are formed within the shear bands. Therefore, for d < d* a regime can be identified where deformation is controlled by the properties of the small-angle and

Figure 4.10 Specific heat increase Δcp (%) in comparison to the unmilled state at room temperature as function of the stored enthalpy ΔH (given as percentage of ΔHf) after 24 hours ball milling of pure elemental powder samples.

138

Processing

Figure 4.11 The stored enthalpy ΔH as function of reciprocal grain size 1/d of Fe at different levels of mechanical attrition. Two distinctively different stages can be observed: stage (i), which is dislocation controlled for d > 40 nm; and stages (ii),(iii) for d < 40 nm, where deformation becomes controlled by grain boundaries.

later high-angle grain boundaries that are developing in stages (ii) and (iii). The slope of the corresponding strain versus 1/d relationship corresponds in the grain boundary regime to 0.1 nm, typical for atomic-level strains.[52]

4.3.3.2 Mechanical Properties As a further consequence of the grain size reduction, a drastic change in the mechanical properties has been observed (for reviews see Refs. 94 and 95). In general, nanocrystalline materials exhibit very similar mechanical behavior to amorphous materials due to shear banding as the prevalent mechanism of deformation.[96] As such, strain hardening is not observed and thus conventional dislocation mechanisms are not operating. The lack of dislocations is the result of the image forces that act on dislocations near grain boundaries. Local mechanical properties can be measured by nano-indentation methods. Here the load as well as the indentation depths are monitored continuously during the loading and unloading process (Fig. 4.12). Typical results for nanocrystalline Fe powder samples exhibit an increase in hardness by a factor of 7 (9 GPa for nx-Fe with d about 16 nm versus 1.3 GPa for annealed px-Fe). In general, the hardness follows a trend

4: Solid State Processing, Fecht and Ivanisenko

139

Figure 4.12 Hardness measurement using a nano-indentation device on polycrystalline (upper curves) and nanocrystalline (lower curves) attrited iron powder samples.

Figure 4.13 Hall–Petch relationship for the hardness of nx- and px-iron.

similar to the Hall–Petch relationship, though the dislocation-based deformation mechanism in the nanocrystalline regime certainly does not apply, as shown in Fig. 4.13. The Young’s modulus can also be measured by this method and shows a decrease by typically about 10% in comparison with the polycrystal. Given this, it is suggested that the mechanical properties of nanophase materials prepared by mechanical attrition after extended periods of milling are no longer being controlled by the plasticity of the crystal due

140

Processing

to dislocation movement but rather by the cohesion of the nanocrystalline material across its grain boundaries. From the considerable increase of hardness and the principal changes of the deformation mechanisms, improved mechanical properties can be expected as attractive features for the design of advanced materials as bulk or layered materials.

4.4 Phase Stability at Elevated Temperatures As a result of the cold work, considerable energy is stored in the powder particles (see Figs 4.10 and 4.11). Therefore, thermodynamically these materials are far removed from their equilibrium configuration and a large driving force toward equilibrium exists. On the other hand, for a number of applications the nanocrystalline powder has to be compacted to a uniform bulk sample. Since cold compaction is usually not sufficient, hot isothermal pressing (HIP) processes[97,98] are usually applied. It is thus important to understand the atomic processes occurring during annealing over extended periods of time. The stored energy is released during heating to elevated temperatures due to recovery, relaxation processes within the boundaries and to grain growth. As a consequence, during annealing at elevated temperatures, relaxation and grain growth processes will occur, leading to a concomitant increase of the grain size. This behavior has been investigated in detail for iron.[99,100] For extended periods of milling time, a decrease of the average grain size to nanometer dimensions is observed with a stationary average grain size d = 16 nm and 0.7% microstrain, as shown in Fig. 4.2. The enthalpy release during a DSC heating experiment spreads over the entire temperature range of the scan, as shown in Fig. 4.14. The very broad signal does exhibits no distinct events but a further increase of the exothermic signal for T > 250–300°C. X-ray diffraction of powder samples annealed for 80 min at each temperature revealed the evolution of grain size and strain as function of annealing temperature as shown in Fig. 4.15. The microstrain is decreasing rapidly below 200°C, while the grain size remains nearly constant. As such, the enthalpy release during the first exotherm in Fig. 4.14 is only related to relaxation and not to grain growth. Grain growth starts to become significant above about 300°C. Furthermore, it has been found that after a fast increase at early times the average grain size d changes from 16 nm to about 30–40 nm. The average grain size remains constant for t ≥ 2400 s and reaches values of 100–200 nm at temperatures of about 600°C.

4: Solid State Processing, Fecht and Ivanisenko

141

Figure 4.14 DSC heating scan at 10 K/min of iron powder after mechanical attrition for 5 and 25 hours.

Figure 4.15 Dependence of stationary grain size and microstrain on annealing temperature.

As such, two regimes, with and without grain growth, can be distinguished. Since the influence of lattice point defects and lattice dislocations is negligible, the enthalpy release can be clearly assigned to the existence of grain boundaries. The reduction of the microstrains is probably caused by grain boundary relaxation and annihilation of secondary grain boundary dislocations. From elastic theory it is estimated that this

142

Processing

contribution to the overall energy is less than about 5%. On this basis, the grain boundary energy can be estimated. From simple geometric considerations[101,102] the specific grain boundary excess enthalpy is estimated to be about 2.1 J/m2 This would correspond to a value for nonequilibrium unrelaxed grain boundaries, whereas after relaxation the grain boundary energy is reduced to 1.5 J/m2. Values resulting from computer simulations suggest excess enthalpies between 1.2 and 1.8 J/m2.[103] We therefore conclude that grain boundaries in the as-prepared state are characterized by increased values of about 25% due to their unrelaxed atomic structure or secondary grain boundary dislocations. Isothermal DSC measurements allow further analysis of the grain growth processes in nanocrystalline Fe. For example, the isothermal DSC curve shown in the upper part of Fig. 4.16 was measured at 500°C after annealing the sample at 400°C and heating to 500°C at a rate of 50°C/min. A monotonically decreasing signal typical for grain growth is observed. Similar signals are obtained at 200°C, 300°C, and 400°C and clearly differ from those measured in isothermal recrystallization processes controlled by nucleation and growth in conventional polycrystalline metals, which are described by Johnson–Mehl-Avrami type models. Figure 4.16 does not exhibit the expected maximum related to an incubation time for nucleation, but shows only a decrease in the signal. Furthermore, (dH/dt)−2/3 should scale linearly with time if normal parabolic grain growth behavior is assumed. This assumption is well approx-

Figure 4.16 Isothermal exothermic DSC curve at 500°C of nanocrystalline iron (upper part) and plot of (ΔH/dt)−2/3 versus time t (lower part).

4: Solid State Processing, Fecht and Ivanisenko

143

imated for t < 1200 s as shown in the lower part of Fig. 4.16. The upper part of Fig. 4.16 includes a fit to the measured DSC signal assuming parabolic grain growth. Based on the available data, an activation energy of 178 kJ/mol has been determined for grain growth in nanocrystalline Fe, which is comparable to the activation energy for grain boundary selfdiffusion in Fe.[104]

4.5 Severe Plastic Deformation (SPD) 4.5.1 General Similar observations regarding the deformation mechanisms during mechanical attrition have been reported in chips removed during machining,[105] simple metal filings,[106,107] and also extreme deformation of bulk materials.[108,109] In analogy to the mechanically attrited powder at the early stage, large inhomogeneities have been observed in the filings, with the deformation process leading to the formation of small-angle grain boundaries. Here, the dislocation cell size dimensions are basically a function of the acting shear stress τ resulting in an average cell size dimension L = 10 Gb/τ, with G the shear modulus and b the Burgers vector.[110] More detailed studies are obtained from cold rolling and torsion,[111] wire drawing,[112] and cyclic deformation[113] processes resulting in an asymptotic saturation of the flow stresses. This is considered to be a result of the simultaneous occurrence of dislocation multiplication and annihilation, leading to a saturation of the dislocation density. In particular, under cyclic deformation of, for example, Cu at amplitudes above γpl ∼ 10−4, slip becomes highly localized in so-called persistent slip bands (shear bands). These lie parallel to the primary glide plane and are separated by regions containing the original matrix structure. These bands consist of dense walls of dislocations, largely screw dislocations having a density of ∼1013 m−2. The closest spacing between screw dislocations of opposite sign is ∼50 nm, the minimum distance before annihilation occurs. For edge dislocations, which are more relevant for the deformation of fcc crystals, this critical annihilation length is found to be 1.6 nm for Cu. It has been concluded that the annihilation of dislocations can set a natural limit to the dislocation densities that can be achieved by plastic deformation (typically less than 1013 m−2 for screw dislocations and 1016 m−2 for edge dislocations). Steady state deformation is observed when the dislocation multiplication rate is balanced by the annihilation rate. This situation corresponds to the transition of stage (i) to stages

Processing

144

(ii)/(iii) as described above. In this stage the role of dislocations becomes negligible and further deformation occurs via grain boundaries. It is expected that the shear modulus of the grain boundary regions will be lowered by about 40% when the “volume-fraction” of the grain boundaries becomes comparable to that of the crystals.[114,115] Localized deformation then proceeds by the dilation of the grain-boundary layers similarly to superplastic behavior,[116] with the undeformed crystallites moving in a “sea” of dilated grain-boundaries. Furthermore, the relative motion of the crystalline grains within the shear band leads to impingement, which should give rise to large, locally inhomogeneous elastic stresses. As a consequence, in order to relax these strains, formation of nanovoids about 1 nm in diameter is expected to occur, which inevitably leads to crack formation under tensile stress.[117] Such a deformation mode basically also provides a mechanism for the repeated fracturing and rewelding of the fresh surfaces during mechanical attrition, leading to a steady-state particle size.

4.5.2 High-Pressure Torsion Among the methods using mechanical means for microstructure refinement and phase composition modification in bulk materials, the most perspective are those that combine shear deformation and high imposed pressure. Application of high pressure prevents crack formation and fracture of deformed samples, so that practically unlimited strains can be attained. Techniques such as equal channel angular pressing[118] and high pressure torsion (HPT)[119] were evolved about 25 years ago, and since that time they have found very wide application in production of bulk nanostructured metals with grain size in the nanometer or submicrometer ranges [for a review see Ref. 120]. In the case of multiphase alloys and intermetallic compounds, these methods of severe plastic deformation not only lead to grain refinement but can also induce the formation of nonequilibrium solid solutions,[121,122] disordering,[123,124] or amorphization.[125,126] High-pressure torsion is the simplest method of severe plastic deformation (SPD). In this technique the sample is placed between two Bridgman anvils: the upper fixed and the lower rotatable as shown in Fig. 4.17(a). Various strains can be achieved by rotating the lower anvil through a defined angle. The shear strain γ may be simply estimated as γ = 2πRN/d

Eq. (4-1)

4: Solid State Processing, Fecht and Ivanisenko

145

Figure 4.17 (a) Schematic diagram of the high-pressure torsion process; (b) Schematic diagram explaining the geometry of shear deformation in high pressure torsion (HPT).

where R is the distance from the sample center, N is the number of anvil rotations, and d is the thickness of the sample. The simple shear stress conditions are realized at relatively low temperatures under high applied pressure (5–15 GPa) in this method of microstructure refinement (Fig. 4.17(b)).

4.5.2.1 Microstructure Development at High-Pressure Torsion Using high-pressure torsion the first nanocrystalline structures were obtained in single-crystal specimens of Ni and Cu.[119] It was found that the mean grain size had gradually decreased with increasing shear strain and had finally stabilized at the level of ∼100 nm. Further experiments with initially coarse-grained Cu,[127,128] Ni,[129] Fe,[130,131] Cr,[132] Mo,[133] and Ti[134] revealed that strong grain refinement can be achieved in all these metals after 3–5 revolutions(rotations), whereupon the mean grain size usually reaches the steady-state value of 100–200 nm depending on the material (Table 4.1). Somewhat smaller grain size had been reported for single-phase and matrix Fe and Al alloys,[135,136] and significantly smaller grain size, down to 10 nm, can be obtained for some multiphase alloys[137] and intermetallic compounds.[123] In early investigations[119,125,130,133,134] the conclusion about the formation of high-angle misorientations, i.e., granular type of microstructure, in HPT-processed materials was based on the inspection of selected-area electron diffraction (SAED) patterns; that is, it was qualitative. In that case, SAED patterns having many spots randomly distributed along the Debye–Sherrer rings were regarded as evidence for the formation of an

Processing

146

Table 4.1 Mean grain size obtained in several pure metals after highpressure torsion (HPT) deformation

Metal Ni Cu Cu Fe Fe Cr Mo Ti

Details of HPT Processing*

Grain Size (nm)

Reference

P = 6 GPa, γ = 300, RT P = 5 GPa, γ = 300, RT P = 8 GPa, γ = 250, RT P = 5 GPa, γ = 600, RT P = 5 GPa, γ = 420, RT P = 7.8 GPa, γ = 120, T = 623 K P = 6 GPa, γ = 300, T = 623 K P = 5 GPa, γ = 600, RT

170 nm 200 nm 200 nm 100 nm 127 nm 500 nm

[129] [127] [128] [130] [131] [132]

190 nm

[133]

120 nm

[134]

* P is the applied pressure; γ is the shear strain estimated for the corresponding number of rotations using Eq. (4-1); T = temperature; RT = room temperature.

ultrafine grained structure. The development of orientation imaging microscopy (OIM) on the basis of modern FEG-SEM microscopes equipped with an electron backscattered diffraction (EBSD) set-up allows for the quantitative analysis of the misorientations of large ensembles of grain boundaries in SPD materials. Recently, this technique was applied to HPT-processed Cu,[128] Ni,[129] and Fe.[131] These investigations have confirmed that typical misorientations between grains are of high-angle type, and that the mean grain size measured in dark-field TEM images corresponds well to the mean distance between the grain boundaries with misorientation angle ≥15° determined using OIM. At the same time, a large amount of low-angle boundaries (cell walls) was detected inside new grains; for example, the fraction of low-angle boundaries varies from 50% in Fe[131] to 20% in Cu[128] and Ni.[129] HPT-processed materials demonstrate high levels of internal stresses as measured by XRD methods;[127,138] however, this level is lower than that in ball-milled counterparts. As already mentioned, typical atomic-level strains in ball-milled materials can increase up to 1%,[32] but only to 0.4% in HPT-Fe[138] and to 0.12% in HPT-Cu.[127] It was concluded that the grain boundaries resulting from SPD are the main sources of the high internal stresses in HPT materials, because these boundaries are in nonequilibrium

4: Solid State Processing, Fecht and Ivanisenko

147

due to the high density of defects, such as intrinsic grain boundary dislocations[139] and disclinations.[140] As an example of typical microstructure evolution occurring in pure metals and single-phase alloys during high pressure torsion, the results of investigation of Armco iron (a commercial purity iron: 99.7 wt% Fe) under a pressure of 5 GPa at a wide range of shear deformation, γ = 15–630,[130,131] may be considered. It was found that microstructure evolution in HPT of initially coarse-grained iron (d = 30 μm) starts with formation of cellular structure with cells elongated in the shear direction (Fig. 4.18(a)). These cells are delineated by dislocation boundaries with high dislocation density in directions parallel to the shear direction and with lesser dislocation density in directions perpendicular to this direction. High-angle misorientations appear after shear deformations of γ > 150. The size of new grains is about 300–500 nm for the intermediate shear strains (γ = 150), and gradually decreases with increase of shear strain. The mean grain size after very large deformations (γ = 630) is ∼100 nm (Fig. 4.18(b,c)). The grain size distribution in HPT-processed iron is of log-normal type and it becomes narrower with increasing strain, indicating the homogenization of the structure. Analysis of the misorientation spectra of large ensembles of grain boundaries in Armco-iron HPTprocessed up to different shear strains by means of OIM has demonstrated that the misorientation spectra in HPT iron do not change significantly with increase of shear strains. They always have a bimodal character with maxima at low (1–2°) and high misorientation angles ranges (Fig. 4.19).[131] The first maximum is related to the formation of a network of low-angle boundaries with misorientation angle <3° inside new grains (Fig. 4.20). Due to the high density of low-angle boundaries, the fraction of high-angle boundaries in the as-processed material is 50%. Therefore, the structures formed during HPT over a wide strain range are similar and demonstrate scaling behavior, analogously to what was shown for microstructures formed in cold rolled materials.[141]

4.5.2.2 Mechanism of Microstructure Refinement in HPT Observations of microstructure evolution during HPT by means of TEM and OIM confirmed in general the model of microstructure refinement developed for ball-milled materials on the basis of neutron scattering experiments (section 3.3.2). Indeed, the structure evolution at severe plastic deformation begins with formation of narrow areas with high dislocation density (shear bands) delineating areas with low dislocation

148

Processing

Figure 4.18 Different types of microstructure formed in Armco iron after HPT straining under a pressure of 5 GPa. (a) Cellular structure with cells elongated in shear direction; shear strain γ = 90. (b,c) Granular structure, shear strain γ = 420. (a,b) Bright-field TEM images; (c) dark-field TEM image. Selected area diffraction pattern in shown in (b) as insert.

density (cells) (Fig. 4.18(a)). The permanent storage of dislocations in cell boundaries with increasing deformation leads to an increase of their misorientations. In principle, the increase in misorientation angle due to storage of dislocations in a boundary is not unlimited. From an upper bound on the dislocation density as 1013–1014 cm−2,[142] the misorientation angle θ can be estimated roughly[143] as

4: Solid State Processing, Fecht and Ivanisenko

149

Figure 4.19 Grain boundary distribution as represented by misorientation angles in Armco iron after HPT straining for γ = 210 and γ = 420.

Figure 4.20 Spatial distribution of different boundary types in HPT processed Armco iron (γ = 420). Areas that are not indexed are shown in darkest gray. It is seen that high-angle boundaries (shown in black) delineate almost equiaxial volumes (grains), and low-angle boundaries (shown in gray) are distributed inside grains.

θ = b/h = b√ρ

Eq. (4-2)

where b is the magnitude of the Burgers vector, h is the dislocation spacing in the grain boundary, and ρ is the dislocation density in the grain boundary. Typical dislocations in iron are 1/2[111] screws, b = 1.43 Å, and θ = 0.143 rad or 8° for ρ = 1014 cm−2. Achievement of such dislocation density in a cell wall corresponds to the beginning of the second stage of

150

Processing

Figure 4.21 Bright field TEM image of the microstructure of Armco iron after HPT deformation for γ = 210 showing large grain divided into cells. After Ref. 131.

microstructure evolution, namely, relaxation and rearrangements of the dislocation structure in grain boundaries and grain rotations with consequent increase of misorientations as discussed on p. 135. An example of such a process can be seen clearly in Fig. 4.21. An inspection of the large grain in the bottom right corner shows that it is divided into small subgrains/cells. One of these in the middle demonstrates a notably different diffraction contrast from its neighbors. It seems that the boundary misorientation has increased during the deformation and apparently at a more rapid rate than with other cells. After the transformation of a cell boundary into a grain boundary, the creation of low-angle boundaries inside this new grain starts again and the whole process is repeated, as is schematically represented in Fig. 4.22. This tendency leads to a decrease of the mean distance between the high-angle boundaries and to the formation of a narrower grain size distribution. It seems that stage 3, or complete homogenization of the microstructure, cannot be achieved in HPT experiments as, even after very large strains (γ > 420), nanocrystalline grains still contain small subgrains.

4.5.2.3 Metastable States in HPT-Processed Alloys Generally, high-pressure torsion leads to significant alteration of the initial phase composition of multiphase alloys and formation of nonequilibrium solid solutions, atomic disordering in intermetallic compounds,

4: Solid State Processing, Fecht and Ivanisenko

151

Figure 4.22 Schematic diagram explaining the process of grain size refinement during severe plastic deformation.

and polymorphic phase transformations. Approaches developed for the mechanical alloying process[38–43] are applicable for understanding the driving forces and mechanisms of the transformations taking place at HPT. We shall discuss several examples of mechanically driven phase transformations observed at HPT.

4.5.2.3.1 Intermetallic Compounds In a similar way as shown for ball milling (p. 122), brittle intermetallic compounds can be easily deformed by means of HPT. Typically formation of nanocrystalline structure and disordering,[123,144,145] or even amorphization,[146–148] have been observed. For example, microstructure evolution in Ni3Al, an intermetallic compound with an L12 structure which remains ordered up to the temperature of melting, has been studied in detail.[123,144] Those workers found that at the beginning of straining, there occurred formation of blocks and bands containing misoriented fragments in the 100 nm range separated by a three-dimensional array of low-angle boundaries. The inhomogeneous deformation led to localized disorder and a high density of dislocations. Further, dynamic recovery caused a heterogeneous formation of the nanocrystalline structure with a mean grain size of 25 nm. Formation of nanocrystalline structure had been accompanied by complete disappearance of long-range chemical order. Obviously, the microstructure evolution of intermetallics at HPT is morphologically similar to that observed for pure metals.

152

Processing

It is interesting to note that HPT of intermetallic NiTi alloy leads to amorphization.[146–148] The tendency to form the amorphous structure under SPD conditions depends on the initial structure of the alloy: it is strongest when the initial structure is martensitic, and weakest when the initial structure consists of stable austenite.[147] This indicates that the degree of initial lattice imperfection plays an important role in this process.

4.5.2.3.2 Dissolution of Precipitates Severe plastic deformation by high-pressure torsion leads to dissolution of carbides in steels,[149,137] oxides,[150] and precipitates in alloys.[125,151] On the other hand, decomposition of oversaturated solid solutions may also take place.[152] Decomposition of iron carbide cementite was studied in commercial carbon steel UIC 860V (Fe–0.8 wt.% C–1.3 wt.% Mn) with fully pearlitic structure subjected to HPT deformation under a pressure of 7 GPa in a shear strain range of γ = 60–420.[137] The amount of cementite in the HPT-processed steel, estimated using thermomagnetic methods, decreased with increasing of shear strain. Finally, total cementite dissolution was observed together with formation of uniform nanocrystalline structure in ferrite with a mean grain size of 10 nm (Fig. 4.23). Precise determination of the lattice parameters of ferrite using the synchrotron XRD method has not revealed any significant changes after HPT deformation. It was

Figure 4.23 Microstructure of UIC 860V steel after HPT straining for γ = 420: (a) bright-field image and diffraction pattern; (b) dark-field image. TEM. After Ref. 137.

4: Solid State Processing, Fecht and Ivanisenko

153

Figure 4.24 (a) 3D reconstruction of an analyzed volume (5 × 5 × 46 nm3) in pearlite processed by HPT (5 turns). Only carbon atoms are plotted to show plain segregation of carbon along the grain boundary and linear along dislocations. (b) Carbon concentration profile computed from the left to the right across the volume shown in (a). Courtesy of Dr. X. Sauvage.

suggested that carbon is located in areas where it does not cause lattice expansion and does not lead to a change of lattice parameter, i.e., in grain boundaries and dislocations cores. This conclusion has recently been confirmed in 3D atom probe experiments, when plain carbon segregations along nanocrystalline boundaries and linear segregations along dislocations were observed as shown in Fig. 4.24. Similar results were also obtained in experiments with ball-milled pearlitic steel.[153] Very important information about the mechanism of cementite decomposition was obtained thanks to direct determination of chemical compositions of the HPT-processed steel by means of 3D atom probe techniques.[154] In these experiments, areas with different carbon contents were revealed and analyzed. Surprisingly, it appears that cementite decomposition includes not only refinement of the cementite particles and reduction of their amount, but also dramatic change of carbon content in cementite. As such, nano-sized areas with carbon content of only 4 at.% have been revealed. These areas still have the lamellar shape, which indicates that they represent the former cementite lamellae (Fig. 4.25). Decomposition of precipitates at HPT is a special case of mechanical alloying, and approaches developed to explain dissolution of cementite at cold plastic deformation can be extended to similar processes occurring in other alloys. Two approaches are discussed in the literature to explain the cementite dissolution. One of these is based on thermodynamic equilibrium considerations in metastable phases;[155,156] the other is based on ballistic models in which movement of matter is caused by external factors.[40,137,157] Obviously, the degree of thermodynamic instability of the cementite particles increases in the course of straining. This is caused by

154

Processing

Figure 4.25 (a) 3D reconstruction of an analyzed volume (10 × 10 × 37 nm3) in pearlite processed by HPT (5 turns). Only carbon atoms are plotted to show three carbon-rich lamellae. (b) Carbon concentration profile computed from the left to the right across carbon rich lamellae shown in (a). The carbon concentration of these lamellae is only about 4 at.%. Courtesy of Dr. X. Sauvage.

two factors: their very small size, so that capillarity effects become important,[45,158] and accumulation of strains in the precipitates.[159] For small precipitates, the relaxation of interface energy and interface stresses provides a high driving force for their dissolution, because dissolution of precipitates leads to relaxation of these strain fields. According to the above considerations, dissolution of precipitates should also take place after the deformation process is stopped. However, it seems that this process is observed only during the deformation, and not when the deformation process is stopped. It therefore seems that the flow of matter during plastic deformation plays a major role in the dissolution of precipitates. Such an effect can be explained on the basis of models that account for the movement of dislocations as a key factor in this process. In one of these it was proposed that the dissolution of the precipitates takes place due to drag of carbon by dislocations crossing the precipitates.[157] On the other hand, for inhomogeneous distribution of the alloying atoms the distance between them will increase with each dislocation slip. These processes can lead to the formation of so-called driven alloys[40] with compositions far from equilibrium. In the particular case of deformation of pearlitic structure, the shear stresses applied exceed the yield stresses of all phases involved: the ferrite, cementite, and the nanocrystalline ironcarbon alloy resulting from cementite dissolution and heavy plastic deformation. However, the cementite is about 3 times harder than the ferrite matrix. It was therefore proposed to regard the ferrite matrix as a viscous fluid driven by the high shear stress to flow around the harder cementite plates.[137] The friction at the precipitate–matrix interface leads to two

4: Solid State Processing, Fecht and Ivanisenko

155

Figure 4.26 Schematic diagram showing the normal to the axis of rotation, cross-section of the cementite lamella, and explaining the cementite dissolution in the different stages of deformation. (a) Initial stages of straining before formation of cells (N = 1): jump of the carbon atom across the interface boundary (1) leads to its trapping and drag by dislocations (2) or drift together with ferrite (3). (b) Nanocrystalline structure (N = 3 . . . 5): jump of the carbon atom across the interface boundary (1) and rotation of ferrite grains lead to the transportation of C atom away from the ferrite–cementite interface (2). After Ref. 137.

effects. One is an accumulation of high strain energy in the carbides, which may strongly contribute to their thermodynamic instability; and the second is their wear due to the flow of the ferrite as shown schematically in Fig. 4.26.

4.5.2.3.3 Stress-Induced Phase Transitions In conditions of high external stresses realized at high-pressure torsion, the formation of various high pressure- and high-temperature phases has been observed in nanocrystalline metals and alloy. For example, the phase transitions bcc→hcp, bcc→fcc and fcc→hcp were observed in Fe-Ni and Fe-Mn alloys at HPT deformation under a pressure of 25 GPa;[160] the HPT processing under a pressure of 5 GPa promoted the partial transformation of a′ (bcc) martensite to ε (hcp) martensite in AISI 304 steel[161] and martensitic transformation of metastable austenite in stainless steel 18 wt% Cr, 10 wt% Ni.[135]

156

Processing

The most remarkable example of such stress-induced phase transformation is a reverse martensitic transformation of nanocrystalline ferrite at HPT under a pressure of 7 GPa, resulting in formation of high-temperature iron phase austenite.[162] Nanocrystalline ferrite with a grain size of 10 nm has been produced by HPT of pearlitic steel with shear strain γ = 420. As was discussed in the previous section, this processing leads to practically Figure 4.27 Zero-loss filtered complete dissolution of cementite and forselected-area diffraction mation of the uniform ferritic structure with pattern of HPT-processed pearlitic steel showing mainly carbon segregations along grain boundferrite reflections, though aries. Figure 4.27 demonstrates the SAED cementite (Fe3C) and γ-Fe pattern of an as-processed specimen, where (austenite) reflections are together with strong Debye–Scherrer rings also visible and some of of bcc ferrite, many spots of fcc austenite these are arrowed and can be clearly seen. A high-resolution TEM indexed. After Ref. 162. image of austenitic grains is presented in Fig. 4.28. To understand why this nonequilibrium displacive transformation occurs when the nominal martensitic transformation temperature is several hundred kelvins above room temperature, a number of factors affecting the phase equilibrium between ferrite and austenite should be considered. As was shown for mechanically attrited materials (see p. 141), HPT-processed nanocrystalline ferrite is energetically a high-enthalpy state due the effects of nanograin boundaries. This has been demonstrated by DSC heating measurement[163] (Fig. 4.29), which showed that stored enthalpy in HPT-processed specimen amounts to 2.5 kJ/mol; this corresponds to more than 60% of the difference between the Gibbs free energies of austenite and ferrite under normal conditions. Thus, the free energy difference between the two phases is much reduced under nanocrystalline conditions making any such transformation easier to trigger. Secondly, external stresses can stimulate martensitic transformations. Indeed, martensitic transformations are accompanied by dilatational and shear deformation of the parent phase, that is why external normal and shear stresses will aid transformation in parent grains favorably oriented with respect to the action of these external strains. In the case of HPT deformation, strong shear stresses of ∼2 GPa act in combination with quasihydrostatic pressure of 7 GPa to provide the necessary driving force for reverse martensitic transformation.[162]

4: Solid State Processing, Fecht and Ivanisenko

157

Figure 4.28 High-resolution TEM image of the boundary between nanoscale ferrite and austenite grains showing Moiré fringes; fast Fourier transforms of the two grains are shown together with indexed schematics thereof; a FFT of the boundary region is also shown to demonstrate the orientation relationship of the two grains. After Ref. 162.

4.5.2.3 Consolidation of Powders and Synthesis of Nanocomposites High pressure torsion can be effectively applied for consolidation of nanocrystalline and amorphous powders prepared by ball milling, rapid quenching, and other techniques. The grain size in the resulting compacts is strictly dependent on the grain and particle size of the initial powders, and Figure 4.29 DSC curves of UIC 860 steel after HPT deformation. The also on the consolidation condi- integrated total enthalpy release is tions. For example, after room 44.2 J/g for the temperature range temperature HPT consolidation of 100–500°C. After Ref. 163. cryomilled Al-7.5 wt.% Mg powder, bulk nanocrystalline specimens with 97% relative density and 30 nm grain size have been obtained.[164] HPT processing of powder blends of metal and metal oxides results in formation of nanocomposites consisting of nanocrystalline matrix (d ∼ 50–70 nm) and embedded oxide particles.[150,165,166] In some

Processing

158 b

Figure 4.30 The microstructure of bulk nanocrystalline Fe–1 wt.% C compacts after HPT consolidation at 540°C: (a) bright-field images; (b) selected-area diffraction pattern; (c) dark-field image in the (311) Fe3O4 reflection showing oxide particles distributed mainly along grain boundaries. After Ref. 167.

cases when the starting powder is so hard that its room temperature consolidation is impossible, this consolidation can be performed at elevated temperatures and in vacuum to prevent oxidation. For example, after HPT processing of a ball-milled blend of powders of elemental Fe and C (carbon black) at 540°C, bulk specimens with composition of Fe-1 wt% C and relative density of 96% have been obtained.[167] TEM investigations had shown that the microstructure of the compacts consists of nanocrystalline ferritic matrix with grain size of 70 nm and fine carbide and oxide particles distributed basically along the grain boundaries (Fig. 4.30). Thanks to these particles, the grain growth was prevented until the temperature of 700°C was achieved! Recently, HPT was applied to produce bulk specimens from very thin ribbons of amorphous Fe77Al2.14Ga0.86P8.4C5B4Si2.6 and Al88Y7Fe5 alloys obtained by rapid quenching.[168,169] It was shown that after HPT consoli-

4: Solid State Processing, Fecht and Ivanisenko

159

dation under a pressure of 6 GPa, fully dense samples can be obtained. Additionally, the torsion straining process induced significant changes in the short-range order (both chemical and topological) of the metallic glass, which resulted in the formation of dispersed nanocrystallites inside the amorphous matrix. Thus, HPT deformation opens wide perspectives for the consolidation of nanopowders and for the synthesis of bulk nanostructured composites of various types (metal–ceramic, amorphous–crystalline) that cannot be produced using conventional processes.

4.5.3 Cold Rolling of Thin Sheets An alternate route to producing samples with high levels of both plastic deformation and interfacial area is by cold rolling of layered elemental sheets that are folded between each deformation cycle. In this case, the large increase in interfacial area is created internally with absolutely negligible contamination. Similarly, in contrast to mechanical attrition, the uncertainty in the temperature during processing is removed since the sample is in firm contact with the massive rolls and deformation can be performed at a low strain rate to maintain ambient temperatures. This approach has been used to examine amorphous phase formation in several binary alloys such as Zr—Ni,[170] Cu—Er[171] and Al—Pt,[172] and also for the preparation of bulk Fe/Ag nano-multilayers with giant magnetoresistance.[173] In earlier work on amorphous phase formation, deformation rates in excess of 1 s−1 were employed and some annealing was needed to complete the amorphization reaction. Recently, fully amorphous foils of a multicomponent Zr65Al7.5Cu17.5Ni10 alloy have been synthesized at ambient temperatures from a layered array of individual elemental sheets by repeated low-strain-rate (0.1 s−1) cold rolling.[174] Figure 4.31 shows X-ray diffraction patterns from the Zr—Al—Ni—Cu foils taken after (a) 10, (b) 80, and (c) 120 deformation cycles. Each deformation cycle consisted of rolling the multilayer sandwich to a thickness of approximately 80 μm and subsequent folding. High-resolution TEM analysis exhibits further evidence that a true amorphous phase has been formed, as shown in Fig. 4.32. The X-ray spectrum of the amorphous sample is basically identical to that of a metallic glass produced by liquid quenching with the same composition. Thermal analysis of the cold rolled amorphous sample by DSC reveals a distinct glass transition at Tg = 647 K followed by a sharp exothermic crystallization peak at 745 K. It is worthy of note that very

160

Processing

Figure 4.31 XRD of cold-rolled thin foils of composition Zr65Al7.5Cu17.5Ni10.

Figure 4.32 High-resolution TEM image of the nanocrystalline/glass transition during cold rolling of thin foils of composition Zr65Al7.5Cu17.5Ni10.

similarly to amorphization reactions observed in mechanically alloyed Zrbased powder mixtures of similar composition (section 4.3.1.2), the initial stage of cold rolling is characterized by the dissolution of solute into Zr along with a reduction in grain size to about 30 nm before the onset of the crystal-to-glass transition. The formation of similar amorphous phases from two inherently different initial states, i.e., the solid and the liquid state, suggests that compositionally induced static disorder in a mechanically driven system can lead to the same final glass state that is conventionally derived from freezing out the dynamic disorder of a liquid to a glass.

4: Solid State Processing, Fecht and Ivanisenko

161

The present examples demonstrate that cold rolling offers an attractive alternative for the synthesis of nanostructured materials or multicomponent metallic glasses beside more traditional techniques such as mechanical attrition or liquid undercooling. Due to the relatively simple experimental conditions, size restrictions do not limit the formation of bulk samples.

4.5.4 Friction Induced Surface Modifications Many microscopic processes occurring during mechanical attrition and mechanical alloying of powder particles exhibit features in common with processes relevant in tribology and wear. For example, the effects of work hardening, material transfer, and erosion during wear situations result in similar microstructures of the wear surface to those observed during mechanical attrition.[175,176] In particular, during sliding wear, large plastic strains and strain-gradients are created near the surface. Typical plastic shear strain rates can here correspond to several 103 s−1. Close to the surface of wear scars, as well is in the wear debris of Cu, nanocrystalline structures have been observed by high-resolution electron microscopy with an average grain size of 4–5 nm.[177] No defects were observed within the interiors of the grains, suggesting that most of the defects are absorbed by the grain boundaries due to their proximity. However, this type of plastic deformation at high strain rates does not seem to be limited to metals and alloys[178] but has been observed in ceramics[179] and diamond[180] as well. During sliding wear, a special tribo layer develops on the surface of a sliding component being subjected to large plastic strains. This surface layer is often called the Beilby layer, and for a long time was thought to be amorphous because its microstructure could not be resolved with the instruments commonly used.[181] There are indeed some systems in which truly amorphous layers are produced by sliding,[182] but in most cases the subsurface layer with a thickness of several micrometers has a nanocrystalline structure. For example, during ultrasonic shot peening, the formation of nanocrystalline Fe-surfaces has been observed and the initially coarse-grained structure in the surface layer was refined into equiaxed ultrafine grains (about 10 nm) with random crystallographic orientation as shown in Fig. 4.33.[183,184] As a further example of technical relevance, the development of highspeed trains reaching velocities higher than 300 km/h poses a materials

162

Processing

Figure 4.33 Nanocrystalline iron produced by means of ultrasonic shot peening (courtesy of K. Lu).

challenge concerning the mechanical integrity and safety required for the railway tracks.[185,186] In particular, the interaction and slip between wheel and rail has been optimized and is controlled by sophisticated electronics, whereas the materials for the rail have not been changed for two decades. On the steel surface (Fe–0.8 at% C–1.3 at% Mn) where the local pressure typically exceeds 1.0–1.5 GPa, solid-state transformations have been observed that are caused by friction-induced shear forces and have strong similarities with mechanical attrition of powder samples. Corresponding XRD and TEM results indicate that the average grain size of the extremely deformed surface layer corresponds to about 20 nm, whereas a gradient in grain size is observed farther away from the surface reaching values up to 200 nm.[187] For example, Fig. 4.34 shows TEM photographs and the corresponding diffraction patterns of the initial pearlitic structure (a) in comparison with the nanocrystalline layer near the surface (b). As a consequence, hardness measurements have been performed using a nano-indenter at small loads as shown in Fig. 4.35. Steep hardness gradients have been found in cross-section with a lateral resolution of a few micrometers, which are clearly correlated with the change in microstructure. As a result, it is found that the hardness is increased from typical values for the pearlitic steel of approximately 2.5 GPa to 13 GPa next to the surface. This remarkable increase in hardness and mechanical strength of regions near the surface is clearly related to the fact that the average grain size is

4: Solid State Processing, Fecht and Ivanisenko

163

a

b Figure 4.34 TEM images of (a) the pearlitic structure of a low-alloyed Fe—C—Mn steel and (b) the nanocrystalline structure of an extremely worn surface of a high-speed railway track.

considerably decreased by the continuous deformation process. Similar results have been obtained for mechanically attrited α-Fe and α-Fe–Cpowder processed in ball-milling devices. Here also an increase of hardness by a factor 5 has been observed.[188] During mechanical attrition of rail filings (of identical composition to rail), a decrease in grain size to 7 nm after 50 hours of milling time has been observed together with a dissolution of the carbides. The same observation holds for the highly deformed

164

Processing

Figure 4.35 Hardness versus depth (distance from surface) for a high-speed railway track with a nanostructured surface due to extreme wear conditions.

rail surface, i.e., a dissolution of carbides and supersaturation of the α-Fe with carbon. However, from to the powder milling experiments it is obvious that the contribution to the increase in hardness results mainly from the grain size reduction and only partially from the highly strained martensite-like structure of the bcc iron supersaturated with carbon. Moreover, the wear resistance of the nanostructured areas is increased by a factor 2. Fretting wear measurements typical of the type of wear in wheel–rail contact reveal wear rates of 1.55 × 10−5 mm3 m−1 for the nanostructured layer and 3.77 × 10−5 mm3 m−1 for the undeformed surface. As such, the large improvement of the mechanical properties clearly shows the importance of the formation of nanostructures for technologically relevant wear problems.

4.6 Summary and Outlook The solid state processing methods of mechanical attrition and mechanical alloying have been developed as a versatile alternative to other processing routes in preparing nanoscaled materials with a broad range of chemical composition and atomic structure. In this process, lattice defects are produced within the initially single-crystalline powder particles. The internal refining process with a reduction of the average grain size by a factor of 103–104 results from the creation and self-organization of dislocation cell networks and the subsequent formation of small-angle and high-angle grain boundaries within the powder particles during the

4: Solid State Processing, Fecht and Ivanisenko

165

mechanical deformation process. As a consequence, changes of the thermodynamic, mechanical, and chemical properties of these materials have been observed, with the properties of nanophase materials becoming controlled by the grain size distribution and the specific atomic structure and cohesive energy of the grain or interphase boundaries. Such a transition from dislocation-controlled properties to grain boundary-controlled properties is expected for nanocrystalline materials synthesized by other methods as well.[189] Mechanical attrition offers interesting perspectives in preparing nanostructured powders with a number of different interface types in terms of structure (crystalline/crystalline, crystalline/amorphous) as well as atomic bonding (metal/metal, metal/semiconductor, metal/ceramic, etc.). Due to the broad range of possible atomic structures, very different properties in comparison with conventional materials are obtained. For example, nanostructured particles prepared by mechanical attrition can exhibit unusually high values of hardness,[190] enhanced hydrogen solubility,[191,192] enhanced catalytic properties,[193] magnetic spin-glass behavior,[194] etc. This opens exciting possibilities for the preparation of advanced materials with particular grain-boundary or interphase-boundary design. It is expected that the study of mechanical attrition and alloying processes in the future will not only open new processing routes for a variety of advanced nanostructured materials but will also improve the understanding of technologically relevant deformation processes on a nanoscale level.

Acknowledgments The authors would like to thank all the colleagues who have contributed to this topic over the past ten years for their collaboration and stimulating discussions, in particular Drs. W.L. Johnson (Caltech USA), J.H. Perepezko (UW-Madison, USA), H. Gleiter (Karlsruhe, Germany), R. Valiev (Ufa, Russia), W. Lojkowski, (Poland), I. MacLaren (Great Britain) and X. Sauvage (France).

References 1. Fecht, H.-J., in: Nanomaterials; Synthesis, Properties and Applications, (A.S. Edelstein, and R.C. Cammarata, eds.), Institute of Physics Publishing, Bristol and Philadelphia pp. 89–110 (1996).

166

Processing

2. Gleiter, H., Prog. Mater. Sci., 33:223 (1989). 3. Hadjipanayis, G.C., and Siegel (eds.), R.W., Nanophase Materials, Kluwer Academic (1994). 4. Benjamin, J.S., Scientific American, 234:40 (1976). 5. Fecht, H.-J., Hellstem, E., Fu, Z., and Johnson, W.L., Adv. Powder Metall., 1:111 (1989). 6. Koch, C.C., Cavin, O.B., McKamey, C.G., and Scarbrough, J.O., Appl. Phys. Lett., 43:1017 (1983). 7. Schwarz, R.B., Petrich, R.R., and Saw, C.K., J. Non-Cryst. Solids, 76:281 (1985). 8. Hellstern, E., and Schultz, L., Appl. Phys. Lett., 48:124 (1986). 9. Ettl, C., and Samwer, K., Mater. Sci. Eng., A, 178:245 (1994). 10. Johnson, W.L., Prog. Mater. Sci., 30:81 (1986). 11. Fecht, H.-J., and Johnson, W.L., Nature, 334:50 (1988). 12. Shingu (ed.), P.H., “Mechanical Alloying,” Mater. Sci. Forum, 88–90 (1992). 13. Fecht, H.-J., Nature, 356:133 (1992). 14. Benjamin, J.S., Mater. Sci. Forum. 88:1 (1992). 15. Fecht, H.-J., Hellstem, E., Fu, Z., and Johnson, W.L., Metall. Trans. A, 21:2333 (1990). 16. Eckert, J., Schultz, L., and Urban, K., Appl. Phys. Lett., 55:117 (1989). 17. Koch, C.C., Mechanical Milling and Alloying in Materials Science and Technology, p. 193, R.W. Cahn, P. Haasen, and E.J. Kramer (eds.), (Vol. 15, Processing of Metals and Alloys), VCH, Weinheim (1991). 18. Kuhn, W.E., Friedman, I.L., Summers, W., and Szegvari, A., ASM Metals Handbook , vol. 7, p. 56, Powder Metallurgy, Metals Park (OH) (1985). 19. Schaefer, H.-E., Würschum, R., Gessmann, T., Stöckl, G., Scharwaechter, P., Frank, W., Valiev, R.Z., Fecht, H.-J., and Moelle, C., Nanostructured Materials, 6:869 (1995). 20. Rigney, D.A., Chen, L.H., Naylor, M.G.S., and Rosenfield, A.R., Wear, 100:195 (1984). 21. Rigney, D.A., Annu. Rev. Mater. Sci., 18:141 (1988). 22. Bever, M.B., Holt, D.L., and Titchener, A.L., Prog. Mater. Sci., 17:5 (1973). 23. Miani, F., and Fecht, H.-J., Int. J, Refract. Met. Hard Mater., 17:133 (1999). 24. Eckert, J., Holzer, J.C., Krill III, C.E., and Johnson, W.L., J. Mater. Res., 7:1751 (1992). 25. Williamson, G.K., and Hall, W.H., Acta Metall., 1:22 (1953). 26. Wagner, C.N.J., and Boldrick, M.S., J. Mater. Sci. Eng., A133:26 (1991). 27. Moelle, C., Thesis, Ph.D., Mechanical Milling, Structure and Thermal Stability of Nanocrystalline Iron Powder, Technical University Berlin (1995). 28. Friedel, J., Dislocations, p. 418, Pergamon Press, Oxford (1964). 29. Moelle, C., Schmauss, M., and Fecht, H.-J., unpublished research (1992). 30. Guinier, A., X-ray Diffraction, p. 121, W.H. Freeman, San Francisco (1963). 31. Warren, B.E., and Averbach, B.L., J. Appl. Phys., 21:595 (1950). 32. Koch, C.C., Nanostructured Materials, 2:109 (1993). 33. Hellstern, E., Fecht, H.-J., Fu, Z., and Johnson, W.L., J. Mater. Res., 4:1292 (1989). 34. Yamada, K., and Koch, C.C., J. Mater. Res., 8:1317 (1993). 35. Oehring, M., and Bormann, R., J. de Phys (Paris) Colloq., 51, C-4:169 (1990).

4: Solid State Processing, Fecht and Ivanisenko

167

36. Bakker, H., Di, L.M., and Lo Cascio, D.M.R., Solid State Phen., 23&24:253 (1992). 37. Di, L.M., and Bakker, H., J. Appl. Phys., 71:5650 (1992). 38. Gleiter, H., Acta Metall., 16:455 (1968). 39. Bellon, P., and Averback, R.S., Phys. Rev. Lett., 74:1819 (1995). 40. Martin, G., and Bellon, P., Solid State Phys., 50:688 (1997). 41. Pochet, P., Bellon, P., Boulanger, L., Chaffron, L., and Martin, G., Mater. Sci. Forum, 655:269–272 (1998). 42. Ma, E., He, J.H., and Schilling, P.J., Phys. Rev. B, 55:5542 (1997). 43. Pabi, S.K., Das, D., Mahapatra, T.K., and Manna, I., Acta Metall. Mater., 47:1379–1388 (1998). 44. Shingu, P.H., Huang, B., Niskitani, S.R., and Nasu, S., Trans. JIM (Suppl.), 29:3 (1988). 45. Yavari, A.R., Desré, P.J., and Benameur, T., Phys. Rev. Lett.. 68:2235 (1992). 46. Gaffet, E., Louison, C., Harmelin M., and Faudet., F., J. Mater. Sci. Eng., A134:1380 (1991). 47. Veltl, G., Scholz, B., and Kunze, H.D., J. Mater. Sci. Eng., A134:1410 (1991). 48. Fukunaga, T., Mori, M., Inou, K., and Mizutani, U., J. Mater. Sci. Eng., A134:863 (1991). 49. Yavari A.R., and Desré, P.J., in: Ordering and Disordering in Alloys, A.R. Yavari (ed.), p. 414, Elsevier, New York (1992). 50. Eckert, J., Birringer, R., Holzer, J.C., Krill III, C.E., and Johnson, W.L., J. of App. phys., 73:2794 (1993). 51. Bordeaux, F., and Yavari, A.R., Z. Metallkd., 81:130 (1990). 52. Yang, J.S.C., and Koch, C.C., Mater, J., Res., 5:325 (1990). 53. Schlump, W., and Grewe, H., in: New Materials by Mechanical Alloying Techniques, E. Arzt, and L. Schultz (eds.), p. 307, DGM-Oberursel (1989). 54. Shingu, P.H., Ishihara, K.N., and Kuyama, J., Proc. 34th Japan Congress on Materials Research, p. 19, The Society of Materials Science, Kyoto, Japan (1991). 55. Schwarz, R.B., Mater. Sci. Forum, 269:269–272 (1998). 56. W. L. Johnson and H.-J. Fecht: “Mechanisms of Instability in Crystalline Alloys with Respect to Vitrification”, J. Less-Common Metals, 145 (1988) 63. 57. Fecht, H.J., Han, G., Fu, Z., and Johnson, W.L., J. Appl. Phys., 67:1744 (1990). 58. Sagel, A., Wunderlich, R.K., Perepezko, J.H., and Fecht, H.J., Appl. Phys. Lett., 70:580 (1997). 59. Sagel, A., Wanderka, N., Wunderlich, R.K., Schubert-Bischoff, N., and Fecht, H.J., Scipta Mater., 38:163 (1998). 60. Sagel, A., Wunderlich, R.K., and Fecht, H.-J., Mater. Lett., 33:123 (1997). 61. Fecht, H.-J., Mater. Sci. Eng., A, 179/180:491 (1994). 62. Qi, M., and Fecht, H.-J., Mater. Sci. Forum, 187:269–272 (1998). 63. Michel, D., Mazerolles, L., and Chichery, E., Mater. Sci. Forum, 99:269–272 (1998). 64. Tonejc, A.M., and Tonejc, A., Mater. Sci. Forum, 497:225–227 (1996). 65. Tokumitsu, K., Nasu, T., Suzuki, K., and Greer, A.L., Mater. Sci. Forum, 181:269–272 (1998).

168

Processing

66. 67. 68. 69. 70.

Schaffer, G.B., and McCormick, P.G., Appl. Phys. Lett., 55:45 (1989). Shaw, W.J.D., Mater. Sci. Forum 19:269–272 (1998). Pan, J., and Shaw, W.J.D., J. Appl. Polymer Sci., 56:557 (1995). Karttunen, M., and Ruuskanen, P., Mater. Sci. Forum, 849:269–272 (1998). Fu, Z., Fecht, H.J., and Johnson, W.L., Mater. Res. Soc. Symp. Proc., 186:169 (1991). Chalka, A., and Williams, J.S., Mater. Sci. Forum, 787:88–90 (1992). El-Eskandarany, M.S., Sumiyama, K., Aoki, K., and Suzuki, K., Mater. Sci. Forum. 801:88–90 (1992). Calka, A., Appl. Phys. Lett., 59:1568 (1991). Calka, A., and Williams, J.S., Mater. Sci. Forum, 787:88–90 (1992). Schwarz, R.B., Desch, P.B., Srinivasan, S., and Nash, P., Nanostructured Materials, 1:37 (1992). Schulson, E.M., and Barker, D.R., Scripta Metall., 17:519 (1983). Lu, I.R., Moelle, C., Sagel, A., Wunderlich, R.K., Perepezko, J.H., and Fecht, H.J., Mater. Lett., 35:297 (1998). Moelle, C., Lu, I.R., Sagel, A., Wunderlich, R.K., Perepezko, J.H., and Fecht, H.-J., Mater. Sci. Forum, 47:269–272 (1998). Hayter, J.B., J. Appl. Phys., 21:737 (1988). Glatter, O., J. Appl. Cryst., 10:415 (1977). Atkinson, H.H., and Hirsch, P.B., Phil. Mag., 3:213 (1958). Kronmüller, H., Seeger, A., and Wilkens, M., Z. Physik, 171:291 (1963). Hellstern, E., Fecht, H.-J., Garland, C., and Johnson, W.L., Mater. Res. Soc. Symp. Proc., 132:137 (1989). Cottrell, A.H., in: Dislocations and Plastic Flow in Crystals, p.162, (Caledron Press, Oxford 1972). Hall, E.O., Proc. Phys. Soc. London, 643:747 (1951). Petch, N.J., J. Iron, Steel Inst. London, 173:25 (1953). Meyers, M.A., and Chawla, K.K., Mechanical Metallurgy, p. 494, (Prentice-Hall, Englewood Cliffs, NJ, 1984). Karch, J., Birringer, R., and Gleiter, H., Nature, 330:556 (1987). Eckert, J., Holzer, J.C., Krill III, C.E., and Johnson, W.L., J. Mater. Res., 7:1751 (1992). Bever, M.B., Holt, D.L., and Titchener, A.L., Prog. Mater. Sci., 15:5 (1973). Essmann, U., and Mughrabi, H., Phil. Mag. A, 40:40 (1979). Fecht, H.-J., Phys. Rev. Lett., 65:610 (1990). Kuwano, H., Ouyang, H., and Fultz, B., Nanostructured Materials, 1:143 (1992). Morris, D.G., Mechanical Behavior of Nanostructured Materials, Trans Tech Publications Ltd., Switzerland (1998). Koch, C.C., J. Metastable Nanocrystalline Mater., 565:2–6 (1999). Fecht, H.-J., Phil. Mag. B, 76:495 (1997). Porat, R., Berger, S., and Rosen, A., Mater. Sci. Forum., 629:225–227 (1996). Pfullmann, Oehring, T., Bohn, R., and Bormann, R., Mater. Sci. Forum., 757:225–227 (1996). Moelle, C., and Fecht, H.-J., Nanostructured Materials, 6:421 (1995).

71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99.

4: Solid State Processing, Fecht and Ivanisenko

169

100. Fecht, H.-J., and Moelle, C., Nanophase and Nanocomposite Materials II, S. Komarneni, J.C. Parker, and H.-J. Wollenberger (eds.), p. 113, Materials Research Society, Symposium Proceedings, vol. 457 (1997). 101. Tschöpe, A., Birringer, R., and Gleiter, H., J. Appl. Phys., 71:5391 (1992). 102. Chen, L.C., and Spaepen, F., J. Appl. Phys., 69:679 (1991). 103. Wolf, D., Phil. Mag. A, 62:447 (1990). 104. Malow, T.R., and Koch, C.C., Mater. Sci. Forum, 595:225–227 (1996). 105. Turley, D., J. Inst. Metals., 99:271 (1971). 106. Wagner, C.N.J., Acta Metall., 5:477 (1957). 107. Warren, B.E., “X-ray studies of deformed metals,” Prog. Metals Phys., 3:147 (1956). 108. Valiev, R.Z., Korznikov, A.V., and Mulyukov, R.R., Mater. Sci. Eng., A168:141(1993). 109. Valiev, R.Z., Islamgaliev, R.K., Stolyarov, V.V., Mishra, R.S., and Mukherjee, A.K., Mater. Sci. Forum, 969:269–272 (1998). 110. Kuhlmann-Wilsdorf, D., and Van der Merwe, J.H., J. Mater. Sci. Eng., 55:79 (1982). 111. Haessner, F., and Hemminger, W., Z. Metallk., 69:553 (1978). 112. Lloyd, D.J., and Kenny, D., Scripta Metall., 12:903 (1978). 113. Grosskreutz, J.C., and Mughrabi, H., Constitutive Equations in Plasticity, A.S. Argon (ed.), p. 251, MIT Press, Cambridge, MA (1971). 114. Gilman, J.J., J. Appl. Phys., 46:1625 (1975). 115. Donovan, P.E., and Stobbs, W.M., Acta. Metall., 31:1 (1983). 116. Hatherly, M., and Malin, A.S., Scripta Metall., 18:449 (1984). 117. Goods, S.M., and Brown, L.M., Acta Metall., 27:1 (1979). 118. Segal, V.M., Reznikov, V.I., Drobishevski, F.E., and Kopilov, V.I., Izv. SSSR Metall., 1:115 (1981). 119. Smirnova, N.A., Levit, V.I., Pilugin, V.P., Kuznezov, R.I., Davidova, L.S., and Sazonova, V.A., Phys. Met. Metallogr., 6:1170 (1986). 120. Valiev, R.Z., Islamgaliev, R.K., and Alexandrov, I.V., Prog. Mater. Sci., 45:103 (2000). 121. Tcherdyntsev, V.V., Kaloshkin, S.D., Gunderov, D.V., Afonina, E.A., Brodova, I.G., Stolyarov, V.V., Baldokhin, Yu. V., Shelekhov, E.V., and Tomilin, I.A., Mater. Sci. Eng., 888:A375–377 (2004). 122. Sauvage, X., Wetscher, F., and Pareige, P., Acta Mater., 53:2127 (2005). 123. Korznikov, A.V., Tram, G., Dimitrov, O., Korznikova, G.F., Idrisova, S.R., and Pakiela, Z., Acta Mater., 49:663 (2001). 124. Rentenberger, C., Mangler, C., and Karnthaler, H.P., Mater. Sci Eng. A, 387–389:795–798 (2004). 125. Senkov, O.N., Froes, F.H., Stolyarov, V.V., Valiev, R.Z., and Liu, J., Nanostructured Materials, 10:691 (1998). 126. Waitz, T., Kazykhanov, V., and Karnthaler, H.P., Acta Mater., 52:137 (2004). 127. Jiang, H., Zhu, Y.T., Butt, D.P., Alexandrov, I.V., and Lowe, T.C., Mater. Sci. Eng. A, 290:128 (2000). 128. Hebesberger, T., Stüwe, H.P., Vorhauer, A., Wetscher, F., and Pippan, R., Acta Mater., 53:393 (2005).

170

Processing

129. Zhilyaev, A.P., Nurislamova, G.V., Kim, B.-K., Baró, M.D., Szpunar, J.A., and Langdon, T.G., Acta Mater., 51:753 (2003). 130. Valiev, R.Z., Ivanisenko, Yu.V., Rauch, E.F., and Baudelet, B., Acta Mater., 44:4705 (1996). 131. Ivanisenko, Yu., Valiev, R.Z., and Fecht, H.-J., Mater. Sci. Eng. A, 390:159 (2005). 132. Wadsack, R., Pippan, R., and Schedler, B., Fusion Engineering and Design, 265:66–68 (2003). 133. Kolobov, Yu.R., Kieback, B., Ivanov, K.V., Weissgaerber, Th., Girsova, N.V., Pochivalov, Yu.I., Grabovetskaya, G.P., Ivanov, M.B., Kazyhanov, V.U., and Alexandrov, I.V., Int. J. Refract. Met. Hard Mater., 21:69 (2003). 134. Sergueeva, A.V., Stolyarov, V.V., Valiev, R.Z., and Mukherjee, A.K., Scripta. Mater., 45:747 (2001). 135. Rybal’chenko, O.V., Dobatkin, S.V., Kaputkina, L.M., Raab, G.I., and Krasilnikov, N.A., Mater. Sci. Eng. A, 244:387–389 (2004). 136. Sakai, G., Horita, Z., and Langdon, T.G., Mater. Sci. Eng. A, 393:344 (2005). 137. Ivanisenko, Yu., Lojkowski, W., Valiev, R.Z., and Fecht, H.-J., Acta. Mater., 51:5555 (2003). 138. Ivanisenko, Yu., Sirenko, A.A., and Korznikov, A.V., Phys. Met. Metallogr., 87:329 (1999). 139. Valiev, R.Z., Gertzman, V.Yu., and Kaibyshev, O.A., Phys. Stat. Sol (a), 61:K95 (1980). 140. Klemm, V., Klimanek, P., Kolesnikova, A.L., Motylenko, M., and Romanov, A.E., Ann. Chim. Sci. Mater., 27:25 (2002). 141. Hughes, D.A., and Hansen, N., Acta. Mater., 11:2985 (2000). 142. Embury, J.E., Mater. Sci. Eng. A, 175:105 (1994). 143. Hirth, J.P., and Lothe, J., Theory of Dislocations, 2nd ed., Wiley, New York, 1982. 144. Rentenberger, C., and Karnthaler, H.P., Acta. Mater., 53:3031 (2005). 145. Rentenberger, C., Mangler, C., and Karnthaler, H.P., Mater. Sci. Eng. A, 795:387–389 (2004). 146. Sergueeva, A.V., Song, C., Valiev, R.Z., and Mukherjee, A.K., Mater. Sci. Eng. A, 339:159 (2003). 147. Pushin, V.G., Stolyarov, V.V., Valiev, R.Z., Kourov, N.I., Kuranova, N.N., Prokofiev, E.A., and Yurchenko, L.I., Ann. Chim. Sci. Mater., 27:77 (2002). 148. Waitz, T., Kazykhanov, V., and Karnthaler, H.P., Acta. Mater., 52:137 (2004). 149. Korznikov, A.V., Ivanisenko, Yu.V., Laptionok, D.V., Safarov, I.M., Pilyugin, V.P., and Valiev, R.Z., Nanostructured Materials, 4:159 (1994). 150. Korznikov, A.V., Safarov, I.M., Laptionok, D.V., and Valiev, R.Z., Acta. Metall. Mater., 39:3193 (1991). 151. Korznikova, G., Korznikov, A., Pakiela, Z., and Kurzydlowski, K.J., Mater. Chem. Phys., 81:401 (2003). 152. Straumal, B.B., Baretzky, B., Mazilkin, A.A., Phillipp, F., Kogtenkova, O.A., Volkov, M.N., and Valiev, R.Z., Acta. Mater., 52:4469 (2004). 153. Ohsaki, S., Hono, K., Hidaka, H., and Takaki, S., Scripta. Mater., 52:271–276 (2005).

4: Solid State Processing, Fecht and Ivanisenko

171

154. Sauvage, X., Dacosta, G., and Valiev, R.Z., Ultrafine Grained Materials III, p. 31, TMS (The Minerals, Metals & Materials Society) (2004). 155. Languilaumme, J., Kapelski, G., and Baudelet, B., Acta. Mater., 45:1201 (1997). 156. Sauvage, X., Copreaux, J., Danoix, F., and Blavette, D., Phil. Mag., 4:781 (2000). 157. Gridnev, V.N., and Gavrilyuk, V.G., Phys. Met., 4:531 (1982). 158. Weissmüller, J., and Cahn, J.W., Acta. Mater., 45:1899 (1997). 159. Ivanisenko, Yu.V., Lojkowski, W., Valiev, R.Z., and Fecht, H.-J., Solid State Phen., 94:45 (2003). 160. Shabashov, V.A., Nanostructured Materials, 6:711 (1995). 161. Tavares, S.S.M., Gunderov, D., Stolyarov, V., and Neto, J.M., Mater. Sci. Eng. A, 358:32 (2003). 162. Ivanisenko, Yu., MacLaren, I., Valiev, R.Z., and Fecht, H.-J., Adv. Eng. Mater., 11:1011 (2005). 163. Ivanisenko, Yu., Wunderlich, R., Valiev, R.Z., and Fecht, H.-J., Scripta. Mater., 49:947 (2003). 164. Lee, Z., Zhou, F., Valiev, R.Z., Lavernia, E.J., and Nutt, S.R., Scripta. Mater., 51:209 (2004). 165. Stolyarov, V.V., Zhu, Y.T., Lowe, T.C., Islamgaliev, R.K., and Valiev, R.Z., Mater. Sci. Eng. A, 282:78 (2000). 166. Alexandrov, I.V., Zhu, Y.T., Lowe, T.S., Islangaliev, R.K., and Valiev, R.Z., Metall. Trans. A, 29A:2253 (1998). 167. Ivanisenko, Yu., Krasilnikov, N., Banhart, F., Kolesnikov, D., Valiev, R., Lojkowski, W., and Fecht, H.-J., Ann. Chim. Sci. Mater., 3:45 (2002). 168. Sort, J., Ile, D.C., Zhilyaev, A.P., Concustell, A., Czeppe, T., Stoica, M., Suriñach, S., Eckert, J., and Baró, M.D., Scripta. Mater., 50:1221 (2004). 169. Boucharat, N., Hebert, R., Rösner, H., Valiev, R., and Wilde, G., Scripta. Mater., 53:823 (2005). 170. Atzmon, M., Verhoeven, J.D., Gibson, E.D., and Johnson, W.L., Appl. Phys. Lett., 45:1052 (1984). 171. Atzmon, M., Unruh, K.M., and Johnson, W.L., J. Appl. Phys., 58:3865 (1985). 172. Bourdeaux, F., and Yavari, A.R., J. Appl. Phys., 67:2385 (1990). 173. Yasuna, K., Terauchi, M., Otsuki, A., Ishihara, K.N., and Shingu, P.H., Appl. Phys. Lett., 82:2435 (1997). 174. Sagel, A., Sieber, H., Fecht, H.J., and Perepezko, J.H., Phil. Mag. Lett., 77:109 (1998). 175. Ivanisenko, Y.V., Baumann, G., Fecht, H.-J., Safarov, I.M., Korznikov, A.V., and Valiev, R.Z., Phys. Met. Metallogr., 83:303 (1997). 176. Fecht, H.-J., Nanostructured Materials, 6:33 (1995). 177. Ganapathi, S.K., and Rigney, D.A., Scripta Metall., 24:1675 (1990). 178. Doyle, F.D., and Aghan, R.L., Metall. Trans. B, 6:143 (1975). 179. Mehrotra, P.K., Proc. Int. Conf. on Wear of Materials, Reston, VA, p. 194, ASME, New York (1983). 180. Humble, P., and Hannink, R.H.-J., Nature, 273:37 (1978). 181. Beilby, G., Aggregation and Flow of Solids, Macmillan, London (1921).

172

Processing

182. Askenasy, P., Thesis, Ph.D., Tribological Behaviour of Intermetallics, California Institute of Technology (1992). 183. Tao, N.R., Sui, M.L., Lu, J., and Lu, K., Nanostructured Materials, 11:433 (1999). 184. Lu, K., private correspondence (1999). 185. Baumann, G., Fecht, H.-J., and Liebelt, S., Wear, 191:133 (1996). 186. Baumann, G., and Fecht, H.-J., Nanostructured Materials, 7, 1/2:237 (1996). 187. Bürkle, G., Nanostructure Formation of Ferritic Steels under High Shear Conditions, Ulm University (1999). 188. Siegel, R.W., and Fougere, G.E., in: Nanophase Materials, G.C. Hadyipanayis and R.W. Siegel (eds.), p. 233, Kluwer Academic, New York (1994). 189. Nieman, G.W., Weertman, J.R., and Siegel, R.W., J. Mater. Res., 6:1012 (1991). 190. Kehrel, A., Moelle, C., and Fecht, H.-J., in Ref. 3. 191. Moelle, C., and Fecht, H.-J., Nanostructured Materials, 3:93 (1993). 192. Ram, S., Fecht, H.-J., Haldar, S., Ramachandrarao, P., and Banerjee, H.D., Phys. Rev. B, 56:1 (1997). 193. Zaluski, L., Zaluska, A., Tessier, P., Ström-Olsen, J.O., and Schulz, R., Mater. Sci. Forum, 853:225–227 (1996). 194. Zhou, G.F., and Bakker, H., Phys. Rev. Lett., 72:2290 (1994).

5 Nanocrystalline Powder Consolidation Methods Joanna R. Groza Department of Chemical Engineering and Materials Science, University of California, Davis, California, USA

5.1 Introduction In powder consolidation, or sintering, an initial weak porous body made of solid particles is converted into a dense and strong bulk part. The process offers great versatility in the shape, size, and weight of the final parts. The bulk sintered materials with nanoscale structures have the potential to offer novel or enhanced properties. As detailed in Chapter 12 of this book, mechanical behavior will benefit from nanometer-range grain sizes by achieving unusually high strength, hardness, wear resistance, or superplasticity. Other physical behavior may be dramatically changed to achieve novel optical and magnetic properties or enhanced biocompatibility. The challenge in processing nanocrystalline powders is to fully densify them without losing the initial metastable features (nanoscale grain size and, sometimes, metastable phases). Nanoscale powders date back to antiquity. Densification of nanopowders was attempted in the 1960s. Some of these efforts were related to sintering MgO to achieve superplastic behavior.[1] In the 1980s, when nanopowder production was initiated on a larger scale, attention was directed to nanopowder processing as well. To date, the synthesis of raw nanomaterials (powders) has progressed further than the formation of bulk parts. Nanosize powders of metals, ceramics, polymers, and composites are now available, sometimes in tonnage quantitities. However, densification of nanoscale structure is still a challenge, commonly due to grain coarsening and, occasionally, to insufficient densification or inadequate bonding. The 1990s stressed the need to develop reproducible processing methods for manufacturing nanopowders into sizable and net shape parts with nanoscale features. The past decade and a half has brought Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 173–234 © 2007 William Andrew, Inc.

173

174

Processing

significant advances in the practice and theory of nanocrystalline powder sintering to produce fully dense parts with nanometer grain size (considered throughout this chapter as ≤100 nm). The densification process for conventional powders is well known theoretically and practically. However, the densification of nanopowders poses significant additional challenges. Powder agglomeration, high reactivity and therefore contamination, grain coarsening with loss of the nanofeatures, and inability to fabricate large and dense parts are among the main problems. The lower temperatures desired to minimize grain growth may hinder good intergranular bonding, thus compromising the expected high mechanical strength and ductility. Such low sintering temperatures may also interfere with the thermochemical oxide reduction on particle surfaces necessary for subsequent sinter bonding. The efforts so far have been very fruitful in overcoming some of these problems (e.g., agglomeration). This has been accomplished by major improvements in the nanopowder synthesis methods and understanding of aspects of the densification process such as pore effects in nanosintering. For nanopowder consolidation, an intriguing question is whether the sintering mechanisms scale with grain size or whether there are changes in these mechanisms when the nanoscale is reached. This review has been written with this question in mind. More specifically, differences between the sintering of regular and nanosize powders will be highlighted. It will be shown that when smaller scales are approached, atomic mechanisms are more easily unveiled. For instance, the reorientation during early sintering stages becomes evident only when particles become very small. Alternatively, the possibility of new sintering mechanisms will be examined. A number of reviews on nanopowder consolidation and processing of specific nanomaterials[2–8] or general reviews addressing sintering issues[9–11] have been published. This chapter first covers some thermodynamic and kinetic aspects of nanopowder densification: driving force, surface energy, sintering mechanisms, activation energies and scaling laws. Due to the critical effect of surface contamination on small particles, the role of impurities in sintering is described separately. Next, cold compaction with resultant pore size and size distribution and their effects on sintering and grain coarsening are addressed. The presentation of the sintering process of nanoparticles is divided into pressureless (conventional) and pressure-assisted sintering. Finally, methods for full densification of nanopowders and their ability to maintain the nanosize features are presented.

5: Nanocrystalline Powder Consolidation Methods, Groza 175

5.2 Thermodynamics, Mechanisms and Kinetics of Nanocrystalline Powder Densification 5.2.1 Thermodynamic and Kinetic Effects Thermodynamically, nanopowders are highly unstable, sometimes by as much as 10–50 kJ/mol. The sintering process is driven by the tendency to reduce the excessively large surface area per unit volume. The extra energy of a surface with a radius of curvature R may be expressed as a stress (σ) in a Laplace equation: σ = γ/R

Eq. (5-1)

where γ is the surface energy. In nanomaterials, the sintering stress may as large as 300 MPa in 10 nm particles as compared to only 3 MPa for 1 μm particles, if γ has a typical value of 1.5 J/m2.[12] For traditional sintering studies, the surface energy is assumed to be isotropic. The implications and difficulties brought about by this simplifying assumption in conventional powders, particularly ceramics, which are more anisotropic than metals, have been addressed.[13] For nanocrystals with significant surface area, the anisotropy problem becomes even more critical. First, sintering starts at a lower temperature. In this case, the effects of surface energy anisotropy are more pronounced. Usually, if the temperature is sufficiently high, the surface energy anisotropy may be reduced. Second, direct transmission electron microscopy (TEM) studies of γ-Al2O3,[14] ZrO2,[15,16] and CeO2[17] indeed showed that nanoparticles have a faceted appearance, specific for anisotropic surface energies. Even for metals, a departure from the ideal spherical particle concept and, therefore, isotropy is accentuated in the nanosize range. Theory and experiment have indicated that clean fcc metal particles with diameters less than 4–7 nm have an icosahedral configuration with multiple twins in their equilibrium condition.[18] In metals, high-resolution electron microscopy (HREM) studies revealed faceted nanoparticles, which are ascribed to surfaces with minimum energy.[19] The highly anisotropic behavior of nanoparticle surfaces may be responsible for the crystallographic alignment that has often been observed. Dynamic simulations and experiments indicated that nanoparticles rotate and align when necking occurs between loose particles.[15,20,21] An atomic-force microscopy study showed that

176

Processing

nano-TiO2 powders obtained by gas condensation have a preferred alignment in a chainlike structure for specimens compacted at room temperature.[22] Although not specifically shown, some crystallographic alignment may be implied. In some cases, this behavior is exploited to control the particle assembly into ordered arrays for specific nanostructures with complex patterns.[23] The altered local atomic arrangement at the surface of nanocrystals may result in surface energy values different from those in conventional powders. This may be the case with amorphous layers, which have been observed on some Al nanoparticle surfaces.[24,25] As Cahn infers, the different nature of the surface oxide on nanopowders may be an indication of a different surface structure.[26] Experimentally, Trudeau et al. showed highly nonstoichiometric surface oxides on nanocrystalline catalysts.[27] BaTiO3 has a cubic surface layer on the tetragonal bulk nanocrystal.[28] Although some caution is in order, Payne also showed that interfaces display electronic states and, therefore, properties different from bulk crystals.[29] Molecular dynamics simulations indicate highly disordered particle surfaces (coordination number 3.5 as compared to 4 in the bulk) in nanosize SiC.[30] The interfacial energy may be also modified by any distortion of surface structure, such as by segregation or adsorption of impurities. Impurities dictate the thermodynamics of surfaces and surface behavior of nanoparticles. In the Al2O3 system, calorimetric studies by Navrotsky and co-workers indicated that hydration at surfaces of nanocrystals results in equal surface energies of α and γ polymorphs, despite the lower surface energy of anhydrous γ-Al2O3.[31] This lower surface energy induces the stabilization of γ-Al2O3 in nano-aluminas instead of the α-Al2O3, which is stable in conventional crystals.[32] Surface hydroxyls on γ-Al2O3 surfaces, which are present up to high temperatures, induce sintering difficulties with this ceramic.[31] More support of the effects of surface structures on nanoparticle sintering comes from modeling, which showed a high dependence of grain boundary formation on initial surface structures in ceramics (e.g., SiC).[33] Experimentally, this was confirmed by a large effect of initial crystallographic surfaces on the microstructural development during sintering.[34] Macroscopically, stabilization of particle surfaces, i.e., lower surface energy values when oxides are formed, results in a decrease of the driving force for sintering. This has been observed in nanosize particles, which typically contain large amounts of adsorbates (e.g., high oxygen content in TiN[35]). In addition to the inherent morphological metastability related to fine grain size, nanocrystalline materials often display topological (e.g., alternate crystal structures from those at equilibrium) or compositional (e.g.,

5: Nanocrystalline Powder Consolidation Methods, Groza 177 extended solid solubility or amorphous phases) metastabilities.[36] Alternate crystal structures are found in materials with high-pressure polymorphism, such as cubic BaTiO3,[37] monoclinic Y2O3,[38] and tetragonal ZrO2[39] owing to the Gibbs–Thomson effect. Alternate crystal structures due to interfacial energy effects were shown by Krauss and Birringer in nanocrystalline refractory metals (the A15 modification of β-W and βTa).[40] These metastable phases are obtained only in powder form. When films of W and Ta have been synthesized with the same grain size, only the equilibrium structures have formed. In contrast to the high-pressure phases in ceramics, the metal structures have a lower density than the equilibrium crystals. The retention of initial powder metastability into the bulk parts is relevant to sintering. To retain the unique properties of nanocrystals throughout sintering, grain size must remain below the critical value at which property conversion occurs. Some critical grain sizes at which nanomaterial structure or physical properties revert back to those of conventional materials are compiled in Table 5.1. For some other properties, i.e., mechanical, such a threshold value has not been demonstrated or this value is property dependent. Moreover, there is not yet consensus on the strengthening benefits in the nanocrystalline range, but the current efforts are to push the limit toward increasingly smaller grain sizes. Therefore, an arbitrary limit for the maximum grain size in the nanocrystalline materials is often considered at ≤100 nm.

Table 5.1 Critical Grain Size of Nanocrystals Below Which Structure/Property Characteristics are Different from Those of Conventional Materials

Critical Grain Size (nm) ∼13

Structure/Property Nano Monoclinic

Material

Reference

Conventional Cubic

Y2O3

38

8–26

Tetragonal

Monoclinic

ZrO2

39

24–30

Reducedthermal conductivity

Regular

ZrO2 (YSZ)

41

50

Anatase

Rutile

TiO2

42

>80

Superparamagnetic

Antiferromagnetic

Cr2O3

43

120

Cubic

Tetragonal

BaTiO3

37

120

Variable TCurie

Constant TCurie

BaTiO3

28

Processing

178

Significantly enhanced kinetics are expected for processes that display direct grain size dependence. For sintering, this dependence may be illustrated using the equation for the normalized densification rate (dL/L dt) developed by Johnson and co-workers for all stages of sintering:[44] −

dL γΩ ⎛ δDb Γb Dv Γv ⎞ = + 3 Ldt kT ⎝ d 4 d ⎠

Eq. (5-2)

where γ is the surface energy, Ω is the atomic volume, δ is the grain boundary width, Db and Dv are the grain boundary and bulk diffusivities, Γb and Γv are functions of density, kT has the usual meaning, and d is the grain size. From this equation, it is seen that decreasing grain size by three orders of magnitude (e.g., from micrometer to nanometer) could enhance sintering rates by up to 12 orders of magnitude. Consequently, sintering of nanopowders may be accomplished at significantly lower temperatures and shorter duration than for conventional powders. This has been noted for numerous examples of real nanoparticle sintering, as will be detailed later in this section. A more dramatic effect of the depression of sintering temperature in nanopowders is seen in the liquid-phase sintering systems, such as WC-Co.[45] Due to the accelerated sintering at lower temperatures in the nanoregime, full densification may take place entirely during solidstate sintering.

5.2.2 Sintering Mechanisms The densification process, driven by capillarity, consists of solid particle bonding or neck formation followed by continuous closing of pores from a largely open porosity to essentially a pore-free body. For simplicity, solid-state densification is considered to be accomplished through three stages: initial, intermediate, and final. Classical sintering theory describes the multiple mechanisms involved throughout these stages as evaporation-condensation, surface diffusion, grain boundary diffusion, bulk diffusion, viscous flow, and plastic deformation. Each transport process exhibits a particular dependence on the particle/grain size and defect density. The highest sensitivity on particle size is that of surface and grain boundary diffusion. Although simultaneous mechanisms participate, the common sintering models attribute a predominant mass transport path to a specific sintering stage. For instance, surface diffusion is considered the principal mechanism during the initial stage when the main event is neck formation. So far, a similar partition of the sintering process is accepted in the existing sintering theories applied to nanopowders.

5: Nanocrystalline Powder Consolidation Methods, Groza 179 For nanoparticles with excessive surface area and highly curved surfaces, surface diffusion is expected to be extremely rapid in the early sintering stages. Indeed, the contribution of surface diffusion to the neck formation is evident in sintering of a variety of nanopowders (TiO2,[46] ZrO2,[15,47] Y-TZP,[48] Al2O3[14]). Surface diffusion was the main mechanism for neck growth observed by in-situ TEM studies of alumina[14] and zirconia.[15,47] Rankin and Sheldon also considered the lack of change in the particle center spacing as an indication of a surface diffusion mechanism during neck formation and growth in nano-zirconia.[15] Surface diffusion is known to occur by a terrace-ledge-kink mechanism, which was observed on the faceted and anisotropic surface structure of alumina nanoparticles.[14] This unveiling of atomic mechanisms is an additional benefit of the study of smaller scales. From classical sintering theory, it is known that surface diffusion leads not to densification but to grain coarsening.[49] Then, if surface diffusion is the most sensitive to particle size, reduced low-temperature densification should be observed in nanoparticle sintering. Abnormal grain growth and pore coarsening without densification were indeed noticed at low sintering temperatures in ultrafine silicon.[50] However, this is more an exception, since enhanced low-temperature densification of nanoparticles has typically been observed.[2,4,6,7] Computer simulations also indicate extremely fast sintering behavior of nanoparticles.[20,21,30,51,52] (see also Chapter 6). Surface diffusion cannot explain this behavior. Furthermore, the usual approach of using high heating rates to avoid deleterious surface diffusion does not seem always to apply to small powders. In some cases, slow heating rates induced surprisingly high density.[53] In other cases, no effect of the heating rates on densification or grain growth has been observed.[54] Although in the latter instances the heating rates used (2– 200 K/min) may be too slow to suppress grain coarsening, the contribution of surface diffusion to nanosintering is not clear. The ratio between bulk (and/or grain boundary) and surface diffusion contribution to overall densification is not yet fully known and may be different from that in conventional powders. Wang et al. calculated surface diffusion rates of 2 × 10−6 cm2/s in the neck region during sintering of Si3N4.[55] These rates are about five times larger than bulk diffusion. Furthermore, evaluations of surface diffusion contribution are more difficult due to powder contamination and agglomeration (e.g., Ref. 56). Surface anisotropy may also fundamentally change the surface transport kinetics. Even for well-known model systems, large uncertainties in the surface energy values exist and add further complications.[57] Additional systematic work is required to quantify the impurity and faceting effects on surface diffusion to better

180

Processing

detect changes in the sintering mechanisms of nanoparticles, primarily in the initial sintering stages. Some other mechanisms for neck formation in nanopowders have been suggested: grain boundary sliding, dislocation motion (shear deformation), mechanical rotation, viscous flow, and even grain boundary melting or amorphization. Thuenissen et al. suggested that grain boundary sliding assisted by surface diffusion is the main contributor to early densification in Y-TZP.[48] For the particular grain and pore size conditions, the calculated sintering stress (70 MPa) is considered sufficient to cause grain boundary sliding and particle rearrangement. A low calculated activation energy, smaller than for any known diffusion mechanism in the context of large porosity and small grain size, may be compatible with grain boundary slip assisted by surface diffusion. This grain boundary sliding, similar to superplasticity, may be also effective in the densification of nanoparticles by severe plastic deformation.[58] Dislocation motion, particle/grain rotation (or rearrangement), amorphization, and viscous flow in nanoparticle sintering were identified by molecular dynamics (MD) simulations.[20,21,30,51,52,59] Most commonly, particle rotation was predicted to minimize grain boundary energy, but was also driven by large adhesion forces.[20,21,59,60] TEM studies amply confirmed particle rotations in both metals (Cu, Ag on Cu[61,62]) and ceramics (Al2O3, ZrO2, CeO2[14,15,17]). Chen and Chen applied the new sintering mechanism based on particle rearrangement to explain the remarkable sinterability of ceramic nanoparticles, even starting with low green density.[63] Thunissen et al. documented the formation of single-crystal aggregates (14 nm) from several initial 8 nm particles[48] as predicted by simulations.[52] Sintering of nanoparticle pairs or of nanowires on a substrate was shown to take place in picoseconds.[52,64] Averback and co-workers calculated that diffusion could not explain this rapid sintering.[52] To account for the observed shrinkage, the grain boundary diffusivity in Cu at 700 K has to be 10−2 cm2/s, a value that is two orders of magnitude larger than liquid diffusivity in Cu. This rapid shrinkage simultaneous with neck formation was attributed to fast dislocation activity driven by the contact Hertzian stresses that exceed the ideal shear strength. An experimental confirmation of such high stresses was provided by dipole stress fringes observed at the contact of two Co nanoparticles.[65] A similar fast generation of dislocations and shearing was predicted by Zeng et al. in sintering of Cu nanofibers on substrates.[21] When three or more particles are sintered, dislocation motion cannot account for full densification. Grain rotation is also restricted, thus creating large residual stresses. Further

5: Nanocrystalline Powder Consolidation Methods, Groza 181 densification is accomplished by grain boundary sliding.[52] Alternatively, the atomic events may culminate in transient or permanent amorphization, particularly if the particles are too small to support crystallinity.[21,62] Experimentally, Bonevich and Marks showed some possible amorphization during epitaxial re-growth of a small particle in sintering nanoalumina.[14] A critical consideration of the nanoparticle melting observed by TEM may be found in a review by Buffat.[66] On a different theoretical basis, Trusov et al. demonstrate that vacancies generated at grain boundaries and pores may reach premelting levels in a very short time in the early stages of sintering of nanoparticles due to a viscous flow mechanism.[67] Dominguez et al. also describe sintering of nanosize Cu and Fe based on grain boundary amorphization or melting mechanisms.[68] For materials with difficult dislocation motion, such as intermetallic compounds, MD simulations indicate viscous flow to be the main mechanism for neck formation and growth.[52] While there is theoretical and experimental evidence to support these new mechanisms in nanoparticle sintering, there is still much uncertainty in their individual or collective contribution to the overall sintering process. Whether these new mechanisms are generally applicable or are system-specific is largely unknown. If new mechanisms contribute to nanosintering, a simple scaling down is not applicable. Careful reinspection of the simplifying assumptions used to derive the sintering equations is required. In particular, surface effects and particle anisotropy should be reconsidered. Sintering diagrams or models indicating different mechanisms as a function of sintering stage and powder size would be most helpful.

5.2.2.1 Activation Energy Commonly, the calculation of activation energy values provides insight into the microscopic process mechanisms. As already shown, sintering is accomplished by multiple mechanisms. As a result, the interpretation of the Arrhenius plots in sintering is difficult, as emphasized in the sintering literature.[49,69] This is even more true for nanoparticles as seen in only the few activation energy values available. In addition to the difficulty of identifying a dominant mechanism, sintering mechanisms change with particle size. The largest departure from conventional sintering is seen in the initial sintering stages. In these stages, lower activation energies than for conventional diffusion have been reported. For instance, Trusov et al. report a lower activation energy (134 kJ/mol) than for surface diffusion

182

Processing

(∼300 kJ/mol[49]) in 40 nm tungsten.[67] Vernon et al. cite activation energies of 234 kJ/mol for 13.5 nm Al2O3 and 96 kJ/mol for 11.5 nm TiO2.[70] Thunissen et al. also found an atypically low activation energy (100 kJ/mol) in the early sintering of 15 nm Y-TZP,[48] while the activation energy has the usual value of 275 kJ/mol when particle size is 50 nm. Activation energy comparable to that of liquid diffusion was reported for nanoparticle iron and copper sintered in hydrogen by Dominguez et al.[68] Some of these low activation energy values can not unambiguously be assigned to surface diffusion due to the lack of comparable literature data, particularly for ceramics. In addition, a more valid comparison would use nanomaterial diffusion data rather than diffusion data on regular materials. As already mentioned, the low activation energy values in early sintering stages may be indicative of some different mechanisms occurring in sintering nanopowders. For late sintering stages, activation energies are in better agreement with values determined for sintering of conventional powders.[48,71,72] For instance, Messing and Kumara found activation energies that correspond to volume diffusion for sintering α-Al2O3 at 90% density.[71] Only slight changes, such as the predominance of grain boundary diffusion instead of volume diffusion, have been noticed in the densification by power-law creep of Fe10 wt% Cu during sinter forging.[72] This shift to grain boundary diffusion is not unreasonable since sintering temperatures are lower than in large-grain powders. By late sintering conditions, the initial nanopowders have already lost some of the more conspicuous initial nano features and therefore their densification behavior converges toward that of regularly grained materials.

5.2.2.2 Scaling Laws As already mentioned, the onset of sintering is consistently reported to be significantly lower in nanoparticles compared to conventional materials. The sintering of both metal and ceramic nanoparticles was found to start at temperatures of 0.2–0.5Tm as compared to 0.5–0.8Tm for conventional powders. Some examples of the variation of sintering onset with particle size are shown in Table 5.2. Full densification of nanopowders is also completed at temperatures lower than those for conventional powders. Densification of 15 nm TiN starts at 1170 K and is completed below 1823 K, whereas the microcrystalline powder is only 63% dense at 1823 K.[35] In Si3N4, Peacenik et al. fully sintered nanopowders at 700 K lower than the coarse grained

5: Nanocrystalline Powder Consolidation Methods, Groza 183 Table 5.2 Examples of Dependence of Sintering Onset on Nanoparticle Size

Material a

Fe W Y2O3 ZrO2 Al2O3 TiO2b SiCc ZrO2 a,b c

Particle Size (nm)

T (K) (T/T0)

Reference

30 9 4 8–9 50–100 13 3 70

393 (0.21) 900 (0.24) 600 (<0.25) 870–920 (∼0.3) 1273 (0.55) 823 (0.4) 1500 (0.48) 1370 (∼0.5)

67 76 73 73 75 46 30 74

Sintering onsets are: a 900 K(0.5Tm) for 2 μm Fe; Molecular dynamics simulation results.

b

950 K(0.46Tm) for 40 nm TiO2.[77]

analogues.[78,79] Nanometer-sized TiO2 (12–14 nm) completes sintering at ∼1300 K compared to more than 1670 K for 1.3 μm TiO2.[46] For gelderived TiO2 powders, near full density is achieved at 875 K.[80] Presently, the lower sintering temperatures of nanopowders are well recognized and fully exploited in numerous applications. Technologically, there are significant benefits from the lower sintering temperatures of nanopowders, such as the possibility to avoid sintering aids, phase decomposition, deleterious interfacial interactions, and undesirable phase transformations (e.g., the monoclinic to tetragonal ZrO2 transformation, which inevitably results in cracking).[73] Different scaling laws have been applied to rationalize this decrease in the sintering temperature. For particles under 100 nm, Alamov et al. developed an empirical relationship for the dependence of the sintering onset temperature, Ts, on the mean particle size, d:[81] ln Ts = 1/d

Eq. (5-3)

They calculated a ratio of sintering onset to melting temperature that is in good agreement with experimental data for sintering of nonmetals. In 1950, Herring developed the classical scaling rule for the effect of the particle size on sintering time.[82] The time, t, to achieve the same sintering condition is related to the powder particle size, d, if the same sintering mechanisms operate, by the equation:

Processing

184 t1/t2 = (d1/d2)n

Eq. (5-4)

where the exponent n has a value of 3 if the main sintering mechanism is volume diffusion and a value of 4 if the main sintering mechanism is grain boundary diffusion. Considering the Arrhenius expression for densification, the sintering temperature dependence on the particle size becomes n ln(d1/d2) = Q/R [(1/T1) − (1/T2)]

Eq. (5-4a)

where Q is the activation energy for the predominant sintering mechanism, R is the gas constant, d1 and d2 are the different powder particle sizes, and T1 and T2 are their respective sintering temperatures. Reasonable agreement of experimental and calculated data was found if certain diffusion mechanisms were assumed.[63,71,77] For instance, the quantitative agreement in TiO2 was better when grain boundary diffusion was considered as the sintering mechanism.[77] The fitting value for the activation energy used (139 kJ/mol) is close to half of the activation energy for lattice diffusion in TiO2 (251 kJ/mol), a typical ratio for grain boundary to lattice diffusion. Chen and Chen also verified the validity of Herring’s law for grain boundary diffusion mechanisms in CeO2 and Y2O3.[63] For Al2O3 the values calculated by Kumara and Messing are in closer agreement with the experimental values again if grain boundary diffusion control is assumed.[71] But an activation energy for sintering close to volume diffusion (543 kJ/mol) was measured experimentally. When applying Herring’s law to Andrievski’s data[83] on sintering pure nickel particles of different sizes to 60% density, the calculated activation energy value (239 kJ/mol) was closer to volume diffusion (284 kJ/mol) than to grain boundary diffusion (115 kJ/mol) in nickel. However, volume diffusion is less likely to be the principal sintering mechanisms since temperatures are low and particles are fine. These discrepancies point to the uncertainties in using the scaling law to determine the activation energy values and thus to deduce the sintering mechanisms. Conversely, the predictions of sintering temperatures as a function of particle size may not be accurate, most likely giving only qualitative estimates. The main concern in using Herring’s scaling law is related to a possible change of sintering mechanisms from conventional size to nanopowders and with sintering temperature. Sintering at lower temperatures may induce a mechanism shift even if particles are of conventional size.[84] MD simulations showed that Herring’s law does not fully apply to nanoscale powders.[85] For nanoparticles, indications of possible changes in the sintering mechanisms have been already

5: Nanocrystalline Powder Consolidation Methods, Groza 185 shown. Furthermore, in using Eq. (5-4a), the assumption is that particle size ratio is maintained throughout the sintering process. This implies that grain growth trajectories in nanopowders are similar to those in conventional powders, an assumption that may not always hold. One typical factor is powder agglomeration, more likely in nanocrystalline powders, which may induce unpredicted grain coarsening. Finally, as in any extrapolation, a careful inspection of the other initial assumptions is in order. Herring cautioned that the degree of contamination on the surface should be kept the same for all particle sizes. This is the case for MD simulations, but is definitely questionable in experiments. Not only have higher contamination levels been reported,[83] but also different types of oxides have been identified on nanoparticles.[24,25,27] In addition, when sintering occurs at lower temperature, the complete reduction of oxides or hydroxides may be hindered. The specific effect of the surface contamination on sintering will be discussed in the next section.

5.2.3 Role of Impurities As a surface-controlled process, sintering is critically dependent on particle surface conditions. Surface properties become even more dominant in the sintering of nanoparticles. Studies of this topic abound since particle surface properties affect not only individual particle properties and sintering behavior, but also final properties. This explains, for instance, the precautions taken to eliminate contamination by in-situ consolidation of metal nanopowders produced by inert-gas condensation for careful mechanical property characterization[86] (see Chapter 12). In ceramics, sinterability and mechanical properties are also strongly affected by surface impurities. When the inherent surface SiO2 is present on Si3N4 particles, sintering is enhanced but creep strength and toughness are impaired during the service of the Si3N4 part.[87] The excessive surface area of nanoparticles justifies the focus on powder contamination. Using photoacoustic infrared spectroscopy, Ying documented larger amounts of adsorbates on nanopowders than on conventional γ-Al2O3.[88] Andrievski measured about 120 cm3/g of absorbed gases such as N2, H2, and H2O in 50 nm pure Ni powders as compared to 40 and 8 cm3/g for 5 μm and 50 μm powders, respectively.[83] Even in very controlled preparation conditions such as in inert-gas condensation, nanopowders may retain high levels of adsorbates.[86] Generally, the synthesis path has a significant effect on nanocrystalline microstructure and hence sinterability.[2,9,89] For instance, metal powders obtained by

186

Processing

mechanical attrition are inevitably more contaminated than those produced by other processes (e.g., by inert gas condensation). Oxides, nitrides, and other compounds are often found in consolidated parts made of attrition-milled nanopowders.[90,91] Although products of contamination, these compounds may be used to advantage when present as fine dispersions to prevent grain coarsening.[91] Conversely, it was more difficult retain a fine grain size in high-purity powders or during consolidation in high-purity conditions.[86] The influence of contamination on sintering was studied by in-situ TEM under controlled oxygen levels as compared to ultrahigh vacuum conditions.[18] Studies of the early sintering stages in ultraclean conditions have been performed for both metal and ceramic nanopowders.[14,15,92] All results indicate that ultraclean nanoparticles sinter very rapidly even at room temperature. This rapid sintering is in agreement with MD simulation results.[20,21,30,52] As shown in section 5.2.2, neck growth occurs by surface diffusion, dislocation motion, and grain rotation of clean particles. In contrast, when sintering of nanoparticles takes place in the presence of controlled oxygen traces, very little or no neck growth was observed.[18] This oxygen contamination decreases the surface energy and slows down the sintering kinetics. As expected, TEM and X-ray diffraction studies showed that sintering of metal nanoparticles starts only when oxide layers are reduced.[93,94] In ceramics, reducing the amount of oxygen in TiN nanopowders also decreases the sintering temperature.[35] The sintering environment appears to have a similar effect in nanopowders as in conventional ones. Dominguez et al. calculated a lower activation energy for the sintering nano-iron in hydrogen as compared to vacuum.[68] In these results, the lower activation energy corresponds to liquid-like diffusion, while the higher energy is for more conventional grain boundary diffusion. Sintering of Fe-Ni occurred above 500 K in hydrogen, while in vacuum no sintering took place up to 725 K.[93] Consistently, ceramics sintered better–higher densities at same temperatures or similar densities at lower temperatures–in vacuum than in air.[73,74]

5.2.4 Green Density of Nanopowders The initial step in most densification processes is to compact powders at room temperature, or cold compaction, to form a green body. Final sintering results are largely dictated by the green compact structure. A large number of initial point contacts, small pores in a high green density compact, and a uniform pore distribution favor high final density. Sinter-

5: Nanocrystalline Powder Consolidation Methods, Groza 187 ing times may be shorter and lower sintering temperatures may be used. Conversely, many sintering defects may be traced to the green compact structure. Inhomogeneities in density, packing, and particle size in green compacts limit the final sintered density. Such an example is the crack generation in ceramics upon sintering of inhomogeneous cold compacts. Generally, nanocrystalline materials are less forgiving of defects in green compacts than are conventional powders. The most common problem is the elimination of large pores that originate from the green compact. This elimination requires high temperatures upon subsequent sintering, promoting undesired grain growth and loss of the desired nanosize features. Cold compaction comprises specific stages that involve sliding and rearrangement of particles, elastic compression at particle contact points, and plastic yielding for metals or fragmentation for brittle materials. Similar steps are seen in nanopowder compaction. The principal differences in nanoparticles arise from the specifics of small particle sliding and friction, as well as from the changes in dislocation-assisted deformation behavior. Sliding and rearrangement in nanopowders are severely restricted owing to large frictional forces among powder particles.[2,83,95] These forces are the result of mechanical, electrostatic, van der Waals, and surface adsorption phenomena that become much more significant as particle size decreases. Mechanical friction resistance is substantial due to numerous interparticle contact points. Irregular particle boundaries favor agglomerate formation. As a result, particle rearrangement is hindered and lower green densities are likely to be achieved as the particle size decreases (Fig. 5.1). For Y-TZP ceramics, Bourell et al. reported 45% density for 70 nm particles and only 29% for 40 nm.[97] Conversely, when particle sliding is facilitated, high green values are obtained. This may be accomplished by using lubricants and coatings. The latter technique has been predominantly used in ceramic compaction to isolate individual particles by encapsulation or sheathing.[2] As an example, rare-earth carbide or ferromagnetic nanocrystals have been carbon coated to prevent degradation by oxidation or hydrolysis.[98] Although no result on packing or consolidation behavior of coated particles is presented, the carbon surface layer and the spherical morphology thus obtained should improve powder flow and alleviate agglomeration problems. For both metals and ceramics, the lubrication action of liquid nitrogen during low-temperature compaction has been shown to improve rearrangement, resulting in higher green density values (note the cryogenic compaction data point for Al2O3 in Fig. 5.1).[95] As a natural extension, wet or colloidal processing provides significantly improved packing uniformity of ceramic nanopowders.[2,6,99] Tape and slip

Processing

188 80 Ni Si3N4 Al2O3 cryogenic compaction Al2O3 dry pressing

Green Density (%)

70

60

50

40 0.1

1

10 1/d (1/micrometers)

100

1000

Figure 5.1 The effect of particle size on green density upon conventional dry compaction by uniaxial pressing for 50 nm Ni and 50 nm Si3N4.[83] Also shown are data points for dry (open triangle) and cryogenic compaction (open diamond) of 20 nm γ-Al2O3.[95,96] All compactions performed at 1 GPa.

casting, centrifugation, electrophoretic deposition, osmotic and pressure filtration are a few examples of such largely used wet processes.[2,6,7,100–102] As an example, a green density (74%) close to the ideal value that results by packing monosize spheres was achieved in 12.5 nm Y2O3-doped ZrO2 by centrifugation from a slip suspension.[101] A better initial particle rearrangement facilitated by good wetting avoids agglomerates and enhances sinterability even without increasing green density. For instance, Yan and Rhodes achieved 99% density by pressureless sintering of only 41% dense green compact of TiO2 nanopowders prepared by the centrifugation method.[77] Similar full densification with 80 nm final grain size after wet compaction is reported by Mayo in ZrO2– 3 mol% Y2O3.[2] These results are to be contrasted with dry compaction of the same powders, which requires 0.5–1 GPa pressure to achieve similar densification characteristics. More results on cold compaction are discussed in a few reviews.[2,6,8,83]

5.2.4.1 Cold Compaction of Metals To achieve high green densities of metal nanopowders, plastic yielding is necessary which, in turn, requires high compaction stresses. These high stresses are rationalized on the basis of the Hall-Petch dependence of yield stress, σy, upon grain size, d, (i.e., σy ∝ d −1/2).[103] According to the pres-

5: Nanocrystalline Powder Consolidation Methods, Groza 189 sure densification models, plastic yielding takes place when the effective pressure is several times the yield strength of the particle material.[9,72] High stresses may be commonly achieved using high-pressure equipment such as cubic anvil cells. Gutmanas et al. obtained high densities up to 97% at room temperature and 99% at 575 K in Ni (65 nm) and Fe (30 nm) at 3 GPa in a high-pressure cell made of high-speed steel.[104] Historically, high-pressure consolidation has been the preferred in-situ consolidation method for classical nanopowders obtained by inert-gas condensation.[105–108] Specimen size is usually small, not exceeding 1 mm in thickness. Sometimes, densities close to the theoretical have been achieved by low-temperature sintering (e.g., 575 K) after cold compaction under high pressure of nanocrystalline metals (Ni, Fe, Al).[104] The usual dependence of green density of nanometals as a function of applied pressure has two regions (Fig. 5.2). At lower pressures, there is a linear increase of density with pressure. At larger pressure values, the slope is considerably less. This region is attributed to work hardening.[8,103] As seen in Fig. 5.2, little work hardening is experimentally observed, particularly for small particle sizes (15 nm Ni and 26 nm Fe as compared to a flattened curve at high pressures for 50 nm Ni). This is in contrast to submicrometer or micrometer powders for which strain hardening of 200–300 MPa has been observed in cold compaction.[8] The difficulty for strain hardening in nanomaterials is explained by restricted dislocation generation and flow. For a more detailed discussion of the role of strain 50 nm Ni (Uniaxial Pressing) 15 nm Ni 26 nm Fe 50 nm Ni (Cold Isostic Pressing) Ni stored in Air Ni stored in Argon

0.9

Density (%)

0.8 0.7 0.6 0.5 0.4 0.3 0.2

0

0.2

0.4

0.6 0.8 Pressure (GPa)

1

1.2

1.4

Figure 5.2 Pressure effects in the cold compaction of 15 nm and 50 nm Ni, and 26 nm Fe powders (from Refs. 83 and 103); filled symbols, uniaxial pressing; open symbols, cold isostatic pressing. The two separate data points are for air stored (star) and argon stored (full symbol) 50 nm Ni powders.[83]

190

Processing

hardening and high pressure in cold compaction of nanopowders, the reader is referred to Gutmanas’ review.[8] A particular case is represented by mechanically alloyed powders, which most commonly consist of micrometer-sized particles with multigrains in the nanometer range. In this case, the particle size dictates the cold compaction behavior and high green densities are therefore achieved. For instance, a green density of ∼90% has been reached under 1 GPa pressure in mechanically alloyed Fe3Al.[109] As shown in section 5.2.1, the particle surface becomes increasingly important in nanopowders due to the dramatic increase in the surface area. Literature examples illustrate the interdependence of green density value on surface chemistries, adsorption characteristics, and reactivity. Andrievski measured high-absorbed gas levels on 50 nm Ni powders and, consequently, attained low final green densities even at high pressure.[83] As seen in Fig. 5.2, the green density values reached in these powders at 1 GPa pressure by uniaxial pressing are only ∼58%. Even lower green density values were obtained for smaller Ni powders (e.g., 15 nm Ni). Cold isostatic pressing may increase the green density of nanoparticles (open diamond symbols for 50 nm Ni in Fig. 5.2). For comparison, 50 μm Ni powders reached 82% density in uniaxial pressing and 86% in cold isostatic conditions at 1 GPa. The degradation of surface condition by oxidation of metal nanopowders due to storage further limits the green density values obtained. Andrievski showed such a decrease of the final green density of metal nanopowders after storage (see the two data points for stored Ni in Fig. 5.2). Also noteworthy is the minor density difference after storage in air and argon.

5.2.4.2 Warm Compaction To eliminate adsorbates and enhance interparticle bonding, warm compaction at temperatures up to 675 K has been widely applied to nanopowders. For most metal nanopowders, these temperatures are still in the cold working regime. Selection of such temperatures below recrystallization is purposely sought to prevent grain growth.[104] Only for low-melting-point materials may compaction at room temperature induce recrystallization and sintering effects. For instance, in a Bi–Sb system, cold pressing at 350 MPa yielded an 89% dense specimen while grain size doubled.[110] Significant work in warm compaction has been carried out for metal nanopowders by Weertman and coworkers.[86,111–113] The powders obtained by inert-gas condensation were densified under high pressure application

5: Nanocrystalline Powder Consolidation Methods, Groza 191 (1.4 GPa) for 5–900 min at temperatures between 375 and 575 K.[110,112] Densities obtained exceeded 95%. Higher densities were obtained when volatile contaminants were carefully eliminated based on studies of gas desorption temperatures and kinetics.[86,113] The main purpose of such studies was to minimize the consolidation artifacts, which hindered the accurate characterization of the mechanical properties of nanomaterials. Ceramic nanoparticles have also been densified under high pressures at low temperatures. Warm compaction at 675 K of 70–80 nm TiN powders resulted in densities exceeding 90%.[114] Magnetic pulses with durations of 100–500 μs and amplitudes up to 2.5 GPa were applied to consolidate metal composites and ceramics to high green density values. For instance, Al/Al2O3 nanocomposites were compacted by magnetic pulses applied between 293 and 673 K.[115] Densities achieved in Al2O3 and ZrO2 by this method are 80–82%, about 15% higher than in conventional uniaxial cold compaction.[116]

5.2.4.3 Cold Compaction of Nanocrystalline Ceramics In dry compaction, ceramic powders are more difficult to compact than metals (compare Ni and Si3N4 compaction lines in Fig. 5.1). An equation to calculate the pressure necessary to achieve a certain density in cold compaction based on volume changes with pressure was developed by Chen and Malghan.[96] Experimentally, the literature reports indicate room temperature densities reaching >95% for nanometals but only 75–90% for nanoceramics.[117] Only occasionally do ceramics pack better than metals by compaction at room temperature (see Al2O3 data points in Fig. 5.1). These high densities were achieved by applying a very high pressure of 1 GPa. Indeed, similarly to metals (Fig. 5.2), green density increases with pressure applied. Conventional cold compaction results for 70–80 nm TiN nanopowders show that green densities up to 65% may be reached. At very high pressures (7 GPa), the density obtained was close to 80%.[114] In cold isostatic pressing at very high pressures (up to 5.6 GPa), densities in excess of 90% for nanocrystalline Al2O3 and 80% for SiO2 have been achieved by Gallas et al.[117] A major inconvenience in applying high pressures, other than equipment limitations, is the high level of residual stresses in ceramic powders that may result in fracture upon subsequent handling.[2] At lower pressure values, the hard agglomerates formed in ceramics cannot be fractured. To break these agglomerates, a critical pressure is necessary (Fig. 5.3). On a density–pressure plot, the transition point, Py,

192

Processing

Figure 5.3 Compaction behavior of ceramic powders (data from Refs. 118 and 119). Curve (1), weakly agglomerated ZrO2–17% Y2O3; (2), hard agglomerates in 5%Y-TZP.

where the slope change occurs, is interpreted as the strength of the agglomerates (curve 1 in Fig. 5.3). The Py pressure level is dependent on the agglomerate type. The weak agglomerates break at lower pressures than the hard ones.[120] When agglomerates are strong and dense, they do not break and, consequently, no transition point is seen (curve 2, Fig 5.3). This is the case of the agglomerates in calcined water-washed yttria-stabilized zirconia that survive a cold isostatic pressing at 400 MPa.[118] If powders contain no or weak agglomerates, the final result of the pressure application is a uniform distribution of small pores with favorable consequences for final sintering. Such a gradual change from a bimodal pore distribution in agglomerated powders to only small pores by pressure application has been experimentally documented by Van de Graaf et al. in 8 nm 17% Y2O3–ZrO2 ceramics using porosimetry techniques.[119] A high value of the green density may not always reflect a uniform pore size and distribution. When large agglomerates are present, the green density may be larger than in nonagglomerated powders, similarly to bimodal size powder packing. For instance, a high density of 75% in ascompacted nano-TiO2 at 425 K under 2 GPa pressure for 2 hours was

5: Nanocrystalline Powder Consolidation Methods, Groza 193 explained by easier packing of large agglomerates (50 nm) rather than individual smaller particles (14 nm).[46] For practical purposes, transparency of green ceramics may be used as a quick check of the compaction results.[2,117] A translucent green compact is an indication that pores are smaller than the visible light wavelength and consequently may reach high densities upon further densification.

5.2.5 Pore Size and Its Effects on the Densification Behavior Similar to conventional powders, full density or rapid sintering of nanosize powders is achieved when the green structure contains a narrow pore size distribution. Conversely, densification is retarded or inhibited when pore size distribution is wide. In this case, big pores become larger and only small pores shrink. The removal of large pores is a lengthy process and requires higher temperatures. Therefore, the overall effect of large pores is to slow densification and induce undesirable coarsening. A thermodynamic model of the pore shrinkage was developed by Kingery.[121] Based on curvature considerations, he predicted the pore size for the transition between pore shrinkage and pore growth. A complicating event is the separation of large pores from the grain boundaries that occurs in final sintering stages. Brook developed pore size-grain size maps to designate the regions where such pore separation from grain boundary occurs.[122] When pore–grain boundary breakaway takes place, the detached pores will no longer benefit from easy transport paths such as grain boundaries. Instead, transport is by slower volume diffusion and, consequently, densification occurs at a slower rate. This is the stage when open porosity breaks down and pores become closed. Since grain boundary migration is no longer restricted by pores, grain coarsening takes place; or pinning of grain boundary migration due to residual pores is no longer effective. Therefore, in the final sintering stage, densification is hindered and significant grain growth occurs. Mayo et al. extended Kingery’s treatment to second-stage sintering of nanoceramics and demonstrated that they also obey Kingery’s critical ratio for pore shrinkage.[123,124] Full densification is likely to occur when all pores are smaller than the critical pore-to-grain size ratio. Large pores are more likely to undergo the pore-boundary separation, which prevents high densities to

194

Processing

Figure 5.4 Exaggerated grain growth in TiO2 when porosity becomes of closed type (square symbols) (from Ref. 125) and the effect of dopants on grain growth of TiO2 (triangles) (from Ref. 126).

be achieved. The effect of large pores from a nonuniform green structure on final sintering of nanopowders is even less forgiving than for normally grained powders. If coarsening occurs in late sintering stages, the overall notion of nanograin size is compromised. Experimentally, a strong correlation between the closure of open porosity and the onset of exaggerated grain growth was noted in both ceramic and metal nanopowders.[46,73,81,123–129] As exemplified in Fig. 5.4 for pure TiO2, accelerated grain growth up to about micrometer grain size occurs at densities above 90% or when the pores become closed. Large pores usually originate from agglomerated powders. As already shown, nanoparticles are inevitably linked to a high agglomeration tendency. Although agglomerates are more often reported in ceramic nanopowders, they have been also observed in metals.[67,129] Trusov et al. gave experimental evidence that aggregates of ∼1 μm may be found in 10 nm metallic powders.[67] Agglomerated powders have a bimodal pore distribution with small interparticle and large interagglomerate pores (Fig. 5.5(a)). The removal of interagglomerate pores based on vacancy diffusion requires significantly higher temperatures and longer sintering times. Mayo et al. compiled evidence for three TiO2 powders, showing that sintering temperature scales with agglomerate size, rather than with size of the individual powder particle (Fig. 5.5(b)). Also noteworthy in Fig. 5.5(b) is that the smaller the particles, the larger the agglomerates. The nonagglomerated powders sinter at the lowest temperature, despite their largest

5: Nanocrystalline Powder Consolidation Methods, Groza 195

a

b Figure 5.5 Schematic of pore sizes in agglomerated powders. (b) The effect of agglomerate size on the sintering temperature of nano-TiO2. Agglomerate size (particle size): (1) nonagglomerated (<40 nm); (2) 80 nm (16 nm); (3) 340 nm (8–10 nm). (After Ref. 123, with permission from Elsevier Science.)

nanoparticle size (see nonagglomerated (<40 nm) data points in Fig. 5.5(b)). As early as 1981, Rhodes demonstrated that deagglomeration of a commercial ZrO2 reduced sintering from 4 hours at 1775 K to 1 hour at 1375 K.[101] The benefits of a small pore size and narrow size distribution in reaching high densities and reduced grain coarsening have been shown by many researchers in nonagglomerated Al2O3,[71,75] Y2O3 and ZrO2,[73,74,130] SnO2,[131] and TiN[35] nanopowders. Zhao and Harmer performed model experiments to demonstrate rapid densification when increasing the number of pores.[132] Lately, more

Processing

196

models have considered the pore size effects on the densification rate in nanocrystalline materials,[2,133–135] as was suggested by Kuczinski in 1979.[136] Mayo developed a modified sintering law that directly accounts for pore size effects on densification rate, dρ/dt:[135] dρ −Q ⎞ 1 1 1 ∝ n exp⎛ ⎝ RT ⎠ ρ(1 − ρ) dt d r

Eq. (5-5)

where ρ is the density, d is the particle size, n is a constant dependent on the sintering mechanism, r is the pore radius in the green compact, Q is the activation energy, R is the gas constant, and T is the absolute sintering temperature. This equation predicts fast sintering kinetics when pore sizes are small. It was found to hold throughout the sintering process and for a large range of initial pore sizes, at least when pore size distribution is uniform. However, the densification rate is also dependent on pore size evolution, not only the initial pore size.[134,137] To maintain a fast sintering rate in late sintering stages, the pores should remain small even in late sintering stages. A small pore size throughout the sintering process is also critical in controlling the final grain size, as will be detailed below.

5.2.6 Grain Growth Success in nanopowder consolidation is intimately related to the control of the competition between densification and coarsening. As sintering progresses and grain boundaries are created, the driving force, Δp, for mass transport may be expressed as a Gibbs–Thomson effect:[138] Δp =

2γ b 2γ ± d r

Eq. (5-6)

where γb is the grain boundary energy, d is the average grain size, γ is the surface free energy, and r is the radius of curvature of pore surfaces, which may be positive or negative. In a simple form, the first term is the grain coarsening tendency, while the second is the sintering driving force. When pores are small, sintering predominates or the pores control grain growth. When particles bond and grain sizes are in the nanosize range, the driving force for coarsening becomes important.

5: Nanocrystalline Powder Consolidation Methods, Groza 197 The thermal stability of nanocrystalline materials has been the subject of recent reviews[139,140] or has been covered by a number of general reviews.[9,10] Here the attention is mainly on grain growth during sintering. In nanomaterials, grain coarsening is described by a power law equation similar to that of conventional materials: dn − d0n = kt

Eq. (5-7)

where d is the average grain diameter, d0 is the initial grain diameter, n is the grain growth exponent, k is an Arrhenius-type constant, and t is time. Koch and co-workers summarized the n values for the isothermal grain growth of nanocrystals and compared them to those for conventional polycrystalline materials.[140] At low temperatures, n values are high or grain growth is slow, similar to regular materials. Such increased n values are rationalized by restricted grain boundary mobility, which is most commonly due to pore or solute effects. In nanocrystalline materials such as iron, Malow and Koch found an average n value of 12 at low temperatures but the n value decreased with increasing temperatures in the same way as for the coarse-grained materials.[141] In ZrO2–3 mol% Y2O3 ceramics, n is larger than 4 at low temperatures and decreases to 3 at intermediate temperatures.[2] In other cases, constant n values were reported, e.g., n = 4 in grain growth during sintering of Al2O3,[142] or n ≈ 3 in TiO2 and yttria-stabilized ZrO2.[2,48,143] Activation energies for grain coarsening are generally difficult to assess, particularly when the exponent n changes. Malow and Koch compiled the existing activation energy data for grain coarsening in consolidated nanosize metals, intermetallic compounds, and ceramics.[140] In most cases, the activation energy is close to that of grain boundary diffusion, which is similar to that for large-grained materials. Some exceptions were found for elemental nanometals with activation energies equal to that of lattice diffusion.[141,144] Two apparent activation energy values were calculated in nanocrystalline Fe: a low value at low temperatures (125 kJ/mol) and a high value (248 kJ/mol) corresponding to high temperatures.[141] The latter value was close to that of volume diffusion in iron and is similar to the value found in coarse-grained iron. The low-temperature value is less than that of the activation energy for grain boundary diffusion. A low activation energy value for grain coarsening in early sintering stages was also calculated by Thunissen et al. in ZrO2–5.8 mol% Y2O3, although some uncertainty may be due to an unknown n value.[48] Eastman reports a low activation energy for grain growth (83 ± 40 kJ/mol) in nano-TiO2.[143] However, any speculation on a new mechanism based on these low

198

Processing

activation energy values is risky because of the reported uncertainties in impurity or porosity levels, as well as the lack of specific diffusion data in nanomaterials for direct comparison. The influence of pores on grain coarsening has been well documented theoretically[99,133,135,145–149] and experimentally, particularly in ceramics.[46,123,150,151] As already shown, the presence of open pores in nanopowders inhibits grain growth in a similar way that pores prevent grain coarsening in ceramics of ordinary grain size. As soon as the pore isolation occurs, grain growth becomes unrestricted (Fig. 5.4), but the pinning action of the pores is difficult to predict. It follows the same principles as in coarse grain materials. For instance, Gupta established an empirical direct relationship between density and grain size in intermediate stage sintering.[152] Liu and Patterson found a linear relationship between the inverse of grain size and the pore surface area per unit volume.[153] Grain growth was specifically modeled in nanoparticle sintering by Iyer and Sastry.[134] This model predicts the grain growth behavior in sintering of nanocrystalline compacts powders with bimodal pore size distribution (i.e., for the agglomerated powders shown in Fig. 5.5(a)). Similarly to classical microstructural development upon sintering, this model shows negligible grain growth in early sintering stages even near the large interagglomerate pores but an accelerated coarsening in the final sintering stages (i.e., grain size are stabilized by pores). The simplest way to control grain coarsening is to take advantage of the pinning effect of open pores and limit the density to ∼90%, if this density is acceptable for the final product. To limit grain size beyond 90% density, a small and uniform pore size distribution in the green compact is required. As already shown, small pores enhance densification as densification rate scales with the inverse of pore size (Eq. 5-5), while a large number of pores (i.e., many small pores rather than few large ones) pin the grain boundaries in an effective manner.[2,124,131,141] Other means to control grain growth after pore closure include kinetic and thermodynamic approaches.[10] One effective strategy is to decouple grain boundary diffusion (important for sintering) from grain boundary migration (responsible for grain growth) in final sintering stages.[154] The approach is a two-stage sintering process, first to bring the ceramics to a temperature at which pores are unstable against shrinkage and then to decrease the temperature. In the second step, densification continues by eliminating the unstable pores while retaining the small grain size achieved in the first step. Experiments on sintering 30 nm Y2O3 showed a strong grain size dependence of the normalized sintering kinetics (Eq. 5-2), implying grain boundary diffusion control. Studies of grain growth in the second stage

5: Nanocrystalline Powder Consolidation Methods, Groza 199 identified a kinetic “window” to achieve densification at constant grain size structure. At temperatures below this window, densification is not complete, while grain growth occurs at higher temperatures. The latter implies that the activation energy for grain boundary migration is higher than for diffusion along the boundary. Therefore, densifying the body at constant grain size may occur at some sufficiently low temperatures capable of eliminating the residual pores. A different approach to slow down the coarsening kinetics is by restricting grain boundary mobility using a solute drag or particle pinning effect. Small particles or second phase precipitates stabilize a fine grain size upon heating at high temperatures, as long as particles do not dissolve or coarsen. The maximum grain size (dmax) is dependent on the particle size, rp, and volume fraction, f, by a relationship given as Zener drag: dmax =

4rp 3f

Eq. (5-8)

The solute drag effect is based on the decrease of grain boundary mobility and free energy when solute segregates at grain boundaries. Theoretically and experimentally, it was shown that grain boundaries in nanomaterials have increased solubilities for solute atoms.[10,155,156] The enrichment of the grain boundary with solute atoms has been shown to diminish or even reverse the free energy available for grain growth, as described by Weissmuller (i.e., segregation at grain boundaries reduces the grain boundary specific energy, σ, to zero or negative values[10,156]). In these cases, the nanocrystal becomes thermodynamically metastable with respect to grain growth. The experimental verification of the grain growth metastability theory was performed in the Y—Fe system and a good stability of grain size was found in alloys that contain up to 20 at% Fe.[155] Experimentally, dopants in ceramics have been widely used to suppress grain growth by forming secondary phase particles or providing a drag force on the moving grain boundaries.[6,46,126,150,157–159] For instance, Hahn et al. found that 6% Y addition to TiO2 retarded grain coarsening up to >95% density, thus retaining a grain size less than 100 nm[126] (see Fig. 5.4). Terwilliger and Chiang showed that Ca and Sn are grain growth inhibitors in 95% dense nano-TiO2.[157] Perhaps one of the most dramatic effects of dopants is the addition of 6% Ca to CeO2, which retained a 30 nm grain size after full sintering by heating at 10 K/min to 1625 K.[159] In the same sintering conditions, undoped CeO2 reached a grain size of 400 nm. Grain boundary chemical segregation of 5% Nb was very effective in suppressing grain growth of milled Fe3Si.[160] Promotion of

200

Processing

chemical ordering in intermetallics (Ni3Al, Fe3Si and (Fe, Mn)3Si) also retards grain growth by reducing grain boundary mobility.[161] The dopant level required to limit the grain boundary migration has recently been quantified.[6] For instance, 7000 ppm MgO dopant is needed to restrain grain growth to 50 nm in alumina. Further understanding of the doping effects on grain boundary mobility and their interplay with initial powder synthesis and agglomeration is a fruitful area of further research with high potential for a better control of final grain size. If no dopants are desired because of their negative effect on the final properties (e.g., on dielectric[77] or mechanical[87] properties), sintering under pressure may be another effective means to suppress grain growth. This approach will be discussed in sections 5.3.3 and 5.3.4.

5.3 Methods for Full Densification of Nanopowders 5.3.1 Characterization of Nanomaterials Densification: Density and Grain Size Measurements To investigate any consolidation results, two main characteristics are usually examined: density and grain size. In analyzing different nanopowder densification methods, caution is required when comparing reported density and grain size values because of the diversity of measurements methods and their inherent experimental errors. Some issues related to these measurement techniques for nanoparticles and theoretical density values will be briefly addressed below.

5.3.1.1 Density Measurements The density may be displayed as a physical value or as a percentage of the theoretical density. The theoretical density value of nanocrystalline materials is controversial. Earlier arguments were that a lower density value in nanocrystals may arise due to a large volume fraction of grain boundaries, which in turn may have lower densities than adjacent crystals (e.g., Refs. 162–165). Another source of theoretical density error is the uncontrolled presence of a secondary phase such as remaining oxides in metals or dopants in ceramics (e.g., Ref. 90) Experimentally, density measurements are usually performed using Archimedes’ principle. The

5: Nanocrystalline Powder Consolidation Methods, Groza 201 errors associated with this method derive mainly from thermal and surface effects, presence of pores, and specimen size. The reproducibility of density values is within ∼2%. To increase the accuracy of density measurements, some higher-density immersion liquids (e.g., ethyl phthalate or methylene iodide) may be used. Larger specimens result in less density error. Prior impregnation is required for specimens with interconnected porosity.[166] The ultimate indication of pore elimination is by positron annihilation spectroscopy. This technique is very sensitive to ultrafine pores that are otherwise difficult to observe even in TEM. It has been used to study densification by monitoring the pore sizes as a function of sintering parameters.[167] Further information on nanopores and their distribution may be furnished by small-angle X-ray and neutron scattering, and porosimetry techniques such as Brunauer, Emmett, Teller (BET) nitrogen adsorption methods.[2,35,48,124,127,168] Porosimetry techniques are suitable to measure only open porosity. More sophisticated pore size characterization has been performed using three-dimensional scanning tunneling microscopy (STM) in cold compacted metals produced by the condensation method.[169]

5.3.1.2 Grain Size Measurements This type of measurement is commonly carried out by X-ray diffraction (XRD) and electron microscopy techniques (TEM or high resolution SEM). In TEM and SEM, grain sizes are measured directly and grain size distribution may also be obtained. In contrast, XRD techniques provide only an average grain size. They rely on the dependence of the broadening of Bragg reflections on grain size. Corrections for instrumental broadening must be made. Scherrer analysis is straightforward and assumes that the broadening of X-ray peaks is due only to the small crystallite size. Other techniques (e.g., Warren–Averbach (WA), Halder–Wagner, or Williamson–Hall) are more sophisticated and account for some other factors such as broadening caused by internal stresses. For instance, the WA analysis may distinguish between broadening due to grain size and to intrinsic faults and twins. Therefore, measurable differences may be found between Scherrer (larger values) and WA (smaller values) results and also when using different X-ray peaks for grain size calculations. The Scherrer method yields a volume-weighted average, while the WA yields an area-weighted average grain size. Typically, these two averages differ by a factor of 2.[10] Nieman et al. report a smaller error (10–25%) in grain size determination by the WA method for grain sizes up to 25 nm, but the error may be up to 100% when the grain size exceeds 50 nm.[106]

202

Processing

Generally, average TEM grain size and X-ray results are in good agreement, particularly for the WA analysis with strain and fault corrections.[10] As expected, the best agreement is found when grain size distribution is narrow (e.g., Ref. 72). When this is large, the TEM method is preferable. Sometimes, discrepancies may occur between grain size values by microscopy and XRD techniques. For instance, the Halder–Wagner XRD value in B4C refers to the twin size, while TEM measurements reflect the true grain size.[170] More details about grain size measurement methods in nanocrystalline materials may be found in Refs. 10 and 139.

5.3.2 Conventional Sintering Both ceramic and metal nanopowders have been fully densified by conventional (pressureless) sintering. As pointed out in section 5.2, sintering of nanoparticles is expected to take place at significantly lower temperatures due to enhanced driving force and kinetics in nanosize powders. These lower temperatures may minimize or even eliminate the use of sintering additives or grain growth inhibitors. For instance, Siegel et al. sintered nanophase TiO2 at a temperature 600 K lower than for conventional sintering, thus avoiding the need for sintering aids.[171] In some other systems such as WC-Co, the low sintering temperatures required to consolidate nanopowders may render the use of grain growth inhibitors unnecessary.[45] However, little research has been performed in understanding the interplay of dopants and pore effects in inhibiting grain growth. A better understanding of the mechanisms by which grain growth is inhibited may result in significant payoffs.

5.3.2.1 Ceramic Sintering The benefits of fine grain size in ceramic sintering were reported almost 40 years ago when the sintering onset for very fine-grained MgO was found to be as low as 625 K.[1] Low-temperature sintering was achieved in fine-grained ceramics in the 1980s.[77,101,172,173] Rhodes sintered the centrifugally cast ZrO2–6.5 mol% Y2O3 to 99.5% density at 1375 K to a final grain size of 200 nm.[101] Colloidally processed alumina sintered to 99.5% density at 1425 K.[173] Similar attempts to sinter ZrO2 achieved 98% density at 1493 K when Y2O3 was the additive and 93% at 1208 K when the additive was Bi2O3.[172] In both the latter cases, the final grain size was 600–700 nm. Lately, intensified efforts on sintering nanograined ceramic

5: Nanocrystalline Powder Consolidation Methods, Groza 203 powders (ZrO2, Al2O3, CdO, and TiO2) by pressureless sintering resulted in fully dense ceramics with restrained grain growth. The sinterability of ceramic nanopowders has been greatly enhanced by improvements in nanopowder synthesis, particularly the control of nanopowder agglomeration.[2,71,173–177] Examples of dense ceramics that have retained nanosize grains up to 100 nm by conventional sintering are provided in Table 5.3. As seen in this table, full densification without grain coarsening has been achieved predominantly in the ZrO2 system. This system has received substantial attention for two main reasons:[2] first, ZrO2 has the slowest coarsening tendency among ceramics; second, the pore size does not seem to affect grain coarsening. Certainly, a better understanding of sintering and coarsening competition in other systems will improve our ability to fully densify them while retaining nanosize grains. Other strategies to control grain size evolution in nanocrystalline ceramic bodies are based on phase transformations and control of heating rates. In the sol-gel titania, the final nanometer grain size is attributed to a transformation-enhanced sintering.[80] In contrast, Hague and Mayo report that phase transformation during sintering of titania promoted grain growth.[178] A complicating factor in this comparison is powder agglomeration. The latter powders were agglomerated, while the former had a low degree of aggregation. The exact role of phase transformation and powder agglomeration in pressureless sintering is certainly worth further attention. One such worthwhile area is the possibility of enhancing the nucleation rate during phase transformations by chemical manipulations upon powder synthesis (most likely by minimizing the interfacial energies).[80] Control of heating rates throughout the densification process, known as rate-controlled sintering, has been successfully used to retain nanograin size in some fully dense ceramics (e.g., 75 nm in 3YSZ and 50 nm in TiN).[179] In conventional sintering of coarsely grained ceramics, high densities and smaller grain sizes may be achieved by using fast heating rates in systems with higher activation energy for densification than grain growth. The explanation is based on bypassing the lowtemperature range where the effect of surface diffusion is predominant and grain coarsening takes place. The theoretical predictions require heating rates on the order of 100 K/s to avoid coarsening by surface diffusion.[180] Results consistent with this theory were achieved in sintering alumina and titania nanoparticles to high densities (95%) by using a heating rate of ∼100 K/s.[70] Sometimes, the effect of heating rate on nanoceramic densification is more controversial. On one hand, no effect on grain size of heating rate in the range 2–200 K/min and retarded densification was observed in 35 nm yttria-stabilized zirconia.[54,181] On the

c

b

a

NR, not reported. Centrifugally consolidated. Sol-gel prepared.

TiO2 TiO2c TiO2–6% Y

CeO2 CeO2–6 at% Ca ZrO2 ZrO2 ZrO2–3 mol% Y2O3 ZrO2–3 mol% Y2O3 ZrO2–3 mol% Y2O3b ZrO2–5.8 mol% Y2O3 ZrO2–5.8 mol% Y2O3

Material

1523 1623 1400 1250 1343 1273 1373 1273 1323 1173 873 1173

8 14 ∼6 ∼15

Temperature (K)

7 15–20 NR 15–20

2 10°C/min. 1.3 0.67 0.2 4 1 100

Time (h)

Sintering Parameters

10–15 10–15 6–9 6–9 6 8 <10 8

Initial Grain Size (nm)

Air Oxygen NR Oxygen

Air Air Air Vacuum NRa Air NR Air

Atmosphere ere

96 ∼90 99 ∼98

Near full Full Full Near full Near full 99 99.9 >95

Density (%)

Final Properties

60 50 <60 30

∼100 30 80 60 <100 80 80 80

Grain Size (nm)

Table 5.3 Density and Grain Sizes in Nanoceramics Consolidated by Conventional (Pressureless) Sintering

48 46 80 126

159 159 74 74 176 177 2 48

References

204 Processing

5: Nanocrystalline Powder Consolidation Methods, Groza 205 other hand, sintering with a similar heating rate (>100 K/min) resulted in high densities and small grain sizes in the same zirconia system.[182] A practical problem in the fast sintering of low-thermal-conductivity nanoceramics is flaw formation as a consequence of thermal gradients.[54] These results, along with enhanced sintering at low temperatures, emphasize the complex and unclear role of surface diffusion in the sintering of nanopowders, which certainly warrants further studies.

5.3.2.2 Metals Conventional sintering has been applied less to nanosize metals and intermetallics than to nanoceramics. Full densities have been achieved in a few cases such as pure Ni, Cu, and Fe-carbide alloys.[94,183,184] Pure Cu was sintered to 92% density and 70 nm grain size.[94] Rate-controlled sintering of pure Ni resulted in 99% dense specimen with 70–80 nm grain size.[183] For some other metals, the final grain size was 500–600 nm since sintering was carried out at regular sintering temperatures (>0.5Tm). Consequently, warm compaction has been the preferred way to combine high densities and grain sizes less than 100 nm in nanometals, as shown in section 5.2.4. Nanograined WC-Co powders have been fully consolidated with minimal grain growth by conventional sintering, but the final grain size was on the order of 100–200 nm.[185–187] This final grain size is mainly attributed to the liquid-phase mechanism, which typically promotes grain growth. Approaches to restrict grain growth in WC-Co are the use of inhibitors or exploration of the sintering temperature depression in the nanoregime to possibly achieve densification only in the solid state.[45] Intermetallics have been more commonly densified into nanocrystalline structures, such as TiAl sintered to >95% density at low temperatures (0.4Tm) with grain size less than 15–20 nm.[188,189] Al3Zr-based compounds were sintered to full density and grain sizes less than 40 nm in a biphasic structure.[190]

5.3.2.3 Nanocomposite Densification Handwerker et al. summarized the sintering behavior of ceramic composites.[173] Three different behaviors are identified in the sintering of viscous matrices. The common rule of mixture applies only when the volume fraction of the second phase is less than 10%. Scherrer’s theory for viscous sintering with rigid inclusions applies between 10% and 12%. At higher volume fractions, deviations from both models are observed.

206

Processing

For crystalline matrices, densification rates are lower than theoretical at all volume fractions, indicating that the second phase retards matrix densification. Sintering of nanocomposites seems to conform to the same mechanisms. For instance, Prabhu and Bourell observed that the experimental shrinkage for a nanocomposite of ZrO2–3 mol% Y2O3 with 20 wt% Al2O3 was lower than for the zirconia matrix.[191] The densification rate of the nanocomposite is not a linear function of the densification rate of the two components and each phase in the mixture has its own densification and coarsening rate. The nanocomposite sintered to a density of 97% at 1573 K for 1 hour only using hot pressing, while monolithic zirconia sintered to a density of 98% density at 1498 K. The grain size exceeded the nanometer range (200–300 nm). Similarly, the addition of SiC particles delayed the densification of the Al2O3 matrix.[192] In ZrO2—Al2O3 systems, a homogeneous distribution of alumina second phase maintained a grain size of 40 nm in nearly fully dense composites.[193] Similarly, in a Cu— Nb composite, the Nb phase successfully restricted grain growth upon conventional sintering, which achieved a high density and retained a grain size of 100 nm.[194]

5.3.3 Pressure Effects in Nanopowder Consolidation The classical temperature–pressure trade-off for sintering brings distinct advantages to nanopowder densification due to the ability to restrict grain growth. Numerous examples of pressure-assisted sintering of nanopowders demonstrate enhanced densification with reduced grain growth as compared to pressureless (conventional) sintering.[46,72,97,130,158,196–199] An illustration of the pressure effects during nanopowder densification is shown in Fig. 5.6. An applied stress adds a new component to the curvature-related driving force for densification. The applied mean stress, known as the hydrostatic stress, induces new plasticity-driven sintering mechanisms, as well as a stress-assisted diffusion mechanism. Both mechanisms enhance the densification rate. The shear component of the applied stress causes particle rearrangement and collapse of large pores. The particle rearrangement and macroscopic deformation of pores increase the number of particle contacts. This different pore size and distribution is more amenable to pore removal than in pressureless sintering.[99,151] The main benefit is the elimination of large pores in agglomerated powder, which is the cause of exaggerated grain coarsening at densities above 90% (see sections 5.2.5

5: Nanocrystalline Powder Consolidation Methods, Groza 207 100

Relative Density (%)

95 90 85 80 75 70 65 60

0

10

20

30

40

50

60

70

80

90

100

Pressure (MPa)

Figure 5.6 Pressure effect on the densification of nanoceramics: density improvement with the increase in the applied stress during sinter forging of yttria stabilized tetragonal zirconia (Y-TZP) at 1275 K for 20 min. (Data points from Ref. 199.)

and 5.2.6). Generally, the capability to collapse the pores in pressureassisted sintering scales with the shear stress level. This stress is minimal in hot isostatic pressing (HIP) and increases gradually from quasi-isostatic pressing to uniaxial pressing in a die (hot pressing), to uniaxial pressing without a die (sinter-forging), and finally to extrusion.[200] A shear stress is also beneficial for mechanical disruption of surface oxide layers, which provides better interparticle contacts for subsequent bonding. In nanocrystalline powders, the effect of the applied stress is noticeable only if it exceeds the intrinsic curvature-driven sintering stress. Skandan et al. analyzed the contributions of the intrinsic and external pressure.[74,130,201] The applied pressure is independent of the particle size. The intrinsic sintering pressure increases when the particle size is reduced and may reach high values as the particles become increasingly small (Eq. 51). For small particles, the applied pressure has to be higher than the curvature-driven pressure. Consistent with this threshold effect, no external pressure effect was observed in the densification of ZrO2,[74,125] TiO2,[123,202,203] or CdO[204] unless a certain pressure level was reached. The intrinsic curvature stress in nanopowders (Eq. 5-1) may be on the order of hundreds of megapascals. However, the experimentally observed threshold pressure in nanopowders is an order of magnitude less than the calculated value.[74,123,203] This discrepancy may suggest a wide particle size distribution with large particles, which may give only a small

208

Processing

contribution to the intrinsic sintering stress. Alternatively, grain coarsening may occur by the time the pressure is applied, thus lowering the intrinsic sintering stress. Finally, pore size must be also considered along with pressure effects. Yier and Sastry showed that the threshold stress concept applies only when pore and grain size are similar.[134] When pore size is larger than grain size, pressure application is beneficial to break the pores and enhance densification.[137] This is the case with agglomerated powders discussed in section 5.2.5. Large interagglomerate pores (Fig. 5.5(a)) are more difficult to fracture even at very high pressures when the temperature is low. Therefore, pressure application is deliberately carried out only at high temperatures where the resistance to deformation is less and the pores are more likely to collapse.[74,123,205,206] For instance, Schwarz et al. achieved full density and retained fine grain size (44 nm) in mechanically alloyed Al5CuZr using a hot forging technique with pressure applied when the system achieved the sintering temperature.[205] The classical pressure-assisted sintering diagrams developed by Ashby and co-workers[207,208] have been used to describe nanopowder consolidation.[72,97,209,210] Yier and Sastry extended Ashby’s model and developed a comprehensive micromechanistic model for the pressure densification of nanocrystalline powders.[134] Ashby maps reveal changes in the densification mechanisms of nanocrystalline materials.[97,209] For the same pressure level, Bourell et al. indicated a change from stress-assisted densification in coarse grained Y-TZP to curvature-driven grain boundary diffusion in nanocrystalline zirconia (Fig. 5.7).[97] Similarly, McKimpson showed an enhanced boundary diffusion contribution in the hot isostatic pressing of nanosize (Al, Cr)3Ti, Cu, and Al2O3 powders.[209] However, comparison with experimental HIP results indicate that the calculated densification rates are consistently overestimated. For instance, 20 nm Al2O3 hot isostatically pressed for 1 hour at 1625 K under 300 MPa pressure resulted in a sample only 84% dense. Among the possible reasons is the retarded densification due to large pores in agglomerated powders. HIP induces only local shearing stresses, and at lowest levels among the pressureenhanced sintering methods. However, even these small deviations from a purely hydrostatic stress in HIP have been shown to contribute to enhanced sintering rates.[210] Shaik and Milligan modified the Ashby model for application to rapid sinter forging.[72] They modified the stress state and also used the experimental green density (0.8) as compared to the theoretical value (0.64) used by Ashby. Noteworthy is the good agreement with the experimental results for milled Fe—Cu nanopowders when grain boundary diffusion was considered to be the mass transfer mechanism in nanograined powders within micrometer particle sizes. Creep was the dominant densification mecha-

5: Nanocrystalline Powder Consolidation Methods, Groza 209

a

b Figure 5.7 Densification maps of regular size (a) and nanocrystalline Y-TZP powders (b). (Courtesy of D. L. Bourell.)

nism. The contributions of yielding and diffusion rates were about two orders of magnitude lower than those for creep densification. When the applied pressure is high, such as in the piston–cylinder method or explosive forming, plastic deformation may contribute to densification of nanocrystalline materials, despite their higher yield stress values than for

210

Processing

conventional materials. Other mechanisms, such as grain boundary sliding and grain rotation, may also play a role in nanopowder deformation sintering (e.g., in sinter forging).[125,193,211] Grain boundary sliding is assumed to be the main densification mechanism in severe plastic deformation consolidation.[58]

5.3.4 Pressure-Assisted Consolidation Methods With only a few exceptions, all known pressure-assisted consolidation methods have been applied for nanopowder densification: hot pressing, sinter forging, HIP, extrusion, and high-pressure techniques. Hot pressing and sinter forging involve a uniaxially applied pressure. In HIP, a fluid transmits a hydrostatic stress. Variants of HIP with quasi-isostatic stress distribution may be obtained if a fluid-like pressure transmitting medium is used such as in the Ceracon process.[200,212]

5.3.4.1 Hot Pressing Hot pressing gives distinct advantages in comparison to pressureless sintering in achieving full densities and minimal grain growth in nanograined metals and ceramics.[46,191,213,214] Remarkable results–consistent theoretical or near-theoretical densities and grain sizes less than 100 nm– have been achieved by hot pressing of mechanically alloyed powders (e.g., Fe–(2–10)% Al,[90,91] Al–10% Ti,[215] Fe, Fe3Al and Ni3Al,[216,217] Fe— Ni,[218] Ag—Ni,[219] and TiAl,[220]). A grain size of only 16 nm has been retained in cryomilled Fe–10 wt% Al, which was hot pressed at 823 K and heat treated for 1 hour at 1223 K.[91] As already shown, this unusual grain size stability is attributed to nanometer dispersoids of γ-Al2O3 and AlN particles. Similar densification results with final nanosize structures have been reported from hot pressing of composites (W—Ti and metalnitrides.[221,222]). Hot pressing retained an amorphous structure in (Fe, Co, Ni)B alloys.[223] Hot pressing of ceramics such as Al2O3,[224] ZrO2,[74] TiO2,[225,226] and CeO2[227] or composites such as ZrO2/Al2O3,[228] Si3N4/SiC,[229] and Al2O3/Ni[230] achieved full densities and grain sizes below 100 nm, as well. Generally, the pressures used in hot pressing span a large range from low (<100 MPa)[90,97,191] to moderate (100– 500 MPa),[74,91,226] and high pressure levels (>0.5 GPa).[46,73,213,216,231,232] Increasing the pressure diminishes the final grain size. Some meaningful

5: Nanocrystalline Powder Consolidation Methods, Groza 211 examples are given by Hahn, who sintered nano-TiO2 to its theoretical density at 725–825 K (∼0.35Tm), applying 1 GPa, with no grain growth,[73] and by Araki et al., who densified mechanically alloyed Al–10.7 at% Ti powders to 98% under 2 GPa at 573 K with virtually no grain growth and retention of the initial Al supersaturation.[231] A very high-pressure application (>1 GPa) retained grain sizes less than 75 nm in nearly fully dense ceramics such as TiO2 and Al2O3,[233] and aluminides (Fe3Al and Ni3Al) sintered to 91–95% at 775 K.[216]

5.3.4.2 Hot Isostatic Pressing (HIP) By design, HIP densification involves longer times and therefore grain growth is more likely to occur than in hot pressing. The final grain size in dense materials obtained by hot isostatic pressing of nanopowders may easily reach micrometer size.[234–236] However, a careful control of HIP parameters, particularly temperature, resulted in grain sizes of 100– 300 nm in ball-milled TiAl and Ti3Al,[237] Fe–10% Cu,[197] and Si3N4—SiC nanocomposites,[238] and sometimes less than 100 nm in TiAl[239] and Cu— NbC.[240] The latter dense nanocrystalline materials were obtained at temperatures less than 0.3Tm and pressures between 175 and 300 MPa. Grain sizes of 7–12 nm have been achieved by transformation-assisted refinement during hot isostatic pressing of amorphous Fe86Zr7B6Cu1 ribbons, but the density was not reported.[241] HIP was applied to nanoceramic densification in only a few cases where densities of 91–92% were reached such as in SiC at 2175 K[235] and Si3N4 at 1925 K.[234] Grain sizes well exceeded nanometer range. One limitation in using HIP for ceramic densification is the lack of adequate encapsulation materials with softening temperatures close to ceramic sintering temperatures.[2] HIP was used as a final sintering step to increase the density of pressureless sintered ceramics.[100] Hot isostatic pressing of composites results in a decrease of density with increasing amounts of the second phase, similar to conventional sintering of composites.[242]

5.3.4.3 Sinter Forging Generally, the stress levels required in sinter forging may be lower than in hot pressing or HIP. This method received substantial attention both theoretically and experimentally as applied to nanoceramic densification.[151,195,203,211,243–247] The most attractive benefit in using the sinterforging technique is the capability to densify green compacts with large

212

Processing

interagglomerate pores. As shown in section 5.3.3, the high shear stresses associated with uniaxial pressure application contribute to the closure of large pores that cannot otherwise be eliminated by diffusion alone. The plastic strains necessary to close large pores are high. Beyond pore closure, the application of high plastic strains induces dynamic grain growth. Therefore, to eliminate large pores and prevent grain coarsening, large strains must be applied before small pores are eliminated by diffusion, i.e. during the intermediate sintering stage.[151] For this purpose, a two-stage sinter-forging (high, then low strain rate) has been developed and resulted in the highest densities and smaller grain sizes.[130,203] Conversely, if large pores are not fractured, such as in some constant-load sinter-forging cases, densification is not complete.[151,193] Sinter forging has been extensively applied to particle consolidation in nanoceramics such as zirconia,[74,151,158,193,199,201,245] zirconia-toughened alumina,[245] alumina,[243,246] titania,[123,125,178,203] and SiC.[247] As an illustration of the efficiency of sinter forging in retaining nanometer grains, Mayo et al. sinterforged nano-TiO2 with a grain size of 87 nm, while pressureless sintering gave a 400 nm grain size for the same final density (91%).[123] Grain sizes even less than 50 nm have often been retained, such as in nanocrystalline ZrO2 at 0.4Tm[74] or TiO2 at 0.43Tm.[203] The benefits of sinter-forging of metals in comparison to HIP may be seen in Fe–10% Cu powders, in which the former consolidation process achieved 45 nm grains at 800 K and 525 MPa whereas HIP at 975 K and 170 MPa yielded a final grain size of 130 nm.[72,197] Practically, high strain rates may be used in sinter-forging of metals,[72] but precautions have to be considered to avoid ceramic specimen fracture at high strain rates at low temperatures.[151]

5.3.4.4 Extrusion Hot extrusion usually involves higher stresses that may be applied at relatively lower temperatures than in other pressure-assisted techniques, such as HIP. Good grain boundary strength is achieved only if sufficient diffusion and bonding occur at the extrusion temperature. A compromise is usually sought between the strength, which requires high temperatures, and the final grain size, for which low temperatures are desired. Hot extrusion has primarily been used to consolidate metal nanoparticles.[248–251] Grain sizes less than 100 nm were achieved at 1120 K using a 0.5 GPa stress in Ni and Fe,[248] and in Mg–5% Cu–10% Y alloys at 323–773 K with extrusion ratios of 5 : 1 to 10 : 1.[251] If the extrusion temperature is sufficiently low, an initial amorphous structure may be retained in the latter alloys.

5: Nanocrystalline Powder Consolidation Methods, Groza 213

5.3.4.5 Ultrahigh-Pressure Sintering Molecular dynamics simulations have been used to investigate the neck formation and densification upon high-pressure sintering.[252] Full densification of 5.6 nm Si3N4 was achieved at 2000 K under 15 GPa pressure within 100 ps, while sintering under 1 GPa pressure left numerous large pores. Practically, pressure levels up to 5–7 GPa have been attained using piston–anvil devices, cubic anvil cells, and toroidal- and belt-type apparatuses. Similar high pressure levels may be achieved by a new technique known as severe plastic deformation consolidation (SPDC). Shockwave or dynamic consolidation may reach even higher pressures, up to tens of gigapascals. This latter technique will be detailed in the next section. Some examples of high-pressure sintering have already been shown in the section on hot pressing. Misra et al. consolidated Al2O3 to near full densities (>99%) and a final grain size of 93 nm at 1223 K using a piston–cylinder method to reach 1 GPa.[253] Further reduction of grain sizes in ceramics has been achieved by application of pressure during phase transformations to maximize the nucleation rate.[254] The final grain size is inversely related to the pressure level. Grain sizes between 10 and 70 nm have been obtained in dense Al2O3 and TiO2 ceramics under pressures <8 GPa. A bulk nanostructure with 22 nm grain size was achieved in Al—Fe (Mo, Si, B) alloys by high-pressure consolidation (3 GPa) at 1073 K of alternate amorphous layers of constituent metals.[255] Fully dense Ni and Cu–50 wt% Ag specimens with ≤20 nm grain size were produced by SPDC.[256,257]

5.3.5 Nonconventional Sintering Methods A number of nonconventional consolidation methods have been applied to nanopowder densification: microwave sintering, shock or dynamic consolidation, magnetic and electrical field-assisted sintering. The main purpose in using these methods is to enhance densification, thus reducing the sintering temperature or time with the ultimate benefit of preserving final fine grain sizes.

5.3.5.1 Microwave Sintering The processing time in microwave sintering is reduced compared to conventional heating from external sources due to the direct energy

214

Processing

coupling with electric dipoles within the body being heated. Reviews of microwave processing summarize the physical principles and benefits of this process.[258,259] Temperature gradients are reduced and an overall short sintering time minimizes grain growth. In addition, the rapid heating rate can bypass the low-temperature region where the rate of grain growth is higher than the rate of densification. Since grain boundaries are the primary sites for electric dipoles, microwave sintering appears particularly attractive for the densification of nanocrystalline powders. The method has been applied mainly to ceramic nanoparticles such as TiO2, Al2O3, and PZT.[259–261] In some of these materials, only densities less than 95% have been achieved when grain size was less than 100 nm. For instance, microwave sintering had to be restricted to 1425 K to maintain a nanometer grain size in γ-Al2O3, which fully transformed to α-Al2O3.[260] The final density was 93%. Contrary to expectations, the density–grain size combinations in γ-Al2O3 were almost identical to those achieved in more conventional pressureless sintering experiments at the same temperature.[261] The sequence of phase transformation in Al2O3 was also similar to that in conventional sintering.

5.3.5.2 Field-Assisted Sintering Variants of field sintering that involve pulsed current discharge and resistance heating, sometimes termed field-activated sintering technique (FAST), spark plasma sintering, or pulse electro-discharge consolidation, have been effectively applied to fully densify conductive or nonconductive materials into bulk nanocrystalline structures.[262–264] The enhanced densification is most noticeable at lower temperatures or when multiple discharges of current through the powder compact are applied.[206,265] For nonconductive ceramics, a tentative explanation of the enhanced densification by multiple discharges is related to the increase in the dielectric constant with the temperature level at which the discharge is applied.[87] Generally, it is believed that pulsed current application promotes the removal of adsorbates or surface oxides due to possible electrical discharges.[264] This enhances the subsequent sinter bonding of “surface cleaned” particles. In the last decade, field-activated sintering seems to have been the most active processing technique for nanopowder densification. Literature is abundant and only a few of examples are given here. Dense BaTiO3 ceramics with 50 nm grain size were FAST-sintered at 800°C for 2 min under 100 MPa pressure, while conventional sintering of the same powders resulted in grain sizes >500 nm.[266] Kimura and

5: Nanocrystalline Powder Consolidation Methods, Groza 215 Kobayashi fully sintered mechanically alloyed TiAl powders and retained nanosize grains by spark sintering at 1051–1312 K under modest pressures (29–147 MPa).[267] Similarly to conventional sintering, they showed that the densification temperature scaled inversely with the applied pressure. Zhan et al. fully sintered Al2O3–carbon nanotube composites with 100 nm grain size by field sintering at 1200°C for 3 min.[268] An alternate electric pulsed power method, termed dynamic magnetic compaction (DMC), induces short high-pressure pulses that result in powder densification.[269,270] Ivanov et al. obtained up to 5 GPa pressure and thus consolidated alumina powders to 83% density while retaining a nanocrystalline structure.[269] The DMC method was applied to sintering of large parts made of nanocrystalline iron and alumina.[270]

5.3.5.3 Shockwave Consolidation Dynamic or shockwave consolidation proceeds with the passage of a large-amplitude compressive stress generated by plate impact or explosion without any external heating.[212,271] The peak pressure values may be on the order of tens of gigapascals, thus providing densification by plastic yielding for both metals and ceramics.[271–275] Localized heating, possibly up to melting temperatures, due to interparticle friction occurs and enables good interparticle bonding. In nanosize powders the heat may transfer throughout the entire particle, thus providing an advantage over coarser materials where the heating is only superficial. Best results are achieved when high temperatures are reached before the shockwave passes. If particles are heated, they may deform rather than fracture when the stress is applied. This very short high-temperature exposure provides the best means to retain fine grain size or out-of-equilibrium conditions such as amorphous structures[240] or supersaturated solid solutions.[275] The major drawback is the difficult coordination of these short stress and heat application events, which result in frequent specimen fracture (cracking). Shockwave consolidation has been applied to consolidation of both ceramic and metal nanoparticles. For instance, fully dense specimens with grain size of 20 nm were obtained in ball-milled Fe—N solid solution.[275] In the Ti—Si system, shock consolidation yielded 30–40 nm grain size of crystalline TiSi2 and Ti5Si3 phases.[273] Only limited grain coarsening took place upon subsequent annealing at 800°C for 1 hour. Full densification was also reported in mechanically alloyed TiAl specimens with final grain sizes of 15 nm.[272] These results are compared with HIP, which provided full consolidation at 1075 K/207 MPa/2 hours but grain sizes about

216

Processing

100 nm. Another comparison is with a pseudo-HIP process, Ceracon consolidation, at 1225 K/1240 MPa/1 min, which resulted in grain sizes of 60 nm in TiAl and 120 nm in Ti3Al.[212]

5.4 Summary The most distinctive features of the sintering process of nanosize powders are the high driving force and enhanced kinetics due to large curvature effects. Densification of nanopowders takes place at temperatures consistently below those of larger-grained powders by up to several hundreds of degrees. There are numerous benefits from using lower sintering temperatures: small final grain sizes, elimination of sintering aids, avoiding undesirable phase transformations, and deleterious decomposition or interfacial reactions. Analysis of the nanopowder sintering process provides new insights into features that are less conspicuous for regular sintering such as interface structure, anisotropy, or the role of contaminants and defects. The atomic mechanisms become more obvious when smaller scales are approached. When particle size decreases, new sintering mechanisms such as grain boundary slip, dislocation motion, grain rotation, and grain boundary melting may also become operative. Densification behavior is largely dictated by pore size and distribution. Densification of nanopowders has led to a new and better understanding of the role of pores in sintering and grain growth. A compact with uniform and fine pore structure sinters to full densities without uncontrolled grain coarsening. Most of the time, such a pore distribution is associated with a high green density in nonagglomerated or weakly agglomerated powders. The efforts to synthesize nonagglomerated nanopowders have reflected positively on their final densification behavior. Better control of synthesis and processing enables fabrication of fully dense nanocrystalline parts, particularly ceramic, by conventional sintering. Pressure-assisted densification applied to nanopowders has also shed light on new stress effects on sintering, such as the threshold effect. The science and practice of pressure application have significantly improved with positive results to achieve sound and dense nanocrystalline parts. This is particularly the case for sinter-forging. Other pressure-assisted sintering techniques are still in the development stage. Although some distinct differences in the densification of nanometervs. micrometer-grained powders have been shown and a better understanding of nanosintering has been accomplished, the specific effect of densification variables on final density and properties of nanomaterials is

5: Nanocrystalline Powder Consolidation Methods, Groza 217 not well understood yet. The interdependence of the sintering behavior on size, chemistry, and structure of nanoscale particles is an area of fruitful future research. More fundamentally, there is a tremendous need for specific materials data in the nanoregime, such as diffusion coefficients and surface energies.

Acknowledgments The author’s research on nanocrystalline materials densification has been supported by National Science Foundation grants. Heather Schembari is gratefully acknowledged for her help in the manuscript preparation.

References 1. Morgan, P.E.D., “Superplasticity in Ceramics,” in: Ultrafine-Grained Ceramics (J.J. Burke, L.R. Norman, and V. Weiss, eds.), pp. 251–271, Syracuse University Press, Syracuse (1968). 2. Mayo, M., “Processing of Nanocrystalline Ceramics from Ultrafine Particles,” Int. Mater. Rev. 41:85–115 (1996). 3. Bourell, D.L., and Groza, J.R., “Consolidation of Ultrafine and Nanocrystalline Powder,” in: Powder Metal Technologies and Fabrication, ASM Handbook, pp. 583–589, ASM International, Metals Park, OH (1998). 4. Andrievski, R.A. “Nanocrystalline High Melting Point Compound-Based Materials,” J. Mater. Sci. 29:614–631 (1994). 5. Groza, J.R., “Powder Consolidation,” in: Non-Equilibrium Processing of Materials (C. Suryanarayana, and R.W. Cahn, eds.), pp. 347–372, Elsevier Science, Oxford, UK (1999). 6. Bowen, P., and Carry, C., “From Powders to Sintered Pieces: Forming, Transformations and Sintering of Nanostructured Ceramic Oxides,” Powder Technology 128(2–3):248–255 (2002). 7. Vassen R., and Stover D., “Processing and Properties of Nanophase Ceramics,” J. Mater. Processing Technology 92–93:77–84 (1999). 8. Gutmanas, E.Y., “Materials with Fine Microstructures by Advanced Powder Metallurgy,” Prog. Mater. Sci. 34:261–366 (1990). 9. Suryanarayana, C., “Nanocrystalline Materials,” Int. Mater. Rev. 40:41–64 (1995). 10. Weismuller, J., “Nanocrystalline Materials–An Overview,” in: Synthesis and Processing of Nanocrystalline Powder (D.L. Bourell, ed.), pp. 3–19, TMS, Warrendale, PA (1996). 11. Lu, K., “Nanocrystalline Metals Crystallized from Amorphous Solids: Nanocrystallization, Structure and Properties,” Mater. Sci. Eng. R16:161–221 (1996). 12. Porter D.A., and Easterling, K.E., Phase Transformations in Metals and Alloys, 2nd ed., p. 113, Chapman and Hall, London (1992).

218

Processing

13. Johnson, D.L., “Solid State Sintering Models,” in: Sintering Processes (G.C. Kuczynski, ed.), pp. 97–106, MRS, Plenum Press, NY (1980). 14. Bonevics, J.E., and Marks, L.D., “The Sintering Behavior of Ultrafine Alumina Particles,” J. Mater Res.7:1489–1500 (1992). 15. Rankin, J., and Sheldon, B.W., “In Situ TEM Sintering of Nano-Sized ZrO2 Particles,” Mater. Sci. Eng. A204:48–53 (1995). 16. Carim, A.H., “Microstructure in Nanocrystalline Zirconia Powders and Sintered Compacts,” in: Nanophase Materials, Synthesis–Properties– Applications, (G.C. Hadjinapanis and R.W. Siegel, eds.), pp. 283–286, Kluwer Academic, Dordrecht (1994). 17. Guillou, N., Nistor, L.C., Fuess, H., and Hahn, H., “Microstructural Studies of Nanocrystalline CeO2 Produced by Gas Condensation,” Nanostruct. Mater. 8:545–557 (1997). 18. Olynick, D.L., Gibson, J.M., and Averbach, R.S., “Trace Oxygen Effects on Copper Nanoparticle Size and Morphology,” Appl. Phys. Lett. 68:343–345 (1996). 19. Champion, Y., and Bigot, J., “Synthesis and Structural Analysis of Al Nanocrystalline Powders,” Nanostruct. Mater. 10:1097–1110 (1998). 20. Zhu, H., and Averback, R.S., “Sintering Process of Two Nanoparticles: A Study by Molecular Dynamics Simulations,” Phil. Mag. Lett. 73:27–33 (1996). 21. Zeng, P., Zajac, S., Clapp P.C., and Rifkin, J.A., “Nanoparticle Sintering Simulations,” Mater. Sci. Eng., A252:301–6 (1998). 22. Ying, J. Y., Chi, L.F., Fuchs, H., and Gleiter, H., “Surface Morphology of Nanocrystalline Titanium Oxide by AFM,” Nanostruct. Mater. 3:273–281 (1993). 23. Moore, N.W., and Kuhl, T.L., “Controlled Self-Assembly,” in: CRC Materials Processing Handbook, CRC Press, Boca Raton, FL, to be published in 2007. 24. Luo, P., Nieh, T.G., Schwartz, A.J., and Lenk, T.J., “Surface Characterization of Nanostructured Metal and Ceramic Particles,” Mater. Sci. Eng. A204:59–64 (1995). 25. Phung, X., Groza, J., Stach, E.A., Williams, L.N., and Ritchey, S.B., “Surface Characterization of Metal Nanoparticles,” Mater. Sci. Eng. A359:261–8 (2003). 26. Cahn, R.W., “Nanostructures Come of Age,” Nature 359:591–592 (1992). 27. Trudeau, M.L., Tschope, A., and Ying, J.Y., “XPS Investigation of Surface Oxidation and Reduction in Nanocrystalline CexLa1−xO2−y,” Surf. Interf. Anal. 23:219–226 (1995). 28. Takeuchi, T., Ado, K., Asai, T., Kageyama, H., Saito, Y., Masquelier, C., and Nakamura, O., “Thickness of Cubic Surface Phase on Barium Titanate Single-Crystalline Grains,” J. Am. Ceram. Soc., 77:1665–1668 (1994). 29. Payne, M.C., “Electron Theory Applied to Phases with Low Dimensions,” in Mechanical Properties and Deformation Behavior of Materials Having Ultra-Fine Microstructures, (M. Nastasi, D. Markin, and H. Gleiter, eds.), pp. 37–52, Kluwer Academic, Dordrecht (1993). 30. Chatterjee, A., Kalia, R.K., Nakano, A., Omeltchenko, A., Tsuruta, K., Vashishta, P., Chun-Keung, L., Winterer, M., and Klein, S., “Sintering,

5: Nanocrystalline Powder Consolidation Methods, Groza 219

31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47.

Structure, and Mechanical Properties of Nanophase SiC: A MolecularDynamics and Neutron Scattering Study,” Appl. Phys. Lett. 77:1132–1134 (2000). McHale, J.M., Navrotsky, A., and Perrotta, A.J., “Effects of Increased Surface Area and Chemisorbed H2O on the Relative Stability of Nanocrystalline γ-Al2O3 and α-Al2O3,” J. Phys. Chem. B 101:603–613 (1997). McHale, J.M., Auroux, A., Perrotta, A.J., and Navrotsky, A., “Surface Energies and Thermodynamic Phase Stability in Nanocrystalline Aluminas,” Science, 277:788–791 (1997). Tsuruta, K., Totsuji, H., and Totsuji, C., “Tight-Binding Molecular Dynamics of Ceramic Nanocrystals Using PC-Based Parallel Machines,” Mater. Res. Soc. 581:673–678 (2000). Ferroni, L.P., and Cerrato, G., “Correlation between Surface Nanotopography and Sintering Behaviour of Zirconia Powders,” Nanotechnology 10:90–96 (1999). Rabe, T., and Wasche, R., “Sintering Behavior of Nanocrystalline Titanium Nitride Powders,” Nanostruct. Mater. 6:357–360 (1995). Turnbull, D., “Metastable Structures in Metallurgy,” Metall. Trans. 12A:695–708 (1981). Uchino K., Sadanaga E., and Hirose, T., “Dependence of the Crystal Structure on Particle Size in Barium Titanate,” J. Am. Ceram. Soc. 72:1555–1558 (1989). Skandan, G., Foster, C.M., Frase, H., Ali, M.N., Parker, J.C., and Hahn, H., “Phase Characterization and Stabilization Due to Grain Size Effects of Nanostructured Y2O3,” Nanostruct. Mater. 1:313–322 (1992). Garvie, R.C., “Stabilization of the Tetragonal Structure in Zirconia Microcrystals,” J. Phys. Chem. 82:218–223 (1978). Krauss, W., and Birringer, R., “Metastable Phases by Inert-Gas Condensation,” Nanostruct. Mater. 9:109–112 (1997). Ho-Soon, Y., Eastman, J.A., Thompson, L.J., Bai, G-R., “Grainsize-dependent thermal transport properties in nanocrystalline yttriastabilized zirconia,” Mater. Res. Soc. 703:179–84 (2002). Hague, D.C., and Mayo, M.J., “The Effect of Crystallization and a Phase Transformation on the Grain Growth of Nanocrystalline Titania,” Nanostruct. Mater. 3:61–67 (1993). Balachandran, U., Siegel, R.W., Liao, Y.X., and Askew, T.R., “Synthesis, Sintering and Magnetic Properties of Nanophase Cr2O3,” Nanostruct. Mater. 5:505–512 (1995). Hansen, J.D., Rusin, R.P., Teng, M.-H., and Johnson, D.L., “CombinedStage Sintering Model,” J. Am. Ceram. Soc. 75:1129–1135 (1992). Porat, R., Berger, S., and Rosen, R., “Dilatometric Study of the Sintering Mechanism of Nanocrystalline Cemented Carbides,” Nanostruct. Mater. 7:429–436 (1996). Hahn, H., Logas, J., and Averbach, R.S., “Sintering Characteristics of Nanocrystalline TiO2,” J. Mater. Res. 5:609–614 (1990). Kusunoki, M., Yonemitsu, K., Sasaki, Y., and Kubo, Y., “In Situ Observation of Zirconia Particles at 1200°C by High-Resolution Electron Microscopy,” J. Am. Ceram. Soc. 76:763–765 (1993).

220

Processing

48. Thuenissen, G.S.A.M., Winnubst, A.J.A., and Burggraaf, A.J., “Sintering Kinetics and Microstructure Development of Nanoscale Y-TZP Ceramics,” J. Europ. Ceram. Soc. 11:319–324 (1993). 49. German, R.M., Sintering Theory and Practice, pp. 86–88, Wiley, New York, (1996). 50. Hayashi, K., and Kihara, H., “Sintering Behavior of Ultrafine Metal Powders,” in: Fundamentals of Diffusion Bonding (I. Yshida, ed.), pp. 501–509, Elsevier, NewYork (1987). 51. Zhu, H., and Averback, R.S., “Sintering of Nano-Particle Powders: Simulations and Experiments,” Mater. Manuf. Processes 11:905–923 (1996). 52. Averback, R.S., Zhu, H., Tao, R., and Hofler, H., “Sintering of Nanocrystalline Materials: Experiments and Computer Simulations,” in: Synthesis and Processing of Nanocrystalline Powder (D.L. Bourell, ed.), pp. 203–216, TMS, Warrendale, PA (1996). 53. Yoo, Y.S., Kim, J.J., and Kim, D.Y., “Effect of Heating Rate on the Microstructural Evolution During Sintering of BaTiO3 Ceramics,” J. Am. Ceram. Soc. 70:C322–C324 (1987). 54. Chen, D.-J., and Mayo, M., “Densification and Grain Growth of Ultrafine 3 mol% Y2O3-ZrO2 Ceramics,” Nanostruct. Mater. 2:469–478 (1993). 55. Wang, J., Tsuruta, K., Omeltchenko, A., Kalia, R.K., and Vashishta, P., “Sintering of Amorphous Si3N4 Nanoclusters: a Molecular Dynamics Study of Stress Analysis,” Mater. Res. Soc. 408:573–576 (1996). 56. Panigrahi, B.B., Das, K., and Godkhindi, M.M., “Kinetic Analysis of Sintering of Nanocrystalline Powders,” Proc. Int. Conf. Trends in Mechanical Alloying: Science, Technology and Applications, Jaipur, India, Feb., 2001, pp. 49–60 (2002). 57. Glaeser, A.M., “Model Studies of Raleigh Instabilities via Microdesigned Interfaces,” Interface Science 9:65–82 (2001). 58. Valiev, R.Z., Private communication, UC Davis (1997). 59. Zhu H., and Averback, R.S., “Molecular Dynamics Simulations of Densification Processes in Nanocrystalline Materials,” Mater. Sci. Eng. A204: 96–100 (1995). 60. Tsuruta, K., Omeltchenko, A., Kalia, R.K., and Vashishta P., “Early Stages of Sintering of Silicon Nitride Nanoclusters: a Molecular-Dynamics Study on Parallel Machines,” Europhys. Lett. 33:441–446 (1996). 61. Yeadon, M., Yang, J.C., Averback, R.S., Bullard, J.W., Olynick, D.L., and Gibson, J.M., “In-situ Observations of Classical Grain Growth Mechanisms During Sintering of Copper Nanoparticles on (011) Copper,” Appl. Phys. Lett. 71 (12):1631–1633 (1997). 62. Yeadon, M., Ghaly, M., Yang, J.C., Averback, R.S., and Gibson, J.M., “Contact Epitaxy Observed in Supported Nanoparticles,” Appl. Phys. Lett. 73 (22):3208–3210 (1998). 63. Chen, P.-L., and Chen, I.-W., “Sintering of fine oxide powders: II, sintering mechanisms,” J. Am. Ceram. Soc. 80(3):637–645 (1997). 64. Ogata, S., Iyetomi, H., Tsuruta, K., Shimojo, F., Nakano, A., Kalia, R.K., and Vashishta, P., “Role of Atomic Charge Transfer on Sintering of TiO2 Nanoparticles. Variable-Charge Molecular Dynamics,” J. Appl. Phys. 88(10):6011–6015 (2000).

5: Nanocrystalline Powder Consolidation Methods, Groza 221 65. Yao, Y., and Tholen, A.R., “Adhesion between Nanoparticles,” Nanostruct. Mater., 12:661–664 (1999). 66. Buffat, P.A., “Electron Diffraction and HRTEM Studies of MultiplyTwinned Structures and Dynamical Events in Metal Nanoparticles: Facts and Artefacts,” Mater. Chem. Phys. 81(2–3):368–375 (2003). 67. Trusov, L.I., Lapovok, V.N., and Novikov, V.I., “Problems of Sintering Ultrafine Powders,” in: Science of Sintering, (D.P. Uskokovic, H. Plamour III, and R.M. Spriggs, eds.), pp. 185–192, Plenum Press, New York (1989). 68. Dominguez, O., Champion, Y., and Bigot, J., “Liquidlike Sintering Behavior of Nanometric Fe and Cu Powders: Experimental Approach,” Metall. Mater. Trans. 29A:2941–2949 (1998). 69. Pask, J.A., “Grain Boundaries in Sintering,” in: Science of Sintering, (D.P. Uskokovic, H. Plamour III, and R.M. Spriggs, eds.), pp. 381–393, Plenum Press, New York (1989). 70. Vergnon, P., Astier, M., and Teichner, S.J., “Initial Stage for Sintering of Ultrafine Particles (TiO2 and Al2O3),” in: Fine Particles (W.E. Kuhn, ed.), pp. 299–307, The Electrochemical Society, Princeton, NJ (1974). 71. Messing, G.L., and Kumagai, M., “Low-Temperature Sintering of αAlumina Seeded Bohemite Gels,” Am. Ceram. Bull. 73:88–91 (1994). 72. Shaik, G.R., and Milligan, W.W., “Consolidation of Nanostructured Metal Powders by Rapid Forging: Processing, Modeling, and Subsequent Mechanical Behavior,” Metall. Mater. Trans. 28A:895–904 (1997). 73. Hahn, H., “Microstructure and Properties of Nanostructured Oxides,” Nanostruct. Mater. 2:251–265 (1993). 74. Skandan, G., “Processing of Nanostructured Zirconia Ceramics,” Nanostruct. Mater. 5:111–126 (1995). 75. Wu, S.J., and De Jonghe, L.C., “Sintering of Nanophase γ-Al2O3 Powder,” J. Am. Ceram. Soc. 79:2207–2211 (1996). 76. Alymov, M.I., Maltina, E.I., and Stepanov, Y.N., “Model of Initial Stage of Ultrafine Metal Powder Sintering,” Nanostruct. Mater. 4:737–742 (1992). 77. Yan, M.F., and Rhodes, W.W., “Low Temperature Sintering of TiO2,” Mater. Sci. Eng. 61:59–66 (1983). 78. Pechenik, A., Piermarini, G.J., and Danforth, S.C., “Fabrication of Transparent Silicon Nitrides from Nanosize Particles,” J. Am. Ceram. Soc. 75:3283–3288 (1992). 79. Pechenik, A., Piermarini, G.J., and Danforth, S.C., “Low Temperature Densification of Silicon Nitride Nanoglasses,” Nanostruct. Mater. 2:479–486 (1993). 80. Kumar, K.-N.P., Keizer, K, Burggraaf, A.J., Okubo, T., Nagamoto, H., and Morooka, S., “Densification of Nanostructured Titania Assisted by a Phase Transformation,” Nature 358:48–51(1992). 81. Alymov, M.I., Maltina, E.I., and Stepanov, Y.N., “Correlation Between the Melting Point and the Initial Sintering Temperature of Ultrafine Metallic Powders,” Fizika i Khimiya Obrabotki Materialov 4–5:131–134 (1994). 82. Herring, C., “Effect of Change of Scale on Sintering Phenomena,” J. Appl. Phys. 21:301–303 (1950). 83. Andrievski, R.A., “Compaction and Sintering of Ultrafine Powders,” Int. Powder Metall. 30:59–66 (1994).

222

Processing

84. Song, H., Coble, R.L., and Brook, R.J., “The Applicability of Herring’s Scaling Law to the Sintering of Powders,” Mater. Sci. Res, 6(Sintering Heterog. Catal.):63–79 (1984). 85. Zeng, P., Clapp, P.C., and Rifkin, J.A., “Failure of Herring’s Sintering Law at the Nanoscale,” Computer-Aided Design of High-Temperature Materials, Conf. Santa Fe, N. Mex., July 30–Aug. 2, 1997, pp. 483–490 (1999). 86. Sanders, P.G., Fougere, G.E., Thompson, L.J., Eastman, J.A., and Weertman, J.R., “Improvements in the Synthesis and Compaction of Nanocrystalline Materials,” Nanostruct. Mater. 8:243–252 (1997). 87. Schneider, J.A., Processing and Properties of Silicon Nitrides Ceramics, PhD Thesis, UC Davis (1997). 88. Ying, J.Y., “Sintering of Nanocrystalline Al2O3 Study by Photoacoustic Infrared Spectroscopy,” in: Mechanical Properties and Deformation Behavior of Materials Having Ultra-Fine Microstructures (M. Nastasi, D.M. Parkin, and H. Gleiter, eds.), pp. 565–570, Kluwer Academic, Dordrecht (1993). 89. Koch, C.C., “Processing of Nanophase Materials by High Energy Milling,” in: Nanophases and Nanocrystalline Structures (R.D. Shull, and J.M. Sanchez, eds.), pp. 19–31, TMS, Warrendale PA (1994). 90. Rawers, J., Slavens, G., Govier, D., Dogan, C., and Doan, R., “Microstructure and Tensile Properties of Compacted, Mechanically Alloyed Nanocrystalline Fe—Al,” Metall. Mater. Trans. 27A:3126–3134 (1996). 91. Perez, R.J., Huang, B., and Lavernia, E.J., “Thermal Stability of Nanocrystalline Fe-10 wt.%Al Produced by Cryogenic Mechanical Alloying,” Nanostruct. Mater. 7:565–572 (1996). 92. Olynick, D.L., Gibson, J.M., and Averbach, R.S., “In-Situ Ultra-High Vacuum Transmission Electron Microscopy Studies of Nanocrystalline Copper,” Mater. Sci. Eng. A204:54–58 (1995). 93. Eldrup, M., Bentzon, M.D., Pedersen, A.S., Linderoth, S., Pedersen, N.J., and Larsen, B., “Reduction of Surface Oxide on Ultrafine FeNi Particles,” in: Mechanical Properties and Deformation Behavior of Materials Having Ultra-Fine Microstructures (M. Nastasi, D.M. Parkin, and H. Gleiter, eds.), pp. 571–577, Kluwer Academic, Dordrecht (1993). 94. Champion, Y., Bernard, F., Guigue-Millot, N., and Perriat, P., “Sintering of Copper Nanopowders Under Hydrogen: an In Situ X-Ray Diffraction Analysis,” Mater. Sci. Eng. A360:258–263 (2003). 95. Chen, W., Pechenik, A., Dapkunas, S.J., Piermarini, G.J., and Malghan, S.G., “Novel Equipment for the Study of Compaction of Fine Powders,” J. Am. Ceram. Soc. 77:1005–1010 (1994). 96. Chen, W., and Malghan, S.G., “Investigations of Compaction Equations for Powders,” Powder Technology 81:75–81 (1994). 97. Bourell, D.L., Parimal, and Kaysser, W.A., “Sol-Gel Synthesis of Nanophase Yttria-Stabilized Tetragonal Zirconia and Densification Behavior Below 1600 K,” J. Am. Ceram. Soc. 76:705–711 (1993). 98. Majetich, S.A., Scott, J.H., Brunsman, E.M., McHenry, M.E., and Nuhfer, N.T., “Structure and Magnetism of Carbon-Coated Rare Earth Carbide Nanocrystals,” in: Fullerenes: Physics, Chemistry, and New Directions VI, (R.S. Ruoff, and K. Kandish, eds.), pp. 1448–1462, The Electrochem. Society, Pennington, NJ (1994).

5: Nanocrystalline Powder Consolidation Methods, Groza 223 99. Lange, F.F., “Powder Processing Science and Technology for Increased Reliability,” J. Am. Ceram. Soc. 72:3–15 (1989). 100. Wolff, M., Falk, G., and Clasen, R., “Sintering Behavior of Dense Nanocrystalline Zirconia Ceramics: A Comparative Investigation,” Ceram. Eng. Sci. Proc. 25(4):19–24 (2004). 101. Rhodes, W.H., “Agglomerate and Particle Size Effects on Sintering YttriaStabilized Zirconia,” J. Am. Ceram. Soc. 64:19–22 (1981). 102. Miller, K.T., and Zukovski, C.F., “Osmotic Consolidation of Suspensions and Gels,” J. Am. Ceram. Soc. 77:2473–2478 (1994). 103. Dominguez, O., Phillippot, M., and Bigot, “The Relationship Between Consolidation Behavior and Particle Size in Fe Nanometric Powders,” Scr. Metall. Mater. 32:13–17 (1995). 104. Gutmanas, E.Y., Trusov, L.I., and Gotman, I., “Consolidation, Microstructure and Mechanical Properties of Nanocrystalline Metal Powders,” Nanostruct. Mater. 8:893–901 (1994). 105. Nieman, W., Weertman, J.R., and Siegel, R.W., “Mechanical Behavior of Nanocrystalline Cu and Pd,” J. Mater. Res. 6:1012–1027 (1991). 106. Nieman, G.W., Weertman, J.R., and Siegel, R.W., “Mechanical Behavior of Nanocrystalline Metals,” Nanostruct. Mater. 1:185–190 (1992). 107. Kobelev, N.P., Soifer, Y.M., Andrievski, R.A., and Gunther, B., “Microhardness and Elastic Properties of Nanocrystalline Silver,” Nanostruct. Mater. 2:537–544 (1993). 108. Kumpmann, A., Gunther, B., and Kunze, H.-D., “Thermal Stability of Ultrafine-Grained Metals and Alloys,” Mater. Sci. Eng. A168:165–69 (1993). 109. Bonetti, E., Valdre, G., Enzo, S., Cocco, G., and Soletta, I., “Nanostructured Fe3Al Intermetallic Obtained by Mechanical Alloying and Thermal Ageing,” Nanostruct. Mater. 2:369–375 (1993). 110. Devaux, X., Brochin, F., Dauscher, A., Lenoir, B., Martin-Lopez, R., Scherrer, H., and Scherrer, S., “Production of Ultrafine Powders of Bi—Sb Solid Solution,” Nanostruct. Mater. 8:137–147 (1997). 111. Sanders, P.G., Witney, A.B., Weertman, J.R., Valiev, R.Z., and Siegel, R.W., “Residual Stress, Strain and Faults in Nanocrystalline Palladium and Copper,” Mater. Sci. Eng. A204:7–11 (1995). 112. Fougere, G.E., Weertman, J.R., and Siegel, R.W., “Processing and Mechanical Behavior of Nanocrystalline Fe,” Nanostruct. Mater. 5:127–134 (1995). 113. Agnew, S.R., Elliott, B.R., Youngdahl, C.J., Hemker, K.J., and Weertman, J.R., “Microstructure and Mechanical Behavior of Nanocrystalline Metals,” Mater. Sci. Eng. A285:391–396 (2000). 114. Andrievski, R.A., Torbov, V.I., and Kurkin, E.N., “Nanocrystalline Titanium Nitride,” Proceedings of the 13th International Plansee Seminar (H. Bildstein, and R. Eck, eds.), vol. 3, pp. 649–655, Metallwerk Plansee, Reutte (1993). 115. Rhee, C.K., Lee, G.H., Kim, W.W., Ivanov, V.V., Zajats, S.V., and Medvedev, A.I., “Nanostructured Al/Al2O3 Composite Sintered by Magnetic Pulse Compaction,” J. Metastable Nanocryst. Mater. 15–16:401–406 (2003). 116. Ivanov, V., Paranin, S., and Nozdrin, A., “Principles of Pulsed Compaction of Ceramic Nano-sized Powders,” Key Eng. Mater. 132–136:400–403 (1997).

224

Processing

117. Gallas, M.R., Rosa, A.R., Costa, T.H., and da Jornada, J.A.H., “High Pressure Compaction of Nanosize Ceramic Powder,” J. Mater. Res. 12:764–768 (1997). 118. Sagel-Ransijn, C.D., Winnbust, A.J.A., Kerkwijk, B., Burggraaf, A.J., and Verweij, H., “Production of Defect-Poor Nanostructured Ceramics of YttriaZirconia,” J. Europ. Ceram. Soc. 17:831–841 (1997). 119. Van de Graaf, M.A.C.G., Termaat, J.H.H., and Burggraaf, A.J., “Microstructure and Sintering Kinetics of Highly Reactive ZrO2-Y2O3 Ceramics,” J. Mater. Sci. 20:1407–1418 (1985). 120. Li, J.G., and Sun, X., “Synthesis and Sintering Behavior of a Nanocrystalline α-Alumina Powder,” Acta Mater. 48:3103–3112 (2003). 121. Kingery, W.D., Bowen, H.K., and Uhlmann, D.R., Introduction to Ceramics, 2nd ed., p. 486, Wiley, New York (1976). 122. Brook, R.J., “Pore–Grain Boundary Interactions and Grain Growth,” J. Am. Ceram. Soc. 52:56–57 (1969). 123. Mayo, M., Hague, D.C., and Chen, D.-J., “Processing Nanocrystalline Ceramics for Applications in Superplasticity,” Mater. Sci. Eng. A166: 145–159 (1993). 124. Mayo, M.J., and Hague, D.C., “Porosity–Grain Growth Relationships in the Sintering of Nanocrystalline Ceramics,” Nanostr. Mater. 3:43–52 (1993). 125. Averback, R.S., Hofler, H.J., and Tao, R., “Processing of Nano-Grained Materials,” Mater. Sci. Eng. A166:169–177 (1993). 126. Hahn, H., Averback, R.S., Hofler, H.J., and Logas, J., “Sintering and Deformation of Nanocrystalline Ceramics,” in: Clusters and Cluster-Assembled Materials (R.S. Averbach, J. Bernholc, and D.L. Nelson, eds.), pp. 569–580, MRS, Pittsburgh, PA (1991). 127. Hackley, V.A., Anderson, M.A., and Spooner, S.A., “Small-Angle X-ray Scattering Study of Microstructure Evolution During Sintering of Sol-GelDerived Porous Nanophase Titania,” J. Mater. Res. 7:2555–2571(1992). 128. Wen, T.L., Hebert, V., Vilminot, S., and Bernier, J.C., “Preparation of Nanosized Yttria-Stabilized Zirconia Powders and Their Characterization,” J. Mater. Sci. 26:3787–3791 (1991). 129. McCandlish, L.E., Kevorkian, V., Jia, K., and Fisher, T.E., “Nanostructured WC-Co Composite Powders,” in Advances in Powder Metallurgy and Particulate Materials (C. Lall, and A.J. Neupaver, eds.), pp. 328–337, MPIF, Princeton, NJ (1994). 130. Skandan, G., Hahn, H., Kear, B.H., Roddy, M., and Cannon, W.R., “Processing of Nanostructured Zirconia Ceramics,” in: Molecularly Designed Ultrafine/Nanostructure Materials (K.E. Gonsalves, G.-M. Chow, T.D. Xiao, and R.C. Cammarata, eds.), pp. 207–212, MRS, Pittsburgh, PA (1994). 131. Ahn, J.-P., Huh, M.-Y., and Park, J.-K., “Effect of Green Density on Subsequent Densification and Grain Growth of Nanophase SnO2 Powder During Isothermal Sintering,” Nanostruct. Mater. 8:637–643 (1997). 132. Zhao, J., and Harmer, M.P., “Effect of Pore Distribution on Microstructure Development. III. Model Experiments,” J. Am. Ceram. Soc. 75:830–843 (1992). 133. Kang, S.-J.L., and Jung, Y.-I., “Sintering Kinetics at Final Stage Sintering: Model Calculation and Map Construction,” Acta Mater. 52:4573–4578 (2004).

5: Nanocrystalline Powder Consolidation Methods, Groza 225 134. Iyer, S.R., and Sastry, S.M.L., “Consolidation of Nanoparticles–Development of a Micromechanistic Model,” Acta Mater. 47:3079–3098 (1999). 135. Sweeney, S.M., and Mayo, M.J., “Green Density-Based Sintering Predictions,” J. Mater. Res. 17:89–97 (2002). 136. Kuczynski, G.C., “A Note About So-Called ‘Activated’ Sintering in Sintering–New Developments,” (M. Ristic, ed.), pp. 245–256, Elsevier (1979). 137. Evans, A.G., and Hsueh, C.H., “Behavior of large Pores During Sintering and Hot Isostatic Pressing,” J. Am. Ceram. Soc. 69:444–448 (1986). 138. Raj, R., “Analysis of the Sintering Pressure,” J. Am. Ceram Soc. 70: C210–C211 (1987). 139. Malow, T.R., and Koch, C.C., “Thermal Stability of Nanocrystalline Materials,” Mater. Sci. Forum 225–227:595–604 (1996). 140. Malow, T.R., and Koch, C.C., “Grain Growth of Nanocrystalline Materials–A Review,” in: Synthesis and Processing of Nanocrystalline Powder (D.L. Bourell, ed.), pp. 33–44, TMS, Warrendale, PA (1996). 141. Malow, T.R., and Koch, C.C., “Grain Growth in Nanocrystalline Iron Prepared by Mechanical Attrition,” Acta Mater. 45:2177–2186 (1997). 142. Zhao, J., and Harmer, M.P., “Sintering Kinetics for a Model Final-Stage Microstructure: A Study of Al2O3,” Phil. Mag. Lett. 63:7–14 (1991). 143. Eastman, J.A., “Microstructural Development in Nanophase TiO2 During Sintering,” J. Appl. Phys. 75:770–779 (1994). 144. Eckert, J., Holzer, J.C., Krill, III, C.E., and Johnson, W.L., “Structural and Thermodynamic Properties of Nanocrystalline fcc Metals Prepared by Mechanical Attrition,” J. Mater. Res. 7:1751–1761 (1992). 145. Rahaman, M.N., Ceramic Processing and Sintering, pp. 445–508, Marcel Dekker, NY (1995). 146. Kanters, J., Eisele, U., and Roedel, J., “Effect of Initial Grain Size on Sintering Trajectories,” Acta Mater. 48:1239–1246 (2000). 147. Sivakumar, S., Subbanna, M., Sahay, S.S., Ramakrishnan, V., Kapur, P.C., Pradip, and Malghan, S.G., “Population Balance Model for Solid State sintering, II. Grain Growth,” Ceram. Int. 27:63–71 (2001). 148. Lange, F., “Sinterability of Agglomerated Powders,” J. Am. Ceram. Soc. 67:83–89 (1984). 149. Mortensen, A., “Kinetics of Densification by Solution-Reprecipitation,” Acta Met. Mater. 45:749–758 (1997). 150. Nieh, T.G., and Wadsworth, J., “Dynamic Grain Growth During Superplastic Deformation of Yttria-Stabilized Tetragonal Zirconia Polycrystals,” J. Am. Ceram. Soc. 72:1469–1472 (1989). 151. Hague, D.C., and Mayo, M.J., “Sinter-Forging of Nanocrystalline Zirconia: 1, Experimental,” J. Am. Ceram. Soc. 80:149–156 (1997). 152. Gupta, K., “Possible Correlation Between Density and Grain Size During Sintering,” J. Am. Ceram. Soc. 55:276–277 (1972). 153. Liu Y., and Patterson, B.R., “Grain Growth Inhibition by Porosity,” Acta Met. Mater. 41:2651–2656 (1993). 154. Chen, I.-W., and Wang, X.-W., “Sintering Dense Nanocrystalline Ceramics without Final-Stage Grain Growth,” Nature, 404:168–171 (2000). 155. Weissmuller, J., Krauss, W., Haubold, T., Birringer, R., and Gleiter, H., “Atomic Structure and Thermal Stability of Nanostructured Y—Fe Alloys,” Nanostruct. Mater. 1:439–447 (1992).

226

Processing

156. Weissmuller, J., “Alloy Effects in Nanostructures,” Nanostruct. Mater. 3:261–272 (1993). 157. Terwilliger, C.D., and Chiang, Y.-M., “The Effect of Calcium Segregation on Grain Growth in Nanocrystalline TiO2,” Nanostruct. Mater. 4:651–661 (1994). 158. Owen, D.M., and Chokshi, A.H., “An Evaluation of the Densification Characteristics of Nanocrystalline Materials,” Nanostruct. Mater. 2:181–187 (1993). 159. Rahaman, M.N., “Sintering and Grain Growth of Ultrafine CeO2 Powders,” in: Sintering Technology (R.M. German, G.L. Messing, and R. Cornwall, eds.), pp. 93–100, Marcel Dekker, NewYork (1996). 160. Gao, Z.Q., and Fultz, B., “Inter-dependence of Grain Growth, Nb Segregation and Chemical Ordering in Fe—Si—Nb Nanocrystals,” Nanostruct. Mater. 4:939–947 (1994). 161. Bansal, C., Gao, Z.Q., and Fultz, B., “Grain Growth and Chemical Ordering in (Fe, Mn)3Si,” Nanostruct. Mater. 5:327–336 (1995). 162. Misra, R.S., Valiev, R.Z., and Mukherjee, A.K., “Fully Dense Nanocrystalline Nickel by Severe Plastic Deformation Consolidation,” Mater. Sci. Forum 225–227:605–610 (1996). 163. Gleiter, H., “Nanostructured Materials: State of the Art and Perspectives,” Zeitschr. Metallk. 86:78–83 (1995). 164. Eastman, J.A., and Fitzsimmons, M.R., “On the Two-State Microstructure of Nanocrystalline Chromium,” J. Appl. Phys. 77:522–527 (1995). 165. Groza, J.R., and Dowding, R., “Nanoparticulate Materials Densification,” Nanostruct. Mater. 8:749–768 (1997). 166. ASTM B 328–92 Standard Test Method for Density, Oil Content, and Interconnected Porosity of Sintered Structural Parts and Oil-Impregnated Bearings. 167. Schaefer, H.E., Wurschum, R., Scheytt, M., Birringer, R., and Gleiter, H., “Structure of Nanometer-Sized Polycrystalline Iron Investigated by Positron Lifetime Spectroscopy,” Mater. Sci. Forum 15–18:9545–9554 (1987). 168. Allen, A.J., Krueger, S., Long, G.G., Kerch, H.M., Hahn, H., and Skandan, G., “Small-Angle Neutron Scattering Studies of Ceramic Nanophase Materials,” Nanostruct. Mater. 7:113–126 (1996). 169. Ying, J.Y., Wang, G.-H., Fuchs, H., Laschinski, R., and Gleiter, H., “STM AFM Study of Grain Boundary Migration in Nanostructured Solids,” Mater. Lett. 15:180–185 (1992). 170. Anselmi Tamburini, U., Munir, Z.A., Kodera, Y., Imai, T., and Ohyanagi, M., “Influence of Synthesis Temperature on the Defect Structure of Boron Carbide,” J. Am. Ceram. Soc 88:1382–1387 (2005). 171. Siegel, R.W., Ramasamy, S., Hahn, H., Li, Z., Lu, T., and Gronsky, R., “Synthesis, Characterization, and Properties of Nanophase TiO2,” J. Mater. Res. 3:1367–1372 (1988). 172. Winnubst, A.J.A., Keizer, K., and Burgraaf, A.J., “Mechanical Properties and Fracture Behavior of ZrO2-Y2O3 Ceramics,” J. Mater. Sci. 18:1958– 1966 (1983). 173. Handwerker, C.A., Blendell, J.E., and Coble, R.L, “Sintering of Ceramics,” in: Science of Sintering (D.P. Uskokovics, H. Palmour III, and R.M. Spriggs, eds.), pp. 3–37, Plenum Press, New York (1989).

5: Nanocrystalline Powder Consolidation Methods, Groza 227 174. Boutz, M.M.R., Olde Scholtenhuis, R.J.M., Winnubst, A.J.A., and Burggraaf, A.J., “A Hydrothermal Route for Production of Dense, Nanostructured Y-TZP,” Mater. Res. Bull. 29:31–40 (1994). 175. Murray, M.G.S., Ponton, C.B., and Marquis, P.M., “Processing of Ultrafine Grained Hydroxyapatite Bioceramics,” in: Nanoceramics (R. Freer, ed.), pp. 25–30, Inst. of Materials, London (1993). 176. Duran, P., Villegas, M., Capel, F., Fernandez, J.F., and Moure, C., “Nanostructured and Near Defect-free Ceramics by Low-temperature Pressureless Sintering of Nanosized Y-TZP Powders,” J. Mater. Sci. 32:4507–4512 (1997). 177. Maca, K., and Trunec, M., “Sintering Kinetics of Ziconia Nanoparticle Compacts,” in: Proc. 4th International Conference on Science, Technology and Applications of Sintering (D. Bouvard, ed.), pp. 29–32, Grenoble, France (2005). 178. Hague, D.C., and Mayo, M.J., “The Effect of Crystallization and a Phase Transformation on the Grain Growth of Nanocrystalline Titania,” Nanostruct. Mater. 3:61–67 (1993). 179. Ragulya, A.V., and Skorokhod, V.V., “Genesis of Nanosized Particles, Grains and Interfaces in the Rate-Controlled Processes of Synthesis and Sintering of Ceramics,” in: Nanostructures: Synthesis, Functional Properties and Applications (NATO Science Series, II: Mathematics, Physics and Chemistry, 128), pp. 139–154 (2003). 180. Johnson, D.L., “Rapid Sintering of Ceramics,” in: Advanced Ceramics II (S. Somiya, ed.), pp. 1–6, Elsevier Applied Science, London (1986). 181. Chen D.-J., and Mayo, M.J., “Rapid Rate Sintering of Nanocrystalline ZrO2–3 mol% Y2O3,” J. Am. Ceram. Soc. 79:906–912 (1996). 182. Venkatachari, K.R., Huang, D., Ostrander, S.P., Schulze, W.A., and Stangle, G.C., “Preparation of Nanocrystalline Yttria-Stabilized Zirconia,” J. Mater. Res. 10:756–761 (1995). 183. Ragulya, A.V., and Skorokhod, V.V., “Rate-Controlled Sintering of Ultrafine Nickel Powder,” Nanostruct. Mater. 5:835–843 (1995). 184. Sadangi, R.K., Kear, B.H., and McCandlish, L.E., “Pressureless Sintering of Nanograined Fe-(Fe, Mo)6 Composite Powders,” Powder Metall. 37:277–282 (1994). 185. Wu, L., Kim, B.K., Kear, B.H., and McCandlish, L.E., “Grain Growth Inhibition in Sintering of Nanostructured WC-Co Alloys,” in: Proc. 13th Int. Plansee Seminar (H. Bildstein, and R. Eck, eds.), vol. 3, pp. 667–679, Metallwerk Plansee, Reute (1993). 186. Fang, Z., and Eason, J.W., “Study of Nanostructured WC-Co Composites,” Int. J. Refract. Hard Met. 13:297–303 (1995). 187. Candlish, L.E., Kear, B.H., and Kim, B.K., “Processing and Properties of Nanostructured WC-Co,” Nanostruct. Mater. 1:119–124 (1992). 188. Chang, H., Hofler, H.J., Alstetter, C.J., and Averbach, R.S., “Synthesis, Processing and Properties of Nanophase TiAl,” Scr. Metall. Mater. 25:1161–1166 (1991). 189. Chang, H., Hofler, J., Altstetter, C., and Averback, R.S., “Synthesis, Processing and Properties of Nanophase Aluminide,” Mater. Sci. Eng. A153:676–678 (1992).

228

Processing

190. Moon, K.I., Hong, H.S., Hong, K.T., Lee, K.S., and Kim, S.J., “Consolidation Behavior of L12 Phase (Al + 12.5 at.% Cu)3Zr Powder with Nanocrystalline Structure during CIP and Subsequent Sintering,” Mater. Sci. Eng. A380(1–2):46–51 (2004). 191. Prabhu, G.B., and Bourell, D.L., “Synthesis and Sintering Characteristics of Zirconia and Zirconia-Alumina Nanocomposites,” Nanostruct. Mater. 6:361–364 (1995). 192. Nakahira, A., and Niihara, K., “Sintering Behavior and Consolidation Process for Al2O3/SiC Nanocomposites,” J. Ceram. Soc. Japan, 100:448–453 (1992). 193. Betz, U., Sturm, A., Loffler, J.F., Wagner, W., Wiedenmann, A., and Hahn H., “Microstructural Development During Final-Stage Sintering of Nanostructured Zirconia Based Ceramics,” Mater. Sci. Eng. A281:68–74 (2000). 194. Holtz, R.L., and Provenzano, V., “Enhanced Microhardness of CopperNiobium Nanocomposites,” Nanostruct. Mater. 4:241–256 (1994). 195. Averback, R.S., Hofler, H.J., Hahn H., and Logas, J.C., “Sintering and Grain Growth in Nanocrystalline Ceramics,” Nanostruct. Mater. 1:173–178 (1992). 196. He, L., and Ma, E., “Processing and Microhardness of Bulk Cu—Fe Nanocomposites,” Nanostruct. Mater. 7:327–339 (1996). 197. Carsley, J.E., Milligan, W.W., Hackney, S.A., and Aifantis, E.C., “Glasslike Behavior in a Nanostructured Fe/Cu Alloy,” Metall. Mater. Trans. 26A:2479–2481 (1995). 198. Skandan, G., Hahn, H., Roddy, M., and Cannon, R.W., “Ultrafine-Grained Dense Monoclinic and Tetragonal Zirconia,” J. Am. Ceram. Soc. 77:1706–1710 (1994). 199. Boutz, M.M.R., Winnubst, L., Burggraaf, A.J., Nauer, M., and Carry, C., “Low-Temperature Sinter Forging of Nanostructured Y-TZP and Ce-TZP,” J. Am. Ceram Soc. 78:121–128 (1995). 200. Groza, J.R., “Non-Conventional Pressure-Assisted Powder Consolidation Processes,” J. Mater. Eng. Perform. 2:283–290 (1993). Groza, J.R., “Powder Consolidation,” in Non-equilibrium Processing of Materials (C. Suryanarayana, ed.), pp. 345–372, Pergamon, Elsevier Science (1999). 201. Skandan, G., Hahn, H., Kear B.H., Roddy, M., and Cannon, R.W., “The Effect of Applied Stress on Densification of Nanostructured Zirconia during Sinter-Forging,” Mater. Lett., 20:305–309 (1994). 202. Hofler, H.J., and Averbach, R.S., “Sinter-Forging of Nanocrystalline TiO2,” in: Nanophase and Nanocomposite Materials, MRS vol. 286, (S. Komarneni, J.C. Parker, and G.J. Thomas, eds.), pp. 9–14, MRS, Pittsburgh, PA (1993). 203. Uchic, M., Hofler, H.J., Flick, W.J., Tao, R., Kurath, P., and Averbach, R.S., “Sinter-Forging of Nanophase TiO2,” Scr. Metall. Mater. 26:791–796 (1992). 204. Rahaman, N.R., and De Jonghe, L.C., “Sintering of CdO Under Low Applied Stress,” J. Am. Ceram. Soc. 67:C205–C207 (1984). 205. Schwarz, R.B., Desch, P.B., Srinivasan, S., and Nash, P., “Synthesis and Properties of Trialuminides with Ultra-Fine Microstructures,” Nanostruct. Mater. 1:37–42 (1992).

5: Nanocrystalline Powder Consolidation Methods, Groza 229 206. Yoo, S.H., Field Effects in Plasma Activated Sintering, MS Thesis, UC Davis (1996). 207. Arzt, E., Asby, M.F., and Easterling, K.E., “Practical Applications of HotIsostatic Pressing Diagrams: Four Case Studies,” Metall. Trans. 14A: 211–221 (1983). 208. Swinkels, F.B., Wilkinson, D.S., Arzt, E., and Ashby, M.F., “Mechanisms of Hot-Isostatic Pressing,” Acta Metall. 31:1829–1840 (1983). 209. McKimpson, M.G., “Densification Maps for Nanosized Powders,” Mater. Manuf. Proc. 11:935–949 (1996). 210. Duva, J.M., and Crow, P.D., “The Densification of Powders by Power-Law Creep During Hot Isostatic Pressing,” Acta Metall. Mater. 40:31–35 (1992). 211. Hague, D.C., and Mayo, M.J., “Modeling Densification During SinterForging of Yttria-Partially-Stabilized Zirconia,” Mater. Sci. Eng. A204: 83–89 (1995). 212. Korth, G.E., and Williamson, R.L., “Dynamic Consolidation of Metastable Nanocrystalline Powders,” Metall. Mater. Trans. 26a:2571–2578 (1995). 213. Kear, B.H., and Mayo, W.E., “On the Processing of Nanostructured Ceramics,” Mater. Sci. Forum 437–438:399–402 (2003). 214. Dutkiewicz, J., and Maziarz, W., “Nanocrystalline TiAl-V Intermetallics Hot Pressed From Ball Milled Powders,” Diffusion and Defect Data-Solid State Data, Pt. B: Solid State Phenomena 101–102:117–122 (2005). 215. Guoxian, L., Qingchang, M., Zhichao, Li, and Erde, W., “Consolidation of Nanocrystalline Al—Ti Alloy Powders Synthesized by Mechanical Alloying,” Nanostruct. Mater. 5:673–678 (1995). 216. He, L., and Ma, E., “Nanophase Metallic Alloys Consolidated from Powders Prepared by Mechanical Alloying,” Mater. Sci. Eng. A204:240–245 (1995). 217. He, L., and Ma, E., “Full-Density Nanocrystalline Fe—29Al—2Cr Intermetallic Consolidated from Mechanically Milled Powders,” J. Mater. Res. 11:72–80 (1996). 218. Zhang, Y.D., Wu, M., Ge, S.H., Hui, S., Ma, X.Q., Zhang, H., Budnick, J.I., and Hines, W.A., “Consolidation and Magnetic Properties of Nanostructured Ni75Fe25,” IEEE Transactions on Magnetics 39(5, Pt. 2):3145–3147 (2003). 219. Zhao, J., “Synthesis and Characteristics of Consolidated Nanocrystalline Two-phase Ag50Ni50 Alloy by Hot Pressing,” Corrosion Science and Technology 31(4):287–290 (2002). 220. Aoki, K., Itoh, Y., Park, J., Kawamura, Y., Inoue, A., and Masumoto, T., “Preparation of Nanocrystalline Bulk TiAl by a Combination of HDDR and Hot Pressing,” Intermetallics, 4:S103–S111 (1996). 221. Axelbaum, R.L., Huertas, J.I., Lottes, C.R., Hariprasad, S., and Sastry, S.M., “Nano-Phase W and W—Ti Composite via Gas-Phase Combustion Synthesis,” Mater. Manuf. Proc. 11:1043–1053 (1996). 222. Secondi, J., Drbohlav, O., and Yavari, R., “Metal-Nitride Composites Obtained by Alloy Milling under Nitrogen Gas,” Mater. Sci. Forum 179:287–294 (1995). 223. Saida, J., Inoue, A., and Masumoto, T., “The Formation of Nanostructured Materials by Consolidating Ultrafine Amorphous Alloy Particles Prepared by Chemical Reduction,” Mater. Sci. Eng. A179–A180:577–581 (1994).

230

Processing

224. Chang, S., Doremus, R.H., Schadler, L.S., and Siegel, R.W., “Hot-Pressing of Nano-Size Alumina Powder and the Resulting Mechanical Properties,” Int. J. Appl. Ceram. Tech. 1(2):172–179 (2004). 225. Hahn, H., and Averbach, R.S., “High Temperature Mechanical Properties of Nanostructured Ceramics,” Nanostruct. Mater. 1:95–100 (1992). 226. Terwilliger, C.D., and Chiang, Y.-M., “Characterization of Chemically- and Physically-Derived Nanophase Titanium Oxide,” Nanostruct. Mater. 2:37–45 (1993). 227. Lavik, E.B., Chiang, Y.-M., Kosacki, I., and Tuller, H.L., “Enhanced Electrical Conductivity and Nonstoichiometry in Nanocrystalline CeO2−x,” in: Metastable Phases and Microstructures (R. Bormann, G. Mazzone, R.D. Shull, R.S. Averbach, and R.F. Ziolo, eds.), pp. 359–364, MRS, Pittsburgh, PA (1996). 228. Bose, S., “Application of Nanocrystalline Materials as Matrices of Composites: Processing and Performance Advantages,” Mater. Sci. Eng. A196:105–109 (1995). 229. Lences, Z., Bellosi, A., and Monteverde, F., “Factors Influencing the Crystallization and the Densification of Ultrafine Si/N/C Powders,” Mater. Chemistry Phys. 41:46–54 (1995). 230. Sekino, T., Nakajima, T., and Niihara, K., “Mechanical and Magnetic Properties of Nickel Dispersed Alumina-based Nanocomposite,” Mater. Lett. 29:165–169 (1996). 231. Araki, H., Saji, S., Okabe, T., Minamino, Y., Yamane, T., and Miyamoto, Y., “Consolidation of MA Al-10.7 at% Ti powder at Low Temperature and High Pressure of 2 GPa,” Mater. Trans. JIM, 36:465–468 (1995). 232. Krasnowski, M., and Kulik, T., “FeAl—TiN Nanocomposite Produced by Reactive Ball Milling and Hot-Pressing Consolidation,” Scr. Mater. 48(10):1489–1494 (2003). 233. Liao, S.-C., Mayo, W.E., and Pae, K.D., “Theory of High Pressure/Low Temperature Sintering of Bulk Nanocrystalline TiO2,” Acta Mater. 45:4027–4040 (1997). 234. Danforth, S.C., Symons, W., and Nilsen, K.J., “Processing of Silicon Nitride Powders,” in: Advanced Ceramic Processing and Technology 1 (J.G.P. Binner, ed.), pp. 39–71, Noyes, Park Ridge, NJ (1990). 235. Kaiser, A., Vassen, R., Stover, D., and Buchkremer, H.P., “Heat Treatment of Ultrafine SiC Powders to Reduce Oxygen Sensitivity and Grain Growth,” Nanostr. Mater. 4:795–802 (1994). 236. Borner, I., and Eckert, J., “Grain Size Effects and Consolidation in Ball Milled Nanocrystalline NiAl,” Mater. Sci. Forum, 235–238:79–84 (1997). 237. Suryanarayana, C., Korth, G.E., Chen, G.-H., Frefer, A., and Froes, F.H., “Thermal Stability of Nanostructured Titatnium Aluminides,” Nanostruct. Mater. 2:527–535 (1993). 238. Kaiser, A., Vassen, R., Stover, D., Buchkremer, H.P., Forster, J., and Uhlenbusch, J., “Si3N4//SiC Composites Using Conventional and Nanosized Powders,” Nanostruct. Mater. 6:917–920 (1995). 239. Senkov, O.N., Ovecoglu, M.L., Srisukhumbowornchai, N., and Froes, F.H., “Stability of Nanocrystallyne TiAl-based Alloys,” Nanostruct. Mater. 10: 935–945 (1998).

5: Nanocrystalline Powder Consolidation Methods, Groza 231 240. Murphy, B.R., and Courtney, T.H., “Synthesis of Cu—NbC Nanocomposites by Mechanical Alloying,” Nanostr. Mater. 4:365–369 (1994). 241. Tang, J., Wu, A., Mao, X., Li, S., Gao, W., and Du, Y., “Effects of Hot Isothermal Pressing on the Microstructures and Soft Magnetic Properties of Nanocrystalline Fe86Zr7B6Cu1 Ribbons,” J. Phys. D: Appl. Phys, 37(2): 151–154 (2004). 242. Vassen, R., Buchner, P., Uhlenbusch, J., and Stoever, D., “Production and Densification of Nanosized Cu/SiC-Powders,” Mater. Sci. Forum 269–272(Mechanically Alloyed, Metastable and Nanocrystalline Materials, Pt. 1):213–218 (1998). 243. Kwon, O.-H., Nordahl, C.S., and Messing, G.L., “Submicrometer Transparent Alumina by Sinter-Forging Seeded γ-Al2O3 Powders,” J. Am. Ceram. Soc. 78:491–494 (1995). 244. Kellet, B.J., and Lange, F.F., “Experiments on Pore Closure During Hot Isostatic Pressing and Forging,” J. Am. Ceram. Soc. 71:7–12 (1988). 245. Winnubst, A.J.A., Boutz, M.M.R., He, Y.J., Burggraaf, A.J., and Verweji, J.H., “Plasticity of Nanocrystalline Zirconia Ceramics and Composites,” Ceramics Int. 23:215–221 (1997). 246. Nordahl, C.S., and Messing, G.L., “Transformation and Densification of θ-Alumina during Sinter Forging,” J. Am. Ceram. Soc. 79:3145–3154 (1996). 247. Wetzel, K., Rixecker, G., Kaiser, G., and Aldinger, F., “Preparation of Dense Nanocrystalline Silicon Carbide Ceramics by Sinter Forging in the Presence of a Liquid Phase,” Adv. Eng. Mater. 7(6):520–524 (2005). 248. Alymov, M.I., and Leontieva, O.N., “Synthesis of Nanoscale Ni and Fe Powders and Properties of Their Compacts,” Nanostruct. Mater. 6:393–395 (1995). 249. Han, B.Q., and Lavernia, E.J., “High-Strain-Rate Superplasticity in a Nanostructured Al—Mg Alloy,” Adv. Eng. Mater. 7(4):247–250 (2005). 250. Morris, M.A., and Morris, D.G., “Microstructural Refinement and Associated Strength of Copper Alloys Obtained by Mechanical Alloying,” Mater. Sci. Eng. A111:115–127 (1989). 251. Kato, A., Suganuma, T., Horikiri, H., Kawamura, Y., Inoue, A., and Masumoto, T., “Consolidation and Mechanical Properties of Atomized Mgbased Amorphous Powders,” Mater. Sci. Eng. A179–A180:112–117 (1994). 252. Tsuruta, K., Nakano, A., Kalia, R.K., and Vashishta, P., “Dynamics of Consolidation and Crack Growth in Nanocluster-Assembled Amorphous Silicon Nitride,” J. Am. Ceram. Soc. 81:433–436 (1998). 253. Misra, R.S., Lesher, C.E., and Mukherjee, A.K., “High-Pressure Sintering of Nanocrystalline γ-Al2O3,” J. Am. Ceram. Soc. 79:2989–2992 (1996). 254. Colaizzi, J., Mayo, W.E., Kear, B., and Liao, S.-C., “Dense Nanoscale Single and Multi-phase Ceramics Sintered by Transformation Assisted Consolidation,” Int. J. Powder Metallurgy, 37:45–54 (2001). 255. Yao, B., Ding, B.Z., Wang, A.M., Li, D.J., and Hu, Z.Q.,” Preparation of a Single Phase Bulk Al—Fe(Mo, Si, B) Nanostructured Alloy Under High Pressure,” Mater. Lett. 22:81–86 (1995). 256. Valiev, R.Z., Misra, R.S., Groza, J.R., and Mukherjee, A.K., “Processing of Nanostructured Nickel by Severe Plastic Deformation of Ball Milled Powder,” Scr. Metall. Mater. 34:1443–1448 (1996).

232

Processing

257. Shen, H., Li, Z., Gunther, B., Korznikov, A.V., and Valiev, R.Z., “Influence of Powder Consolidation Methods on the Structural and Thermal Properties of a Nanophase Cu-50wt-Percent Ag Alloy,” Nanostruct. Mater. 6:385–388 (1995). 258. Clark, D.E., and Sutton, W.H., “Microwave Processing of Materials,” Annu. Rev. Mater. Sci. 26:299–331 (1996). 259. Ritzhaupt-Kleissl, H.-J., and Link, G., “Millimeter Wave Sintering of Ceramics,” Ceram. Forum Int. 76(1–2):28–32 (1999). 260. Link, G., Ivanov, V., Paranin, S., Khrustov, V., Bohme, R., Muller, G., Schumacher, G., Thumm, M., and Weisenburger, A., “A Comparison of mm-Wave Sintering and Fast Conventional Sintering of Nanocrystalline Al2O3,” in: Microwave Processing of Materials V (M.F. Iskander, J.O. Kiggans, Jr., and J.-C. Bolomey, eds.), MRS 430, pp. 157–162, MRS Pittsburgh, PA (1996). 261. Freim, J., McKittrick, J., Katz, J., and Sickafus, K., “Microwave Sintering of Nanocrystalline γ-Al2O3,” Nanostruct. Mater. 4:371–385 (1994). 262. Nygren, M., and Shen, Z., “Spark Plasma Sintering: Possibilities and Limitations,” Key Eng. Mater. 264–268(Pt. 1, Euro Ceramics VIII):719–724 (2004). 263. Chen, W., Anselmi-Tamburini, U., Garay, J.E., Groza, J.R., and Munir, Z.A., “Fundamental Investigations on the Spark Plasma Sintering/Synthesis Process,” Mater. Sci. Eng. A394(1–2):132–138 (2005). 264. Groza, J.R., Garcia, M., Schneider, J.A., “Surface Effects in Field-Assisted Sintering,” J. Mater. Res. 16:286–292 (2001). 265. Misra, R.S., Mukherjee, A.M., and Yamazaki, K., “Effect of TiO2 Doping on Rapid Densification of Alumina by Plasma Activated Sintering,” J. Mater. Res. 11:1144–1148 (1996). 266. Zhao, Z., Buscaglia, V., Viviani, M., Buscaglia, M.T., Mitoseriu, L., Testino, A., Nygren, M., Johnsson, M., and Nanni, P., “Grain Size Effects on the Ferroelectric Behavior of Dense Nanocrystalline BaTiO3 Ceramics,” Phy. Rev. B 70:024107/1–024107/8 (2004). 267. Kimura, H., and Kobayashi, S., “Synthesis of High Hardness Intermetallic Compound by Pulse Electro-Discharge Consolidation of Amorphous TiAl Powder,” J. Japan Inst. Met. 58:201–207 (1994). 268. Zhan, G.-D., Kuntz, J.D., Garay, J.E., and Mukherjee, A.K., “Electrical properties of nanoceramics reinforced with ropes of single-walled carbon nanotubes,” Appl. Phys. Lett. 83:1228–1230 (2003). 269. Ivanov, V., Kotov, Y.A., Samatov, O.H., Bohme, R., Karow, H.U., and Schumacher, G., “Synthesis and Dynamic Compaction of Ceramic Nanopowders by Techniques Based on Electric Pulsed Power,” Nanostruct. Mater. 6:287–290 (1995). 270. Chelluri, B., Barber, J.P., Gonzales, E.J., and Thadani, N.N., “Nano Grain Size Component Fabrication Using Dynamic Magnetic Compaction (DMC) Process,” Adv. Powder Met. Particulate Mater. 3:3-65–3-76 (1997). 271. Kondo, K.I., and Sawai, S., “Fabricating Nanocrystalline Diamond Ceramics by a Shock Compaction Method,” J. Am. Ceram. Soc. 73:1983–1991 (1990).

5: Nanocrystalline Powder Consolidation Methods, Groza 233 272. Suryanarayana, C., Korth, G.E., and Froes, F.H., “Compaction and Characterization of Mechanically Alloyed Nanocrystalline Titanium Aluminides,” Metall. Mater. Trans. 28A:293–302 (1997). 273. Glade, S.C., and Thadhani, N.N., “Shock Consolidation of Mechanically Alloyed Amorphous Ti—Si Powders,” Metall. Mater. Trans. 26A:2565–2569 (1995). 274. Hirai, H., and Kondo, K., “Shock-Compacted Si3N4 Nanocrystalline Ceramics: Mechanisms of Consolidation and of Transition from Alpha—to Beta– Form,” J. Am. Ceram. Soc. 77:487–492 (1994). 275. Dogan, C.P., Rawers, J.C., Govier, R.D., and Korth, G., “Mechanical Processing, Compaction, and Thermal Processing of Alpha-Fe powder,” Nanostruct. Mater. 4:631–644 (1994).

6 Electrodeposited Nanocrystalline Metals, Alloys, and Composites Uwe Erb and Karl T. Aust Department of Materials Science and Engineering, University of Toronto, Toronto, Ontario, Canada Gino Palumbo Integran Technologies Inc., Toronto, Ontario, Canada

6.1 Introduction Since their first introduction in the early 1980s, nanocrystalline materials have been made by numerous synthesis techniques. On the basis of the underlying fundamental physical and chemical principles, most of the more than 200 different processing routes that have been developed over the past 20 years can be classified into five distinct groups. The first three groups are vapor phase processing (e.g., physical vapor deposition, chemical vapor deposition, inert-gas condensation), liquid-phase processing (e.g., rapid solidification, atomization, sonication of immiscible liquids), and solid state processing (e.g., equal channel angular pressing, mechanical attrition, annealing of amorphous precursors). The other two groups are chemical synthesis (e.g., sol-gel processing, inverse micelle technology, sonochemical synthesis) and electrochemical synthesis (e.g., electrodeposition, electroless deposition, galvanic conversion). In many cases the latter two general processing routes are closely related in terms of the chemical reactions involved. However, the distinguishing feature of electrochemical synthesis is that it involves an interface at which charge transfer occurs. All electrochemical reactions have well-defined cathodic and anodic reactions. In the case of electrodeposition, an external power supply (direct current or pulsed current) is required to drive the reaction, while in electroless deposition the electrons for metal ion reduction come from a reducing agent added to the base metal electrolyte (e.g., hypophosphite in the case of electroless Ni—P).[1] Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 235–292 © 2007 William Andrew, Inc.

235

236

Processing

Electrodeposition can be used to make many different materials including pure metals, alloys, metal matrix composites, oxides, semiconductors, and conductive polymers.[1] As a surface coating technology, electrodeposition of conventional metals has a very long history, dating back more than 200 years for some metals and alloys.[1,2] Today electrodeposition is much more than just a coatings technology. In addition to application to decorative wear and corrosion-, erosion-, and oxidation-resistant coatings, electrodeposition technologies have been advanced for many different applications including large thick-walled (>1 mm) electroformed products such as molds for thermoplastic composite component manufacture, functional coatings for magnetic and electronic applications, or the manufacture of microelectromechanical system components. The commercial importance of electrodeposition is enormous and can be estimated by looking at the volume of materials used in terms of metal salts and anode materials. For the metal nickel alone, the worldwide consumption is on the order of 100 000 tonnes per year.[1] This chapter addresses the synthesis, structure, and properties of nanocrystalline metals, alloys, and metal–matrix composites made by electrodeposition from aqueous solutions. Not included in this chapter are semiconductors, polymers, compositionally modulated superlattices, oxides, and materials produced by electrodeposition from ionic liquids, molten salts, or organic solutions. In order to treat electrodeposited nanomaterials with sufficient focus, this review is limited to structures in which the dominant phase has a crystal size of less than 100 nm and in which crystals are separated by high-angle (>15°) grain boundaries. Not included are the many papers that explain various phenomena of electrodeposited materials on the basis of nanostructure effects but without presenting sufficient microstructural characterization to substantiate such claims. Also excluded is the research that deals with electrodeposited structures in which nanosize refers to the size of crystallites or domains separated by small-angle boundaries within a matrix consisting of large (typically micrometer-sized) grains. While such deposits have quite interesting structures, their properties are better explained in the context of conventional polycrystalline electrodeposits.

6.2 Synthesis of Nanostructured Materials by Electrodeposition There are numerous early reports in the literature describing electrodeposits with ultrafine structures; many examples are given in Ref. 1. However, no systematic studies were published before the late 1980s[3,4]

6: Metals, Alloys, and Composites, Erb et al.

237

on the synthesis of nanocrystalline materials by electrodeposition in an attempt to optimize certain properties by deliberately controlling the volume fractions of grain boundaries and triple junctions in the material. In fact, the synthesis of nanostructured materials with grain size control during the electrodeposition process can be considered a distinct form of grain boundary engineering in which the grain boundary content (types and quantities of grain boundaries) of a material are controlled during material processing to achieve certain physical, chemical, and mechanical properties.[5–7] The final result is a bulk interfacial material, as originally defined by Gleiter,[8] which does not require any further processing of precursor powder material. In this respect, electrodeposited nanomaterials are quite different from other nanostructures that are based on consolidated particles. To date, a large number of nanostructured pure metals, alloys, and metal–matrix composites have been produced by electrodeposition (see Table 6.1 for examples).[9–40] Crystalline metal electrodeposits exhibit several growth forms including layers, blocks, pyramids, ridges, spiral growth forms, dendrites, powders, and whiskers.[41] These morphologies have been studied extensively and various models have been advanced to correlate specific growth forms with electrodeposition parameters and substrate microstructure.[41,42] Electrodeposition parameters include bath composition, pH, temperature, overpotential, bath additives, etc., while important microstructural features of the substrate are grain size, crystallographic texture, dislocation density, and internal stress.[41,42] Electrocrystallization (Fig. 6.1) occurs either by the build-up of existing crystals or by the formation of new ones.[43] These two processes are in competition with each other and are influenced by different factors. The two key mechanisms that have been identified as the major ratedetermining steps for nanocrystal formation are charge transfer at the electrode surface and surface diffusion of adions on the crystal surface.[44] Earlier, Fischer presented a classification of microstructures typically observed in electrodeposits.[45] One of the key factors in the microstructural evolution of electrodeposits in terms of grain size and shape is inhibition, for example, resulting from reduced surface diffusion of adions by adsorption of foreign species (such as grain refiners) on the growing surface. With increasing inhibition, the deposit structure changes from basis-oriented and reproduction type (BR) to twin transition types (TT), to field-oriented type (FT), and finally to unoriented dispersion type (UD).[46] A large number of grain refiners have been described in the literature (see for example, Ref. 47); their effectiveness depending upon surface adsorption characteristics, compatibility with the electrolyte, and

Processing

238 Table 6.1 Examples of Nanocrystalline Metals, Alloys and Metal Matrix Composites That Have Been Produced by Electrodeposition

Material

References

Ni Co Pd Cu Zn Ni—P Ni—Fe Ni—Zn Co—W Co—Fe

[9, 10, 11, 12] [13, 14] [15] [12, 16] [17] [3, 4] [18, 19] [20, 21, 22] [23] [24]

Pd—Fe Ni—Fe—Cr Fe—Co—Ni Ni—Zn—P Co—Fe—P Ni—Al2O3 Ni—SiC Cu–Al2O3 Ni—P—BN Ni—MoS2 Ni—Al (particles) Ni—Nanocarbon tubes

[25] [26, 27, 28] [29] [30, 31] [32] [10, 33, 34] [10, 35, 36] [37] [38] [32, 38] [39] [40]

temperature stability. For example, saccharin,[48] coumarin,[14] thiourea,[14] and formic acid[49] have all been successfully applied to achieve grain refinement down to the nanocrystalline range for nickel electrodeposits. The second important factor in nanocrystal formation during electrocrystallization is overpotential.[43,44] Grain growth is favored at low overpotential and high surface diffusion rates. On the other hand, high overpotential and low diffusion rates promote the formation of new nuclei.

6: Metals, Alloys, and Composites, Erb et al.

239

x C∞

Me Me

x+

x+

Nucleation

Me

x+

Growth

CMex+

xe

-

xe

-

Surface Diffusion

Figure 6.1 Schematic diagram showing nucleation, growth, and surface diffusion during electrocrystallization. Also shown is the concentration of metal ions in the electrolyte as a function of distance x from the cathode surface.

The latter conditions can be experimentally achieved when using pulse plating, where the peak current density can be considerably higher than the limiting current density attained for the same electrolyte during direct current plating.[11,14] While many of the processes associated with the crystallization stage (see Fig. 6.1) are still poorly understood, previous work has shown that electrodeposition will result in nanostructured materials when electrodeposition variables (e.g., bath composition, pH, temperature, current density, etc.) are chosen such that electrocrystallization results in massive nucleation and reduced grain growth. Under these conditions, the effect of the substrate on the structure of the resulting bulk electrodeposit becomes negligible (for example, see Ref. 50).

6.3 Structure of Nanocrystalline Metal Electrodeposits This chapter deals mainly with equiaxed nanostructured electrodeposits, although grain-shape-modified structures can also be synthesized by electrodeposition.[51] Figure 6.2 shows bright-field, dark-field, and diffraction pattern images, and grain size distribution of a nanocrystalline Ni–2.8 wt% P specimen produced by direct current plating from a

Processing

240

a

100 nm

100 nm

b

As-plated Ni-2.8wt%P d=6.1 nm

Frequency

0.2

0.1

0.0

c

d

2

4

6

8

10

12

14

16

18

20

Grain Size (nm)

Figure 6.2 (a) Bright-field image, (b) dark-field image, (c) electron diffraction pattern, and (d) grain size distribution of a nanocrystalline Ni–2.8 wt% P electrodeposit with an average grain size of 6.1 nm.[52]

modified Watts bath.[52] Electrodeposition of nanocrystals typically operates far from equilibrium conditions. Consequently, the resulting materials are nonequilibrium structures, which is primarily manifested in the small grain size and the associated large volume fraction of high-angle grain boundaries and triple junctions. In addition, alloys produced by this method can show considerable extensions of the solid solubility range similar to what is observed in materials produced by other nonequilibrium processing routes, such as rapid solidification. For example, the roomtemperature solid solubility for P in Ni is negligible.[53] On the other hand, electrodeposited Ni—P can form solid solutions containing phosphorus levels of 10 wt% or more.[3,4] Similarly, extended solubility ranges were also observed in other alloys, such as Co—W,[23] or Zn—Ni.[20] Depending on the electrodeposition parameters, the grain size distribution of electrodeposited nanocrystals can be relatively narrow as shown, for example, for Ni—P in Fig. 6.2(d). The crystallographic texture depends strongly on the electroplating parameters, as demonstrated in Fig. 6.3 for a series of pulse plated Ni—P alloy nanocrystals.[54] In this example, the crystallo-

6: Metals, Alloys, and Composites, Erb et al.

241

graphic texture changes from a (111) (200) double fiber texture to a strong (111) texture with increasing phosphorus content in the deposit. The series of X-ray diffraction scans presented in Fig. 6.3 also show that the phosphorus concentration in the deposit has a strong effect on the grain size of the material. This is evident from the increasing line broadening with increasing phosphorus concentration. High-resolution electron microscopy (Fig. 6.4) has revealed that the grain boundary structure in electrodeposited nanocrystals is similar to the structure found in conventional polycrystalline materials.[55] This finding is in agreement with previous results by Thomas et al.[56] but in contrast to earlier work by Wunderlich et al.[57] who observed extended grain boundary structures in nanocrystalline palladium produced by inert-gas condensation. Using position annihilation spectroscopy, Würschum et al.[15] reported relatively large free volumes in electrodeposited nanocrystalline Pd, which they described as nanopores (4 missing atoms) or nanovoids (10–15 missing atoms) containing light impurity atoms. Similarly, large volumes were also found in commercially available nanocrystalline Ni electrodeposits.[58] For other electrodeposited nanocrystals, however,

Peak Intensity (a.u.)

Ni-5.9wt%P Ni-3.4wt%P Ni-2.5wt%P Ni-1.8wt%P Ni-1.2wt%P Ni-0.5wt%P Poly Ni 40

50

60

70

80

90

100

110

120

130

140

Diffraction Angle (2 θ) Figure 6.3 X-ray diffraction scans showing grain size reduction (line broadening) and changes in crystallographic texture, with increasing phosphorus content in electrodeposited nanocrystalline Ni—P alloys. Also shown for comparison is the X-ray scan for conventional polycrystalline Ni powder.[54]

242

Processing

Figure 6.4 High-resolution electron micrograph of a nanocrystalline Ni–1.9 wt% P electrodeposit. (Courtesy of S. Mehta and D. A. Smith.)

porosity is usually negligible as demonstrated, for example, for Ni deposits by detailed density measurements[59] and positron annihilation spectroscopy.[60] Porosity-free electrodeposited nanocrystals are a distinct form of grain boundary engineered materials. If the intercrystalline region of a material is considered to consist of grain boundary and triple junction defects, the influence of these defects on the bulk properties of nanocrystalline materials will depend upon their relative volume fractions. A three-dimensional treatment involving tetrakaidecahedral grains, where grain boundaries are represented by the faces of the polyhedron and triple junctions by the edges, has been applied,[5] and more recently generalized to any grain shape.[61] Figure 6.5 shows calculated volume fractions for the grain boundary, triple junction, and total intercrystalline components in the grain size range from 2 nm to 1000 nm, calculated for a boundary thickness of 1 nm. The intercrystalline volume fraction increases from a value of ∼0.3% at 1000 nm to ≥50% at grain sizes smaller than 5 nm. In the range 100 nm to 2 nm, the triple junction volume fraction increases by three orders of magnitude, while the grain boundary volume fraction increases by a little over one order of magnitude. The grain boundary volume fraction also shows a plateau at a grain size of ∼3 nm, while the triple junction volume fraction continues to increase and becomes equivalent to the grain boundary volume fraction at a grain size of ∼2 nm. A comparison of cross-sectional structure evolution with increasing deposit thickness reveals the main difference between conventional electrodeposits and nanocrystalline deposits (Figs. 6.6 and 6.7). Conventional

6: Metals, Alloys, and Composites, Erb et al.

Volume Fraction

100

243

Total Interface Grain Boundary Triple Junction

10-1

10-2

10-3

Δ = 1 nm

100

101

102

103

Grain Size (nm)

Figure 6.5 Volume fractions of atoms associated with the interfaces in fully dense nanocrystalline materials as a function of grain size assuming a regular 14-sided tetrakaidecahedron as the grain shape and a grain boundary thickness Δ of 1 nm.

electrodeposits (Fig. 6.6) often show an initial layer of very fine-grained material close to the substrate. However, with increasing deposit thickness, a transition takes place from nucleation-controlled crystallization to growth-dominated deposition. As a consequence, there is a structural transition from fine-grained-equiaxed to coarse-grained columnar structure with strong grain shape anisotropy. The thickness of the initial finegrained layer depends on the electrodeposition parameters, such as metal ion concentration, bath pH, or agitation.[62] On the other hand, if growth conditions are maintained to favor massive nucleation regardless of coating thickness, nanocrystalline deposits showing no grain shape anisotropy can be grown to any thickness (Fig. 6.7).

Processing

244 Conventional plating – strong grain shape anisotropy 5 µm

b

Growth Direction

a

Poly Ni

250 µm

Substrate

Substrate 10 µm

Figure 6.6 Cross-sectional structure evolution for a conventional polycrystalline nickel electrodeposit. Nano plating – no grain shape anisotropy 5 µm

b

Nano Ni

Growth Direction

a

Substrate

Substrate

100 nm

10 µm

Figure 6.7 Cross-sectional structure evolution for a nanocrystalline nickel electrodeposit.

Computer-controlled variations of current density or other deposition parameters (e.g., agitation, bath temperature) during electrodeposition allow for a great variety of grain size-modulated microstructures, as shown in Fig. 6.8. Such structures are of interest in efforts to optimize specific property requirements, e.g., high strength combined with high ductility or high strength at high electrical conductivity.

6: Metals, Alloys, and Composites, Erb et al.

a

245

b

c

Figure 6.8 Schematic diagrams showing structure modulations in nanocrystalline electrodeposits: (a) bimodal grain size structure; (b) layered structure; (c) grain size gradient structure.

a

c

b

d

Figure 6.9 Examples of nanocomposite materials made by electrodeposition using (a) conventional particles, (b) whiskers, (c) nanoparticles, and (d) nanotubes.

Other property optimizations can be achieved by various types of composite nanostructures as shown schematically in Fig. 6.9. There are two possible approaches to synthesizing nanocomposite materials. In the first approach the second phase (e.g., SiC, Al2O3, MoS2, carbon nanotubes) is added to the electroplating solution for the metal matrix (e.g., Ni, Co, Cu) and co-deposited during electrodeposition. Basically, this is the same

246

Processing

approach that has been used in the electroplating industry for the deposition of polycrystalline composite deposits for many years. For nanocomposite synthesis, either conventional large particles[10,32,35,36,38] or newly developed nanopowders, nanotubes, or nanofibers[33,40] can be codeposited. It should be pointed out that the prevention of second phase agglomeration becomes more difficult as their size is reduced to the nanometer range. The second approach to nanocomposite formation is an in-situ two-step process.[63] In the first step, the nanocrystalline matrix is deposited as a supersaturated solid solution. This is followed by an annealing step during which the second phase is precipitated at a temperature at which little grain growth of the matrix occurs. In other words, this process makes use of the age hardenability of specific nanocrystalline alloy systems[64], as will be discussed in section 6.4.2.3.

6.4 Properties Electrodeposited nanomaterials have been studied extensively over the past 20 years because of (i) their relative uniqueness compared with nanostructured materials prepared by other synthesis routes (e.g., fully dense interface-engineered materials in a single-step process) and (ii) their early adoption in numerous industrial applications (see section 6.5). Many physical, chemical, and mechanical properties have been assessed as a function of grain size covering a wide range from less than 10 nm to more than 1 μm (i.e., their polycrystalline counterparts). While the effect of grain size on some properties (e.g., saturation magnetization) is now quite well understood, the understanding of other properties (e.g., some mechanical properties) is still far from complete. Today, the largest database available on properties of electrodeposited nanomaterials is for nanocrystalline nickel and nickel alloys (see ref. 64 for a recent review). Extensive property measurements were performed on these materials during the development of the first large-scale industrial application for structural nanomaterials more than 10 years ago. This so-called electrosleeve technology was developed in the early 1990s as an in-situ repair technology for nuclear steam generator rehabilitation and applied in both Canadian CANDU and US pressurized water reactors.[65,66] In this process, steam generator tubes (e.g., Alloys 600 and 400) whose structural integrity was compromised by localized degradation phenomena (e.g., intergranular stress corrosion cracking, pitting) were repaired by coating the inside of the tubes with a thick (∼1 mm) nanocrystalline Ni

6: Metals, Alloys, and Composites, Erb et al.

247

microalloy to restore a complete pressure boundary. The full approval of this material by both Canadian and US regulatory bodies under ASME Code-Case N-569 required a thorough understanding of many of its properties, including mechanical properties, thermal expansion, corrosion properties, ferromagnetic properties, and thermal stability. This section gives extensive coverage of the properties of nanocrystalline nickel and nickel-based alloys/composites. This will be complemented with properties of other nanocrystalline electrodeposits for which similar data are available in the literature. Comparisons with conventional polycrystalline materials and nanomaterials prepared by other synthesis methods will also be presented. The section is organized in the following way. Properties that show little grain size dependence will be presented first. These include thermal expansion, specific heat, saturation magnetization and Young’s modulus. This will be followed by properties that show strong grain size dependence such as hardness, yield strength, ductility, wear properties, corrosion behavior, electrical resistivity, and thermal stability.

6.4.1 Properties with Weak Grain Size Dependence Figure 6.10 shows that, over a very large range, grain size has no major effect on properties of bulk nanocrystalline nickel such as saturation magnetization, heat capacity, thermal expansion, and Young’s modulus. These properties are affected by grain size only at very small grain sizes (e.g., <20 nm for the case of the Young’s modulus[52]).

6.4.1.1 Saturation Magnetization Conflicting results have been reported regarding the grain size/crystal size dependence of certain magnetic properties. While the understanding of the magnetic structure of nanostructured materials is still far from complete, a clear picture is now emerging regarding the saturation magnetization, Ms, where the early contradictory results can be explained in terms of the chemical composition and microstructure of the nanocrystalline materials. Initially many studies reported that, for nanocrystalline materials, there is a large reduction in saturation magnetization with decreasing grain size.[67–71] Gleiter first reported a 40% decrease in saturation magnetization

Processing

248 0.8

a

b

0.7 HEAT CAPACITY (J/g°C)

SATURATION MAGNETIZATION (kA/m)

600

500

400

0.6 0.5 0.4 0.3 0.2 0.1 0.0

0

100

101

102

103

104

105

100

101

102

103

104

105

GRAIN SIZE (nm)

GRAIN SIZE (nm)

400

14

c YOUNG'S MODULUS (GPa)

-1

THERMAL EXPANSION (K )x10

-6

d

12

10

8

300

200

100

0

0

100

101

102

103

GRAIN SIZE (nm)

104

105

100

101

102

103

104

105

GRAIN SIZE (nm)

Figure 6.10 (a) Saturation magnetization, (b) heat capacity, (c) thermal expansion, and (d) Young’s modulus as a function of grain size for bulk nanocrystalline nickel electrodeposits.

(compared to bulk polycrystalline α-iron) for nanocrystalline iron with 6 nm grain size that was produced by consolidating nanocrystalline particles produced by the inert-gas condensation technique.[67] This behavior was attributed to differences in the magnetic microstructure between nanocrystalline and conventional polycrystalline iron. Similarly, strong effects of particle size on saturation magnetization were observed in the study of uncompacted ultrafine particles produced by the gas evaporation method.[68,69] In the case of ultrafine particles (10–50 nm) of Ni, Co, and Fe, Gong et al.[68] observed a rapid decrease in saturation magnetization with decreasing grain size, which they attributed to antiferromagnetic oxide layers on the ultrafine metal particles. In

6: Metals, Alloys, and Composites, Erb et al.

249

another study on ultrafine particles it was found that the normalized magnetization ratio decreases with decreasing particle diameter.[69] The reduction in saturation magnetization was linked to surface effects, which were considered more important in the case of smaller particles. Schaefer et al.[70] also noted a decrease in Ms in consolidated nanocrystalline nickel powder produced by gas evaporation, which they explained in terms of structural disorder of the interfaces. The magnetic moment of the interfacial atoms was calculated to be nearly half that of the atoms in the bulk material. Furthermore, Yao et al.[71] also found that the saturation magnetization of ultrafine Ni particles decreases drastically with decreasing grain size. Krill et al.[72] reported that the spontaneous magnetization of nanocrystalline Gd samples produced by gas condensation and subsequent compaction was approximately 75% of the value for polycrystalline Gd. It should be noted that all of the previous materials were produced using the gas condensation method, which yields materials with high internal porosity that can provide large surface areas for oxide formation after exposing the samples to air. In contrast, Aus et al.[73] reported that the saturation magnetization of bulk nanocrystalline Ni was not strongly dependent on the grain size (Fig. 6.10(a)). In this study, the grain sizes of Ni varied from 10 μm to 10 nm, and for the Ni samples with the smallest grain size, the observed Ms was only 10% less than for conventional polycrystalline Ni. These results were obtained on bulk nanocrystalline Ni produced by electrodeposition and were explained in terms of the negligible porosity/oxide formation in this material. The results by Aus et al.[73] agree well with results of calculations that assessed the effect of structural disorder, introduced by grain boundaries, on the magnetic properties of nanocrystalline metals.[74–76] In these studies, grain boundary configurations representing various degrees of disorder were generated using molecular dynamics simulations with embeddedatom potentials. The boundaries ranged from Σ3 boundaries with minimum structural disorder through Σ5 and Σ13 special grain boundaries of intermediate structural disorder to random amorphous grain boundaries with maximum disorder. Electronic structure calculations were performed using the tight-binding linear muffin-tin orbital atomic-sphere-approximation method. These calculations have shown that the magnetic moment is rather insensitive to the degree of structural disorder associated with grain boundaries. Even when the entire material was amorphous, the average moment was found to be reduced by only 15%. It was concluded that, for the case of nanocrystalline Ni with a grain size of 10 nm at which the grain boundary atoms comprise about 30% of the volume (Fig. 6.5),

250

Processing

the overall effect of structural disorder on the average moment is very small, in good agreement with the experimental data reported by Aus et al.[73] for electrodeposited nickel. More recently, there have been other reports confirming the early results by Aus et al.[71] for nanocrystalline Ni. For example, Daroczi et al.[77,78] reported for nanocrystalline Ni prepared by ball milling that there is no observable difference in Ms for materials with 7 nm and 50 μm grain sizes. Bakonyi et al.[12] observed the same trend for nanocrystalline Ni, also prepared by electrodeposition. Weissmüller, et al.[79] confirmed the earlier measurements by Aus et al.,[73] reporting only small changes in Ms for electrodeposited nanocrystalline Ni with 18 nm grain size. Kisker et al.[80] presented further results for gas-condensed Ni which, in contrast to their earlier work,[70] now showed the saturation magnetization to be independent of grain size as long as the gas-condensed material was not exposed to air. However, after exposure to air, Ms decreased to about 80% of its original value. Aus et al.,[81] Szpunar et al.,[82] and Cheung et al.[83] have recently presented further experimental evidence and detailed calculations for nanocrystalline Ni—Fe, Ni—P, Co, Co—W, and Co—Fe that further support the earlier findings that structural disorder introduced by grain boundaries and triple junction has an insignificant effect on saturation magnetization compared with chemical disorder introduced by alloying additions.

6.4.1.2 Thermal Expansion and Heat Capacity Thermal expansion, αL, and specific heat, CP, measurements on nanocrystalline electrodeposits[60,84] showed quite different results from those reported earlier on nanostructures produced by inert-gas condensation[85] or crystallization of amorphous precursors.[86] Rupp and Birringer[85] found αL for gas-condensed nanocrystalline copper of 8 nm grain size to be nearly twice that of regular polycrystalline copper. However, the nanocrystalline copper was only 90% dense. Heat capacities of nanocrystalline copper and palladium were also found to be increased by 10% and 40%, respectively.[85] Thermodynamic models based on quasiharmonic approximations used to predict theoretically the thermodynamic properties as a function of grain boundary free volume have produced similar increases in αL (70–85%) and CP (10–25%) for nanocrystalline palladium with respect to conventional polycrystalline palladium.[87] Fecht’s calculations[88] also predict increases by a factor of 2 in both volumetric thermal expansion and heat capacity. Lu et al.[86]

6: Metals, Alloys, and Composites, Erb et al.

251

observed an increase in αL of about 60% and 12% in CP for Ni—P alloys crystallized from amorphous precursors. However, these materials consisted of two phases (fcc Ni and bct Ni3P) and cannot be compared directly with single-phase materials. Gleiter[89] showed that the differences in αL between nanocrystalline and conventional polycrystalline copper depend strongly on the applied pressure during consolidation of gas condensed powder. The general trend was that this difference decreased with increasing compaction pressure. Some compacted nanocrystals even showed a slight reduction of αL from the value of polycrystalline material. In contrast, for electrodeposited nanocrystals, neither thermal expansion nor specific heat showed significant differences when compared to their polycrystalline counterparts.[60,84] Figures 6.10(b) and 6.10(c) show the grain size dependence of heat capacity and thermal expansion for fully dense electrodeposited nickel for grain sizes from 10 nm to about 10 μm. Overall the effect of grain size on these properties is rather small for the temperature range between 200 and 400 K. The weak grain size dependence of the thermal expansion for this grain size range is consistent with results from molecular-dynamics simulations with embedded-atom method potentials, which showed that the thermal expansion of nickel in the presence of special grain boundaries is only slightly affected.[90]

6.4.1.3 Young’s Modulus Initial studies on nanocrystalline metals prepared by the inert-gas condensation method showed significant reductions in the Young’s modulus compared with their conventional polycrystalline counterparts.[91–93] It was speculated that these reductions were due to interfacial effects or residual porosity. Krstic et al.[94] and Zugic et al.[95] showed that large reductions in the Young’s modulus in metals with very high strength resulting from reduced dislocation mobility, and containing residual porosity, can actually be predicted by applying a porosity model previously developed for other strong solids such as ceramics. This model shows that the reduction in the elastic response of the material can be considered a result of the crack opening displacement of annular flaws associated with the pores. Subsequent measurements on gas condensed nanocrystals with lower residual porosity confirmed that the initially reported large reductions in Young’s modulus were indeed mainly the result of residual porosity.[96] As shown in Fig. 6.10(d), the grain size in fully dense, free-standing nanocrystalline nickel electrodeposits has virtually no effect on the

252

Processing

Young’s modulus for grain sizes larger than 20 nm. The Young’s modulus data for the two smallest grain sizes in Fig. 10(d) were for nickel alloys containing about 2.5 wt% phosphorus.[97] In fact, Zhou et al.[98] recently measured the Young’s modulus of a large number of nanocrystalline Ni–2.5 wt% P alloy electrodeposits with grain sizes in the range 4–30 nm. Also included in the study were (i) pure nanocrystalline nickel with a grain size of 62 nm and (ii) a fully amorphous Ni–15 wt% P alloy. The results are shown in Fig. 6.11 in the form of a normalized E/E0 plot, where E0 is the Young’s modulus of conventional polycrystalline nickel. For nickel–phosphorus deposits with grain sizes larger than 20 nm and pure nano-nickel with 62 nm grain size, the Young’s modulus is essentially the same as for polycrystalline Ni. However, a reduction in the Young’s modulus can be seen for grain sizes less than 20 nm, down to a value of about 86% of the value for conventional polycrystalline Ni at the smallest grain size of 4 nm. Zhou et al.[97,98] applied a composite model that was previously developed by Shen et al.[99] to describe the Young’s modulus reductions observed in ball-milled nanomaterials at very small grain sizes. In this model, the nanocrystalline material is considered as a composite material consisting of grain interiors, grain boundaries, and triple junctions. Using the measured Young’s modulus values and the relative volume fractions for grain boundaries and triple junctions (Fig. 6.5), Zhou et al.[97,98] determined the ranges for upper-bound and lower-bound solutions for Ni– 2.5 wt% P as 200–220 GPa, 137–172 GPa, and 132–187 GPa for the grain interiors, grain boundaries, and triple junctions, respectively. With these values, the overall upper- and lower-bound solutions shown in Fig. 6.11 give a reasonable approximation of the Young’s modulus at grain sizes less than 20 nm, similar to the observations by Shen et al.[99] It is interesting to note in Fig. 6.11 that the composite model can also describe the Young’s modulus of the amorphous electrodeposit, suggesting a smooth transition from the nanocrystalline to the amorphous structure. In a recent study, Somekawa et al.[100] presented Young’s modulus data for electrodeposited Ni–19.5 at% W alloys with both amorphous and mixed nanocrystalline (6–7 nm)/amorphous structures. The Young’s modulus of the amorphous material was also about 10% lower than for the mixed nanocrystalline/amorphous material. It should be noted that the Young’s modulus reported for conventional electrodeposits can vary greatly even for the same material.[101] Such variations are mainly due to (i) difficulties in measuring the elastic properties of thin coatings on a substrate,[1] (ii) the presence of pores and codeposited hydrogen, and (iii) most importantly, crystallographic texture.

6: Metals, Alloys, and Composites, Erb et al.

253

1.0

0.9

E / E0

Upper bound maximum

0.8

Lower bound minimum Amorphous Ni-P Ni-P nano Ni

0.7

0

10

20

30

40

50

60

70

80

90

Grain Size (nm)

Figure 6.11 Young’s modulus (normalized with respect to Young’s modulus of polycrystalline nickel, E0) of nanocrystalline Ni–2.5 wt% P alloys, pure nanocrystalline nickel, and amorphous Ni–15 wt% P.

6.4.2 Properties with Strong Grain Size Dependence This section presents an overview of the properties of electrodeposited nanomaterials that exhibit strong grain size dependence. Figure 6.12 summarizes the results of several of our studies which looked at grain size effects in nickel: Vickers hardness,[65,66,102,103] yield strength,[65,66,104] tensile elongation,[65,66,104] wear rate,[103] Taber wear index,[105] coefficient of friction,[103] and electrical resistivity.[106,107] Also shown in Fig. 6.12(a) are the volume fractions for grain boundaries, triple junctions, and total interface component on the same log scale for comparison. It can be seen that the properties shown in Fig. 6.12 scale differently with grain size/interfacial volume fractions.

6.4.2.1 Hardness, Yield Strength, and Tensile Ductility Both hardness (Fig. 6.12(b)) and yield strength (Fig. 6.12(c)) of nickel show considerable increases with decreasing grain size, following initially the well-known Hall–Petch[108,109] behavior. A decrease in grain size by three orders of magnitude results in hardness/strength increases by factors of 4 to 5. Following a maximum in hardness/strength at approximately 10–12 nm, the inverse Hall–Petch behavior is observed as initially

Processing

254 8

a

b

7 VICKER'S HARDNESS (GPa)

VOLUME FRACTION

100

10-1 INTERCRYSTAL GRAIN BOUNDARY TRIPLE JUNCTION

10-2

6 5 4 3 2 1

10-3

0 100

101

102

103

104

100

105

101

1000

103

104

105

60

c

d 50 TENSILE ELONGATION (%)

800

YIELD STRENGTH (MPa)

102

GRAIN SIZE (nm)

GRAIN SIZE (nm)

600

400

200

40

30

20

10

0

0

100

101

102

103

GRAIN SIZE (nm)

104

105

100

101

102

103

104

105

GRAIN SIZE (nm)

Figure 6.12 Summary of interfacial volume fractions (a) and properties of nanocrystalline Ni electrodeposits showing strong grain size dependence: (b) Vickers hardness; (c) yield strength; (d) tensile elongation.

reported by Chokshi et al.[110] for nanocrystalline Cu and Pd made by the inert-gas condensation technique, and by McMahon and Erb[3] for Ni—P alloys made by electrodeposition. As dislocation mechanisms are unlikely to operate at such small grain sizes, Chokshi et al.[110] interpreted their results in terms of room temperature Coble creep, arising from the disorder associated with large intercrystalline volume fractions. Palumbo et al.[111] pointed out that the onset of decreasing hardness in electrodeposited Ni—P nanomaterials, i.e., the deviation from regular Hall–Petch behavior, occurs at a grain size less than 20 nm where triple lines begin to comprise

6: Metals, Alloys, and Composites, Erb et al. 40

150

e

f

125

TABER WEAR INDEX

WEAR RATE (mm3/mm)x10-5

255

100

75

50

30

20

25

0

0

100

101

102

103

104

100

105

101

1.0

103

104

105

25

g

h μΩcm) ELECTRICAL RESISTIVITY (μ

0.9 COEFFICIENT OF FRICTION

102

GRAIN SIZE (nm)

GRAIN SIZE (nm)

0.8 0.7 0.6 0.5 0.4 0.3 0.2

20

15

10

5

0.1

0

0.0

100

101

102

103

GRAIN SIZE (nm)

104

105

100

101

102

103

104

105

GRAIN SIZE (nm)

Figure 6.12 Summary of interfacial volume fractions (e) pin-on-disk wear rate; (f) Taber wear index; (g) coefficient of friction; and (h) electrical resistivity. (cont’d)

a significant fraction of the bulk volume (Fig. 6.12(a)) The concept of triple line softening was earlier reported by Rabukhin,[112] who investigated triple line effects on the room temperature tensile properties of conventional polycrystalline metal wires (Al, Cu, W) having various grain sizes. Triple junctions were eliminated in these wires by electrochemical thinning to a thickness less than the average grain size. In all cases an increase in strength and decrease in ductility was noted on the transition from equiaxed to bamboo grain structure. Triple line softening effects were also reported by Lehockey et al. for polycrystalline Ni.[113] The importance of triple junction contributions to the plastic deformation was further shown by the smooth transition of the decreasing

256

Processing

hardness in Ni—P electrodeposits when the grain size was gradually reduced from about 8 nm, the onset grain size of inverse Hall–Petch behavior, to less than 2 nm and toward the limit of the amorphous structure. Palumbo et al.[111] explained this in terms of the high density of disclinations, a defect that has been used to describe both triple junction structures as well as the amorphous structure.[114–116] Wang et al.[104] presented a comprehensive analysis of the deformation of nanocrystalline nickel for grain sizes between 6 and 40 nm. In addition to presenting a composite model that considered the contributions of the various microstructural defects (i.e., grain boundaries, triple junctions, and quadruple nodes), an Ashby-type deformation map was developed from the experimental data. Wang’s work[104] essentially showed that at high stress levels, grain boundary sliding is the major room-temperature deformation mechanism in nanocrystalline nickel. However, creep mechanisms resulting from the large intercrystalline volume fraction can contribute substantially at smaller grain size. It was suggested that the deviation from the Hall–Petch relationship can be attributed to a dynamic creep process due to diffusion mechanisms. Figure 6.12(d) shows that the increases in hardness and yield strength observed with decreasing grain size for nanocrystalline nickel come at the expense of tensile ductility. While the tensile elongation to fracture is still on the order of 15% for a grain size in the range of 50–100 nm,[66] ductility is reduced to less than 1% at 10 nm.[104] A slight recovery is observed at the smallest grain size of about 6 nm, which corresponds to the softening observed in the hardness (Fig. 612(b)) and yield strength (Fig. 6.12(c)) curves. Generally, greater ductility is observed in bending[66] or during cold rolling.[117] For example, nanocrystalline nickel sheet with 1.3 mm starting thickness and an as-deposited grain size of 40 nm was cold rolled without cracking to a total strain of 680%.[117] After rolling, the grains remained equiaxed with no change in grain size, while the texture changed from a strong (200) fiber texture to a more random texture. Surprisingly, there was also a small reduction in hardness after rolling. A similar observation was presented by Markham et al.[118] for inert-gas condensed Pd with grain sizes between 8 and 35 nm. Such a deformation behavior could be explained in terms of grain boundary sliding and grain rotation.[118] With the exception of the inverse Hall–Petch trend at very small grain sizes, directionally similar trends as shown in Fig. 6.12(b–d) have been reported in a number of other studies on nanocrystalline nickel electrodeposits (see, e.g., Refs. 119–123]. Dalla Torre et al.[120] carried out extensive tensile tests on commercially available nanocrystalline nickel electrodeposits with a grain size of about 21 nm at various strain rates

6: Metals, Alloys, and Composites, Erb et al.

257

between 10−5 and 103 s−1 and compared the results with coarse-grained nickel. They observed that the strain rate sensitivity in nanocrystalline nickel is one order of magnitude higher than in polycrystalline materials. A reduced work-hardening coefficient as previously reported by Wang et al.[104] was also observed. Evidence was presented of shear band formation together with local grain growth events during deformation at high strain rates. In this study it was also shown that commercially available materials can show considerable differences in their mechanical response, which can be traced back to microstructural differences for different batches of material in terms of texture, microporosity, microstrain, and impurities. A strong influence of tensile sample size on the ultimate tensile strength was also reported, with 3 mm tensile specimens yielding much higher strength values compared with 20 mm samples. Kumar et al.[124] studied the deformation and damage evolution in electrodeposited nano-Ni with an average grain size of ∼30 nm by in-situ electron microscopy and ex-situ high-resolution electron microscopy after deformation. For this grain size they showed that dislocation-mediated plasticity played a dominant role in the deformation. Void or wedge crack formation was observed along grain boundaries and triple junctions as a result of dislocation emission from grain boundaries, transgranular slip and unaccommodated grain boundary sliding. These voids can then act as nucleation sites for dimples, which lead to fracture. In these samples fracture was not observed to occur preferentially at grain boundaries. In contrast to the low elongations to fracture in tensile testing observed for nanocrystalline nickel, much greater tensile ductility was observed for other electrodeposited nanocrystals including copper,[125] cobalt,[126–128] and nickel–iron alloys.[129,130] Karimpoor et al.[126–128] reported tensile elongations in the range of 6–9%, yield strength values of 930–1000 MPa and ultimate tensile strengths between 1820 and 2130 MPa, respectively, for nanocrystalline hexagonal cobalt electrodeposits with an average grain size of 12 nm, tested at strain rates between 1 × 10−4 and 2.5 × 10−3 s−1. In comparison, polycrystalline cobalt with an average grain size of 4.8 μm tested in the same strain rate range gave the following results: elongation to fracture 6–24%; yield strength 398–446 MPa; ultimate tensile strength 821– 892 MPa. Examples of stress–strain curves for both nanocrystalline and polycrystalline cobalt are shown in Fig. 6.13. These results are quite remarkable in comparison with nanocrystalline nickel for which, at the same grain size of 12 nm, the tensile ductility is only about 1–2%.[104] The fracture surface in Fig. 6.13 shows considerable ductility, and the formation of shear bands was observed near the fracture surface of the

Processing

258 2000 Nanocrystal Cobalt

Eng. Stress (MPa)

1600

1200

800 Polycrystal Cobalt

400 strain rate = 5x10E(-4) (1/s)

0 0

4

8

12

16

20

24

Eng. Strain (%)

Figure 6.13 Engineering stress–strain curves for nanocrystalline and polycrystalline cobalt (top); side view of fractured nano-Co tensile sample (bottom right); and fracture surface of nano-Co (bottom left).[126]

nanocrystalline material. It was also observed that for nanocrystalline cobalt lower strain rates resulted in higher flow stress and tensile strength, which led the authors to conclude that mechanical twinning plays an important role in the deformation of nanocrystalline cobalt. Similarly, considerable tensile elongations were recently also observed in 0.5–3 mm thick nanocrystalline Ni—Fe deposits with Fe contents between 6 and 56 wt%.[129,130] For example, Fig. 6.14 shows the stress–strain curve of a sample containing 6.3 wt% Fe with an average grain size of 24 nm and a thickness of 0.9 mm,[130] exhibiting a tensile elongation to fracture of about 9%. Lu et al.[131] reported high ductility (>13%) at high strength (∼1 GPa) for electrodeposited copper with average grain sizes of several hundred

6: Metals, Alloys, and Composites, Erb et al.

259

2000

Eng. Stress (MPa)

1600

1200

800

400

0 0%

2%

4%

6%

8%

10%

Eng. Strain [%]

Figure 6.14 Stress–strain curve for a nanocrystalline (grain size 24 nm) Ni– 6.3 wt% Fe alloy tested at a strain rate of 5 × 10−5 s−1.[130]

nanometers. However, each grain contained a very high density of growth twins with average twin lamellae thicknesses in the range from less than 20 nm to just over 100 nm. Both the strength and the ductility were found to increase with decreasing twin lamellae thickness, with the strength following the Hall–Petch relationship when plotted as a function of the inverse square root of lamellae thickness. There is now considerable evidence that the low ductility initially found for relatively pure nanocrystalline nickel can be enhanced by incorporating second-phase particles into the nanocrystalline matrix by the particle co-deposition approach. For example, Zimmerman et al.[36] have shown that small additions (0.5–3 vol%) of submicrometer SiC particles (particle size 0.4 ± 0.2 μm) distributed uniformly throughout a nanocrystalline nickel matrix (grain size 10–15 nm) increase the ductility from 0.6% to 2.1%, as shown in Fig. 6.15. In fact, the nanocrystalline composite material shows almost the same ductility as a conventional polycrystalline Ni—SiC composite, but has significantly higher hardness, yield strength, and tensile strength at a much lower concentration of SiC particles. Chan et al.[132] have recently demonstrated superplastic behavior of Ni/Si3N4 nanocomposite electrodeposits with an as-plated Ni grain size of 50 nm and containing about 5 wt% Si3N4 whiskers with ∼100 nm diameter and ∼5 μm length. Superplastic behavior was observed at temperatures between 673 and 753 K with maximum elongations of 635% for a strain rate of 10−2 s−1. However, since considerable grain growth occurred in the Ni matrix (to about 2 μm) during the tests, it is difficult to assess the roomtemperature ductility for this material.

Processing

260 (a)

Ni (10 µm)

Ni (11 nm)

H = 1.5 GPa σy = 177 MPa σUTS = 192 MPa εfT = 50 %

(c) H = 6.2 GPa σy = 830 MPa σUTS = 1060 εfT = 0.6 %

10 µm

100 nm

(b)

(d) H = 6.0 GPa σy = 920 MPa σUTS = 1430 εfT = 2.1 %

H = 3.2 GPa σy = 570 MPa σUTS = 670 MPa εfT = 3 %

10 µm

100 nm

Ni + 7% SiC

Ni + 1.8% SiC

Figure 6.15 Comparison of hardness (H), yield strength (σy), ultimate tensile strength (σUTS), and elongation to fracture in tensile testing of conventional (a, b) and nanocrystalline (c, d) nickel and Ni—SiC composite materials.

The work to date on the mechanical properties of electrodeposited nanocrystalline materials has provided some answers to the unusual properties exhibited by these materials. However, the understanding of deformation mechanisms, damage accumulation and fracture response is still far from complete. For the time being, the results of many studies remain inconclusive, including the strain rate sensitivity of various nanocrystalline electrodeposits (see, e.g., Refs. 120, 128, 132, 133), as well as the superplastic deformation, up to almost 900% at 0.4Tm, observed for pure nanocrystalline nickel (starting grain size 35 nm) during high-temperature deformation.[134] In the latter study, simultaneous grain growth during the deformation process complicates the interpretation of grain size and interface effects. Clearly, this general article on electrodeposited nanomaterials cannot review all details of mechanical property measurements that have been reported so far on this group of materials, in particular over the past five years. For a more in-depth discussion of the mechanical behavior of nanomaterials produced by electrodeposition and other methods, the reader is referred to the excellent recent review article by Kumar et al.[121] It can be expected that answers to many of the questions regarding the mechanical properties of nanocrystalline materials can be obtained from future in-situ deformation studies and computer modeling work such as presented in the

6: Metals, Alloys, and Composites, Erb et al.

261

review by Kumar et al.[121] Extensive computer modeling has already revealed several new inter-grain and intra-grain deformation mechanisms including the emission from grain boundaries of full or partial dislocations or deformation twins as summarized, for example, in Ref. 135.

6.4.2.2 Fatigue Response In conjunction with the development of the electrosleeve technology, extensive fatigue testing was performed on nanocrystalline nickel with grain sizes in the 50–100 nm range.[66] Tests were performed in fully reversed bending (R = −1) at frequencies in the range 0.5–25 Hz. A comparison of S–N curves for nanocrystalline, fully annealed, and cold-drawn conventional nickel showed that, at room temperature, the fatigue performance of nanocrystalline nickel was basically the same as for conventional nickel. Furthermore, the fatigue performance was not found to be compromised at 300°C. More recently, Hanlon et al.[136] studied the fatigue response for nanocrystalline nickel electrodeposits with a grain size in the 20–40 nm range. This material was subjected to zero-tension-zero (R = 0)-loading at a frequency of 1.0 Hz with a sinusoidal waveform. The S–N curve for the nanocrystalline Ni shown in this study is very similar to the one shown in Ref. 66. Comparison curves were presented for ultrafine-grained (grain size 300 nm) and equiaxed polycrystalline (grain size 10 μm) samples, which clearly showed that nanocrystalline nickel has a higher endurance limit than ultrafine or polycrystalline nickel. It was pointed out, however, that grain refinement can have a deleterious effect on the resistance to subcritical fatigue fracture.

6.4.2.3 Precipitation and Solid Solution Hardening Nanocrystalline alloys can be electrodeposited as supersaturated solid solutions (section 6.3). Upon heat treating, these alloys transform by precipitation of second-phase particles in addition to grain growth. By choosing appropriate heat treatments, the mechanical properties can be enhanced considerably by controlling the grain size as well as the second-phase volume fraction and particle size. In fact, such heat treatments can produce the in-situ composite materials as shown schematically in Fig. 6.9(b). Examples of the effect of heat treatment on the hardness of Co—P alloys[137] and Ni–1 wt% P alloys[63] are shown in Fig. 6.16. For example,

Processing

262 14

Temperature: 400 ºC

12

Ni - 1 wt.% P

12

10 8 6 4 Co - 5.2 % P Co - 3.2 % P Co - 0.8 % P Nano Colbalt

2 0 0

5

10 15 20 25 30 35 Annealing Time [min]

Vickers Hardness [GPa]

Vickers Hardness [GPa]

14

10

400 ºC

8 500 ºC

6 4

600 ºC

2 0 0

20 40 60 80 100 120 140 Annealing Time [min]

Figure 6.16 Precipitation hardening in nanocrystalline Co—P and Ni—P alloy deposits.

the hardness of the Ni—P alloys annealed at 400°C for 15 minutes is almost twice the hardness of the as-plated starting material with a grain size of 10 nm. This is the result of the formation of Ni3P precipitates in the Ni matrix. Longer annealing times or higher annealing temperatures have the effect of overaging or immediate hardness reduction resulting from excessive grain growth and second-phase coarsening. For pure cobalt, Fig. 6.16 shows that annealing at 400°C results in a substantial decrease in hardness after about 5 minutes, which is due to extensive grain growth. However, upon alloying cobalt with P, Co3P second-phase formation impedes grain growth and increases the hardness as for the case of the Ni—P alloy. Overaging effects are also observed for the Co—P alloys. Solid-solution effects have been studied in detail for nanomaterials produced by mechanical attrition.[138] For example, alloys such as Ti (Cu), Nb (Cu), Cu (Ni), Cu (Fe) and Ni (Co) exhibited solution hardening effects, while Cr (Cu) Fe (Cu) and Ni (Cu) showed solution softening. To the best of the authors’ knowledge, to date there has been no comparable systematic study on solution hardening/softening effects for nanocrystalline materials produced by electrodeposition. From the extensive list of alloys studied in the authors’ research group, the only systems that showed clear evidence of solution hardening in addition to grain size hardening were ternary Ni—Fe—Co and Co—Fe—P alloys.[38] Another study explicitly stated that solution hardening was negligible in the Ni—W alloy system.[139] The area of solid-solution effects in electrodeposited nanomaterials clearly needs more attention in future studies.

6: Metals, Alloys, and Composites, Erb et al.

263

6.4.2.4 Wear Properties Numerous reports have addressed the wear properties of nanocrystalline electrodeposits.[103,129,137,140–147] These studies used either the pinon-disk, Taber wear, or the scratch tests to assess the resistance of as-deposited or annealed nanomaterials. One of the earliest studies[103] investigated the tribological properties of pure nanocrystalline Ni as a function of grain size using the pin-on-disk method. The results from this work are shown in Figs. 6.12(e) and 6.12(g), respectively. By reducing the grain size in Ni from 10 μm to about 10 nm, the wear rate was reduced by two orders of magnitude while the coefficient of friction dropped by a factor of 2. While the pin-on-disk test gives a good measure of adhesive wear properties, the Taber wear test is designed for the much harsher conditions of abrasive wear, usually expressed as the Taber wear index which is the weight loss in milligrams per 1000 cycles of rotational abrasion against a test rubber wheel with embedded hard Al2O3 particles. Figure 6.12(f) shows that even under such severe conditions nanocrystalline nickel outperforms polycrystalline nickel by a factor of 2.[140] The higher Taber wear resistance of nanocrystalline nickel is clearly visible on comparing the worn surfaces of polycrystalline (grain size 60 μm) and nanocrystalline (grain size 13 nm) nickel as shown in Fig. 6.17; the wear grooves in polycrystalline nickel are much deeper and wider than for nanocrystalline

10μm

Polycrystalline

10μm

Nanocrystalline

Figure 6.17 Comparison of nanocrystalline (grain size 13 nm) and polycrystalline (grain size 60 μm) nickel surfaces after Taber wear testing.[54]

264

Processing

nickel. The study by Jeong et al.[140] showed clearly that for pure nickel electrodeposits, the Taber wear resistance follows the well-established Archard law,[148] which states that the wear resistance is directly proportional to the hardness of the material. Upon alloying Ni with phosphorus, Jeong et al.[142] reduced the grain size of the electrodeposits down to less than 10 nm, thereby achieving hardnesses values that fall into the range of the inverse Hall–Petch behavior. As expected from Archard’s law, the Taber wear index went through a minimum that corresponded to the maximum in the hardness as a function of grain size curve. In other words, for as-deposited Ni and Ni—P electrodeposits there is a one-to-one relationship between grain size, hardness, and Taber wear index. However, such a direct relationship was not observed for cobalt electrodeposits.[147] Instead, for cobalt, ductility was found to play an important role in the abrasive wear resistance, with lower ductility giving better abrasive wear resistance. Therefore, when comparing nanocrystalline nickel and nanocrystalline cobalt with about the same grain size and hardness, nickel has a much higher abrasive wear resistance than Co because of its lower ductility (section 6.4.2.1). In a study of heat-treated Ni—P alloys containing Ni3P precipitates, it was further shown that there is no simple relationship between hardness and abrasive wear resistance.[146] For these materials the highest hardness did not correspond to the highest wear resistance. As for nanocrystalline cobalt electrodeposits, a better correlation was found between ductility and Taber wear resistance. From these studies it can be concluded that the high hardness of nanocrystalline materials is a prerequisite of high abrasive wear resistance. However, for equal hardness/grain size, the other important factor is ductility, with lower ductility giving better abrasive wear resistance. This phenomenon is not understood to date and requires further study. The highest abrasive wear resistance values found in the literature to date for electrodeposited nanomaterials are for iron, cobalt–iron and nickel–iron–phosphorus alloys.[137,141] Their Taber wear resistance is very close to the wear resistance of hard chromium, one of the best industrial coatings that has been used for many decades for its outstanding wear resistance. In terms of pin-on-disk wear performance, it has been shown that nanocrystalline iron and cobalt–iron–phosphorus alloys have even higher wear resistance than hard chromium.[138] Schuh et al.[144] also observed a good correlation between hardness and wear resistance in a scratch test performed using a nano-indenter on

6: Metals, Alloys, and Composites, Erb et al.

265

nanocrystalline Ni with grain sizes ranging from 22 to 12 nm. In this grain size range they noticed a breakdown in the Hall–Petch relationship which was clearly reflected in the scratch wear resistance.

6.4.2.5 Corrosion Properties In conventional polycrystalline metals, grain boundaries are often prone to intergranular corrosion and stress corrosion cracking because of (i) their enhanced energy relative to the grain interior and (ii) differences in chemical composition resulting from solute segregation and second-phase precipitation (see, e.g., Ref. 149). When nanocrystalline materials were first introduced about 20 years ago, the initial expectation was that these structures would exhibit very poor corrosion properties because of their large volume fractions of interfaces. However, detailed studies over the past 15 years have shown that, for a wide range of electrochemical conditions, grain size reduction in nanocrystalline electrodeposits can actually have a beneficial effect by providing a microstructure that is highly resistant to localized attack.[9,65,66,150–165] Rofagha et al.[9] investigated the corrosion behavior of nanocrystalline nickel electrodeposits in de-aerated 2N H2SO4 (pH 0) solution using potentiodynamic and potentiostatic polarization tests. The results of this study can be summarized as follows. Ultrafine-grained and nanocrystalline nickel electrodeposits (grain sizes of 500, 50, and 32 nm, respectively) exhibited active, passive and transpassive behavior very similar to what was observed for conventional polycrystalline Ni (100 μm grain size). However, the passive current densities of the ultrafine-grained and nanocrystalline materials were about one order of magnitude higher than for polycrystalline nickel. Using X-ray photoelectron spectroscopy[152] it was shown that the higher passive current density was due to a more defective passive film formed on the highly disordered surfaces of the nanocrystalline nickel. Another notable observation was a positive shift in the corrosion potential for the nanocrystalline samples, which was explained in terms of the catalysis of the hydrogen evolution reaction. Despite the somewhat enhanced overall corrosion rate, it was clearly shown in this study that ultrafine and nanocrystalline nickel developed a much more uniform corrosion morphology than the polycrystalline counterpart. Very similar results were obtained when nanocrystalline nickel was corroded in 30 wt% KOH (pH 13) and 3 wt% sodium chloride (neutral pH) solutions.[153] In an attempt to study in detail the localized corrosion resistance of nanostructured Ni electrodeposits, Kim[158] and Kim et al.[155] investigated

266

Processing

the effects of grain size and solute segregation in as-plated and annealed nickel containing about 1000 wt-ppm sulfur in a 0.25 M Na2SO4 solution (pH 6.5). The potentiodynamic polarization curves of nickel containing sulfur showed a similar electrochemical behavior regardless of microstructure (i.e., polycrystalline or nanocrystalline). Surface and cross-sectional examination of the corrosion morphology revealed that nanocrystalline deposits (grain size 20–30 nm) developed a corrosion morphology exhibiting a very high density of shallow corrosion pits, as shown in Fig. 6.18. In contrast, the annealed polycrystalline counterpart (grain size 50–100 μm) showed extensive localized corrosion attack along grain boundaries and triple junctions, also shown in Fig. 6.18. The tremendous increase in the localized corrosion resistance observed for the nanocrystalline nickel was explained in terms of “solute dilution by grain size refinement,” a mechanism originally proposed by Palumbo and Erb.[166] The good corrosion performance of nanocrystalline nickel was also observed when the material was exposed to a salt spray environment as per the ASTM B-117 test.[162] In this study, both polycrystalline and nanocrystalline electrodeposits on mild steel substrates provided the same corrosion protection.

Figure 6.18 Scanning electron micrographs showing surface (top) and crosssectional (bottom) corrosion morphologies of nickel containing 0.1 wt% sulfur: polycrystalline structure (left) and nanocrystalline structure (right).[158]

6: Metals, Alloys, and Composites, Erb et al.

267

Tang et al.[154] compared the corrosion performance of nanocrystalline nickel (grain sizes 5–10 nm), prepared by direct-current plating, pulsecurrent plating, and pulse reverse plating, in immersion tests in 7 M nitric acid, 3 M hydrochloric acid, and 20 g/liter citric acid solutions, as well as exposure to moist SO2 environment. Considerable differences in the corrosion resistance were found among the differently prepared samples, which were explained in terms of distinct crystallographic texture developments in these materials. The corrosion behavior of nanocrystalline cobalt (grain size 13 nm) was studied in two environments: 0.25 M NaSO4 (pH 7) and 0.1 M NaOH (pH 13).[156,157] For the 0.25 M Na2SO4 solution, in which cobalt does not passivate, the overall shape of the potentiodynamic polarization curves for both the nanocrystalline and polycrystalline (grain size ∼10 μm) materials were almost identical and the metal dissolution in the absence of a passive film was only slightly enhanced for nanocrystalline cobalt. However, as for the case of nickel, nanocrystalline cobalt showed a higher resistance to localized corrosion. Aledresse and Alfantazi[165] reported very similar results in a study comparing nanocrystalline Co (50 and 67 nm) with polycrystalline Co (grain size 100 μm). In the 0.1 M NaOH solution both polycrystalline and nanocrystalline cobalt readily formed passive films and the passivation characteristics were not affected by reducing the grain size from 10 μm to 13 nm.[157] The electrosleeve material developed for in-situ nuclear repair technology[65,66] used a nickel–phosphorus microalloy in which ∼3000 ppm by weight of phosphorus was added for enhanced thermal stability of the 50–100 nm grain size structure. Various corrosion tests (ASTM G28susceptibility to intergranular attack; G35, G36, and G44 susceptibility to stress corrosion cracking) have shown that the material is intrinsically resistant to intergranular attack and intergranular stress corrosion cracking. The material was also found to be resistant to pitting attack and only slightly susceptible to crevice corrosion. In another study, it was found that the corrosion behavior of nanocrystalline Ni—P alloys with higher phosphorus contents (1.4 wt% and 1.9 wt%) in 0.1 M H2SO4 (pH 0) approached that of amorphous Ni–6.2 wt% P electrodeposits.[151] X-ray photoelectron spectroscopy revealed that neither the nanocrystalline (1.9 wt% P, grain size 8.4 nm) nor the amorphous (6.2 wt% P) Ni—P alloys formed a passive layer in this environment.[163] Youssef et al.[164] compared the corrosion performance of nanocrystalline zinc electrodeposits (grain size 60 nm) and conventional electrogalvanized steel by potentiodynamic polarization and impedence measurements in 0.5N NaOH solution. It was observed that the corrosion

Processing

268

rate of nanocrystalline zinc was about 60% lower than for the electrogalvanized steel, which was attributed to the passive film that formed on nanocrystalline zinc. A study of ternary Co65Ni12Fe23 electrodeposits (grain sizes 10–40 nm) in de-aerated 2.5 wt% NaCl also showed that the good corrosion resistance of Co—Ni—Fe alloys is not compromised by grain size reduction to the nanocrystalline state.[159] A comparative wear/corrosion study[160] of pure polycrystalline nickel (grain size >1 μm) and Ni—SiC nanocomposite (Ni grain size ∼100 μm; SiC particles <100 nm) using sliding type wear testing and electrochemical impedance spectroscopy in 0.5 M Na2SO4 (neutral pH) showed a 50% reduction in the corrosion rate for the nanocomposite coating compared to the polycrystalline deposit already in the absence of wear action. Even higher differences in the degradation rates between the two samples were found when the materials were subjected to wear-corrosion conditions, e.g., a 90% reduction at a normal load of 30 N. An ongoing current study[161] is concerned with the effect of grain size reduction on the corrosion performance of copper. Results to date have shown that there are no major differences between the polarization curves for both polycrystalline (grain size 3 μm) and nanocrystalline (grain size 70 nm) copper in a 0.1 M NaCl solution (Fig. 6.19). As for nickel and cobalt, nanocrystalline copper deposits exhibited a very uniform corrosion morphology.

2.0 1.5

70 nm

Potential (VSCE)

1.0

3 µm

0.5 0.0 -0.5 -1.0 -1.5 10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Current Density (Amps/cm2)

Figure 6.19 Potentiodynamic polarization curves for nanocrystalline and polycrystalline Cu in 0.1 M NaCl at room temperature.[161]

6: Metals, Alloys, and Composites, Erb et al.

269

6.4.2.6 Electrical Resistivity Grain boundaries have long been known to affect the electrical properties of metals.[167] Conventional polycrystalline metals (grain size >1 μm) contain a relatively small volume fraction of grain boundaries and therefore their effect on the electrical resistivity is only significant at low temperatures.[168–171] It is worth noting that not all grain boundaries affect resistivity in the same way. Considerable differences in the electrical resistivity were observed for structurally different grain boundaries (i.e., lowangle, general high-angle, low-Σ and coherent twin boundaries) in zone-refined aluminum.[171] In nanocrystalline materials, on the other hand, the density of grain boundaries can reach considerable values, such that their influence on the resistivity can be measured even at room temperature. A comparison of results of electrical resistivity measurements performed on materials produced by different synthesis routes (e.g., inert-gas condensation,[67] electrodeposition[12,13,106,107,129,172]) show very similar trends, i.e., increasing resistivity with decreasing grain size, such as shown for nickel in Fig. 6.12(h). This can be attributed to electron scattering effects at grain boundaries and triple junctions. In fact, a linear relationship between excess resistivity–defined as the total resistivity of the nanocrystalline nickel minus the resistivity of conventional polycrystalline nickel with negligible intercrystalline volume fraction–was observed for nickel electrodeposits of varying grain sizes.[106] There is also good agreement in terms of the temperature coefficient of resistivity for materials produced by inert-gas condensation[67] and electrodeposited Ni[106] and Co.[12] In all cases decreasing temperature coefficients with decreasing grain sizes were observed. For electrodeposited Cu, Bakonyi et al.[12] found no effect of grain size on the electrical transport properties, which they attributed to the negligible effect of structural disorder on the density of states around the Fermi level of copper. More recent resistivity measurements on copper by Woo[172] are not in agreement with this observation. A decrease in the grain size of copper from 3.2 μm to 73 nm resulted in a room temperature resistivity increase by 26%. McCrea[107] and McCrea et al.[129] expressed the electrical resistivity of nanocrystalline metals (Ni, Co, and Ni—Fe alloys) in terms of a specific grain boundary resistivity in order to be able to compare results from measurements on nanocrystalline and polycrystalline metals. The specific grain boundary resistivity values were 2.82 × 10−6 μΩ cm2 for nanocrystalline Ni and 3.26 × 10−6 μΩ cm2 for nanocrystalline Co. The values for

270

Processing

nanocrystalline Ni—Fe alloys were 3.01, 3.09, and 2.99 × 10−6 μΩ cm2 for iron contents of 16 wt%, 31 wt%, and 34 wt%, respectively. In comparison, the specific grain boundary resistivity values measured on polycrystalline Cu and Al fall in the range of 1.35 × 10−6 μΩ cm2 and 4.8 × 10−6 μΩ cm2.[173] This observation has led McCrea[107] to conclude that, as far as electron scattering is concerned, grain boundary structures in polycrystalline and nanocrystalline materials are quite similar. Similar results were recently reported for nanocrystalline Cu (grain size 19–350 nm) prepared by magnetron sputtering.[174] The specific grain boundary resistivity value of 2.04 × 10−6 μΩ cm2 reported in this study is close to the very low value of 1.5 × 10−6 μΩ cm2 reported earlier for the specific case of twin boundaries in Cu.[170] McCrea[107] and McCrea et al.[173] have further shown that the changes in electrical resistivity with increasing temperature can be used to monitor the thermal stability of nanocrystalline metal electrodeposits during annealing. By monitoring resistivity as a function of time, McCrea[107] was able to determine activation energies for the growth process in nanocrystalline Ni, Co, and Ni–30 wt% Fe electrodeposits.

6.4.2.7 Thermal Stability As a result of the large interfacial area in nanocrystalline materials there is a strong driving force for grain growth, leading to structural instability with increasing temperature. The fundamental issues that must be addressed in thermal stability studies are the driving forces for grain growth and other phase transformations (e.g., allotropic phase transformations, second-phase precipitation, crystallization from precursor amorphous phases) and dragging forces such as solute drag, particle drag, vacancy drag, and porosity drag. These can vary considerably from system to system and are strongly dependent upon the synthesis method used to prepare the nanostructured material. As a consequence, it is very difficult to compare thermal stability studies performed on materials made by different processing routes. Therefore, this section considers only thermal stability and grain growth studies on fully dense electrodeposited metals and alloys that have no second-phase particles or residual porosity in the starting microstructure. There are numerous reports of studies of the thermal stability and grain growth in electrodeposited nanocrystalline Ni,[60,107,175–186] Ni—P alloys,[52,55,65,66,187–190] Ni—Fe alloys,[60,107,191,192] Pd—Fe alloys,[25,193] Co,[185,194,195] Co—P alloys,[196] Ni—Co alloys,[197] and Ni—W alloys.[198]

6: Metals, Alloys, and Composites, Erb et al.

271

Recent reviews of the thermal stability of nanocrystalline electrodeposits are given in Refs. 199, 200, and 201. Many studies investigated the thermal stability by calorimetric methods such as differential scanning calorimetry (DSC) or modulated differential scanning calorimetry (MDSC). In these measurements, the enthalpy release during phase transformations is recorded as a function of temperature. A typical DSC curve for nanocrystalline Ni—P, recorded at a scan rate of 5°C/min, is shown in Fig. 6.20.[52] The curve shows a broad low energy exotherm followed by a main heat release peak. This type of curve has been observed in several studies for Ni,[60,177–179,182,185] Ni—P,[52,190] Ni—Fe,[60] Co,[185,194] and Co—P[196] and contains a wealth of information. First, the broad low-energy exotherm indicates microstructural transformation events before the onset of major grain growth in the structure. For example, Klement et al.[177,178] have attributed this exotherm to the formation of nuclei for subsequent growth by grain boundary relaxation and subgrain coalescence. Second, in relatively pure materials the main peak corresponds to the major grain growth event. As a result, the maximum in this curve, referred to as the peak temperature, Tp, can be used as a relative measure of the thermal stability of the nanostructure. Third, the total enthalpy release (area under the curve) can be used to derive the interfacial enthalpy if the grain sizes and shapes of the initial nanostructure and the structure after grain growth are known, for example, from electronmicroscopic observations. Of course any other phase transformations

0.10 0

0

420 to 467

DSC @ 5 to 80 C/min

0

Heat Release (W/g)

0.08

0.06

0.04

d = 6.9 nm

0.02

0.00

TP 0

100

200

300

400

500

600

0

Temperature ( C)

Figure 6.20 DSC curve for a nanocrystalline (grain size 6.9 nm) Ni–2.5 wt% P alloy at a scanning rate of 5°C/min. The arrow indicates the shift in peak temperature, Tp, upon increasing the scanning rate to 80°C/min.[52]

Processing

272

(e.g., precipitation, allotropic transformation) occurring simultaneously with grain growth make interfacial enthalpy determinations more difficult. When DSC scans are carried out at different scanning rates (e.g., from 5°C/min to 80°C/min), the peak shift observed in the heat flow versus temperature curves can be used to determine the activation energy for the process responsible for the heat release as per modified Kissinger analysis.[202] Table 6.2 summarizes the peak temperatures measured by calorimetry at a scan rate of 5°C/min (10°C/min for the case of Ni–1.2 wt% P) in various studies for Ni, Ni—P, Ni—Fe, and Co electrodeposits with starting grain sizes ranging from 7 nm to 26 nm. Also shown in this table are the activation energies, Q, for grain growth as determined by the Kissinger-type analysis using DSC scanning rates between 5°C/min and 100°C/min. As can be seen from these values the major grain growth event for Ni occurs between 266 and 296°C, with activation energies ranging from 1.20 eV to 1.46 eV, respectively. The activation energies for lattice self-diffusion and grain boundary diffusion in polycrystalline nickel are 2.9 eV and 1.2 eV, respectively.[203] This has led several authors to the conclusion that grain growth in nanocrystalline nickel is controlled by grain boundary diffusion.[60,179,181] The differences for the peak temperatures and activation energies reported in the different studies are likely due to dif-

Table 6.2 Summary of Calorimetric Studies of Nanostructured Electrodeposits (Note: all Tp values listed were obtained at 5°C/min with the exception of Ni–1.2 wt% P, which was obtained at 10°C/min)

System Ni Ni Ni Ni Ni Ni–1.2 wt% P Ni–1.9 wt% P Ni–2.5 wt% P Ni–20 wt% Fe Co

Grain size (nm)

Tp (°C)

Q (eV)

Reference

20 26 20 15 20 10 9 7 13 20

290 266 269 293 296 432 412 420 379 355

1.36 1.20 1.22 1.42 1.46 2.25 2.63 2.58 2.53 1.63

[179] [60] [60] [60] [182] [55] [190] [52] [60] [194]

6: Metals, Alloys, and Composites, Erb et al.

273

ferences in the starting materials in terms of grain size and, more importantly, impurity content. Alloying Ni with 1.2–2.5 wt% P has the effect of substantially increasing the peak temperatures and activation energies to over 400°C and 2 eV, respectively. This can be understood in terms of P-solute drag by P segregation to the grain boundaries and Zener drag by Ni3P particles, both of which occur simultaneously with grain growth (see, e.g., Refs. 52, 55). In addition, there could already be substantial P segregation to grain boundaries in the as-deposited state, as shown by Hentschel et al.[190] for electrodeposited nano-nickel containing 1.9 wt% P. The strong influence of phosphorus in stabilizing the nanostructure of Ni was effectively used in the electrosleeve technology, which required a long-term (∼20 years) thermal stability at steam generator operating temperatures of 280–320°C. By alloying the nickel electrosleeve (grain size 50–100 nm) with about 3000 wt-ppm phosphorus, a peak temperature of over 500°C and an activation energy on the order of 2.5 eV were achieved.[204] In fact, long-term indirect monitoring of grain growth by hardness measurements at temperatures higher than the operating temperature has shown the electrosleeve material to be thermally stable at 340°C for in excess of 106 minutes.[66] Table 6.2 further shows that the thermal stability of nanocrystalline nickel can also be increased by alloying with 20 wt% Fe. The stabilizing mechanism in this system is currently not understood. One possibility is the ordering transformation to Ni3Fe which, however, is difficult to observe by conventional X-ray or electron diffraction due to the closeness of the atomic scattering factors for iron and nickel. Mössbauer spectroscopy studies on as-deposited and annealed Ni—Fe alloys are currently in progress in the authors’ laboratory to investigate this possibility. For nanocrystalline hexagonal cobalt, the thermal stability is also enhanced compared with nanocrystalline nickel (Table 6.2). In cobalt, the stability issue is more complex because of the allotropic hcp to fcc transformation occurring simultaneously with grain growth at temperatures well below (∼120°C) the equilibrium transformation temperature.[194,195] When comparing the microstructures of annealed nanocrystalline nickel electrodeposits shown in the various studies there is considerable confusion with respect to grain growth mechanisms. Both normal and abnormal growth have been reported depending on the thermal history. This is demonstrated in Fig. 6.21, which shows the microstructures of nanocrystalline nickel (average grain size 20 nm) in the as-plated state (Fig. 6.21(a)) and after various heat treatments (b–d). It is clear that different conclusions can be drawn from such micrographs regarding growth

Processing

274

(a)

(b) 300 nm

(c)

(d)

Figure 6.21 Microstructures of (a) as-deposited (grain size 20 nm) and (b–d) annealed nickel electrodeposits. (b) 1 min at 300°C; (c) 10 min at 300°C; (d) 30 min at 300°C.

mechanisms depending on the grain growth state after which these were taken. As a result several growth mechanisms have been reported in the past including (i) uniform growth followed by stabilization, (ii) abnormal growth at lower temperatures and normal growth at higher temperatures, (iii) a transition from abnormal growth to normal growth with increasing DSC scanning temperature, (iv) transition from normal to abnormal growth with increasing annealing time at constant temperature, and (v) concurrent abnormal and normal growth. Hibbard[185] and Hibbard et al.[199–201] have recently shown that, with the exception of the first of the proposed mechanisms, all of the other reported grain growth events for nanocrystalline nickel can in fact be explained when the microstructural evolution upon annealing is described as a sequential multistage transformation. It is important to note that this transformation sequence is directly linked with sulfur impurities (e.g.,

6: Metals, Alloys, and Composites, Erb et al.

275

100–1000 wt-ppm) that were present in the deposits as a result of using saccharin in the electroplating baths during the synthesis of these materials.[48] The multistage process proposed by Hibbard[185,199] can be summarized as follows. During the first stage of growth a small portion of the nanocrystalline material is consumed by a relatively small fraction of grains growing abnormally into the nanocrystalline matrix. There are a number of reasons why some grains can grow abnormally, including a size advantage they have either in the as-deposited state or after a subgrain coalescence process, or higher mobilities of some grain boundaries resulting from special misorientations. While the details of the early growth selection are not yet completely understood, it is important to note that during this first stage of growth the majority of grains in the matrix retained the starting grain size. Once the initial stage of grain growth is completed, a heavily sulfursegregated structure remains that still has a substantial driving force for grain growth. Continued growth now proceeds more uniformly, and at a significantly reduced rate compared with the initial stage, until the nanocrystalline material is completely consumed. At this stage the structure looks like it has gone through a normal grain growth process. This is followed by another growth stage, referred to as the late stage abnormal grain growth, which is characterized by the migration of planar interfaces into a matrix that now consists of sub-micron-sized grains.[185,199] Energy-dispersive X-ray spectroscopy and scanning transmission electron microscopy showed sulfur-rich second-phase precipitates that appeared to wet these interfaces.[186] Initiation of this second stage of abnormal growth is likely controlled by sulfur redistribution as the system transforms toward the thermodynamic equilibrium with respect to nickel sulfide precipitation and grain boundary structure-dependent segregation. Hibbard further showed[185] that sulfur impurities are also the most important factor in the microstructural evolution of nanocrystalline cobalt electrodeposits with relatively constant starting grain size (∼13 nm) but varying sulfur content (240–610 wt-ppm). In contrast, nanocrystalline nickel deposits that contained phosphorus in concentrations of 1.2, 1.9 or 2.5 wt% have not shown this multistage transformation involving transitions between abnormal and normal grain growth. Instead these materials always exhibited normal grain growth[52,55,187] as shown, for example, in Fig. 6.22 for a nanocrystalline (grain size 6.9 nm) Ni—P alloy containing 2.5 wt% P. At this time it is not clear whether the differences observed in sulfuror phosphorus-containing nickel are the result of chemical effects (sulfur

Processing

276

100 nm

100 nm

(a)

(b)

Figure 6.22 TEM bright-field images of (a) as-plated and (b) DSC annealed (5°C/min to 400°C) Ni–2.5 wt% P with a starting grain size of 6.9 nm.[52]

versus phosphorus) or concentration dependence (hundreds of ppm versus thousands of ppm). Studies are currently under way to follow the microstructural evolution in high-purity nanocrystalline nickel without sulfur or phosphorus solute additions.[205]

6.5 Applications As pointed out in the introduction, electrodeposition has a very long history. Initially this process was used to modify surfaces of finished or semi-finished metallic products with a surface coating (typically less than 100 μm in thickness) to improve the appearance or properties such as corrosion and wear resistance. Some of the most successful coatings in industrial applications are nickel, zinc, and cadmium for corrosion protection; hard chromium for its outstanding wear resistance; copper, gold, and palladium for electronic applications; and nickel-iron or cobalt-based electrodeposits for their excellent magnetic properties.[1] Given the large variety of product shapes and sizes, electrodeposition operations range from large-scale plants for mass finishing of inexpensive parts, to clean room facilities for the processing of high-value-added products, following very strict process steps in terms of cleanliness and deposit thickness, composition, and internal stress control. Many processes have been developed to produce thick coatings (>100 μm), electroformed components (e.g., >1 mm thickness), free-standing products such as sheet, foil, tubes, wires, foam, powders, or components for microsystem applications. Industrial installations exist for batch processing (e.g., rack plating, barrel plating), continuous production (e.g., reel-to-reel plating, continuous foil

6: Metals, Alloys, and Composites, Erb et al.

277

plating), or repair and refurbishing operations (e.g., brush plating). Electrodeposition can be used as a non-line-of-sight process to deposit material in recesses, holes and cavities or even the inside of tubes, which is a considerable advantage over other deposition processes (e.g., vapor deposition or plasma spray technologies). The electrodeposition of nanocrystalline metals, alloys, and composites requires the same basic equipment and electrolyte ingredients as conventional electroplating. The single-step formation of the nanostructure is controlled by the bath chemistry and operating parameters such as current density, stir rate, pH, or temperature. While some modifications may be required to the power supply to allow for pulse current capability, nanoelectrodeposition can basically make use of existing infrastructure in many electroplating facilities. Therefore, this technology can be considered a drop-in technology for many different applications requiring only modest capital investment. Figure 6.23 shows examples of product shapes that can be made as nanocrystalline electrodeposits. As a result of (i) the mature status of established industries utilizing electrodeposition, (ii) the relatively low cost of the process, and, most importantly (iii) the capability to produce fully dense nanostructures free of extraneous porosity in many different shapes and forms, electrodeposition of nanomaterials has advanced rapidly to commercial application. From the outset, the fully dense nanomaterials produced by electrodeposition have displayed properties that, to a large extent, were predictable on the

Nanoplate technology can be applied as a coating or, via conventional net-shape electroforming processes, can be used to cost-effectively produce micro- or macro-scale complex components.

Figure 6.23 Various product forms of nanocrystalline materials made by electrodeposition. (Courtesy of Integran Technologies Inc., Toronto, ON, Canada.)

278

Processing

basis of their increased content of intercrystalline defects (i.e., grain boundaries and triple junctions) such as summarized in Figs. 6.10 and 6.12. In these materials, artifacts arising from residual porosity are virtually eliminated, making grain size control, and therefore the high-angle grain boundary content, the dominant parameter for property optimization. The latter point can be demonstrated by looking at the electrosleeve technology. For this application there were certain minimum property requirements for the sleeve material: (i) strength levels in excess of 600 MPa; (ii) minimum tensile ductility of 10%; (iii) high Young’s modulus; (iv) thermal expansion close to the tube material: (v) ferromagnetic properties for nondestructive testing capability; (vi) excellent corrosion resistance; and (vii) long-term (>10 years) thermal stability at temperatures in the range 280–320°C. While strength optimization alone would have suggested a grain size on the order of ∼10 nm (i.e., the maximum in Fig. 6.12(c)), the ductility and thermal stability requirements dictated a final grain size in the range of 50–100 nm, as well as the addition of about 3000 ppm of phosphorus for enhanced thermal stability. We have reported on a wide range of applications of electrodeposited nanomaterials on numerous occasions. Some of the most important applications are summarized in Table 6.3. From the properties of nanocrystalline metals, alloys, and composites discussed in section 6.4, many of the applications listed in Table 6.3 are obvious and will not be further discussed here in view of their extensive coverage elsewhere. Less obvious are the applications of nanomaterials in printed wiring board and microelectromechanical systems. These two applications will be discussed in more detail in the following sections. Both are currently under development in our laboratories.

6.5.1 Nanocrystalline Copper for Printed Wiring Boards Electroformed copper foil (thickness 10–35 μm) is extensively used as the conductor in the manufacture of printed wiring boards. The continuing trends toward miniaturization of electronic devices have created considerable challenges to produce copper foils that can meet the reliability requirements of the fully assembled printed wiring board. The three major challenges are (i) to provide sufficient strength to prevent failure of the foil during thermal cycling; (ii) to prevent or minimize undercut of the copper foil during the etching process; and (iii) to reduce the trace width for higher wiring density capability.

6: Metals, Alloys, and Composites, Erb et al.

279

Table 6.3 Applications of Electrodeposited Nanocrystals

Application

Materials

Armor laminates

Ni, Fe,Co, Ni—Fe

Battery grids Catalysts for H2 evolution Corrosion-resistant coatings

Pb Ni—Mo Ni, Ni—P, Zn—Ni, Co, Co—P Co—P Ni—Fe Ni Ni—P Cu Ni, Co, Ni—Fe, Co—Fe Ni, Ni—SiC, Ni—Al2O3, Co, Co—P Ni—Fe Ni, Co, Ni—Fe Ni—MoS2, Ni—BN Cu Ni, Co, Ni—Fe Ni—Fe, Co—Fe Ni, Ni—SiC, Ni—P, Co, Co—P

Chromium replacement coatings Electromagnetic shielding Electronic connectors Electrosleeve Foil for printed circuit boards Free-standing soft magnets Hard facing applications Magnetic recording heads Microelectromechanical systems Self-lubricating coatings Shaped charge liners Structural applications Tranformer core materials Wear-resistant coatings

References [38] [206] [207] [38, 207, 208, 209, 210] [38, 206, 211, 212] [129] [207] [65, 66, 207] [38, 206] [38, 206] [38, 206] [207] [213, 214, 215] [38, 206, 213] [206] [38, 206] [38, 129] [38, 141, 147, 206, 211, 212]

The first challenge is related to the strength of copper. Failure of the copper foil in printed wiring boards is a direct result of the difference in thermal expansion between the copper foil and the epoxy board. Thermal cycling creates considerable stresses in the copper foil that can lead to foil cracking or separation from the board. In fact, copper foil cracking is one of the most common failures in electronic packages.[216] The best commercially available electrodeposited copper foils typically have average grain sizes of 300–400 nm[217,218] with a micro-hardness on the order of 130 kg/mm2.[218] Woo[172] and Woo and Erb[218] have shown that grain size reduction can substantially increase the hardness (strength) of such foils

Processing

280 Circuit trace etched copper

5 μm

Average grain size of 1 μm

5 μm

d

Average grain size of 100 nm

Figure 6.24 Schematic diagrams showing the potential improvements in wiring density by grain size reduction in copper foil used for printed wiring board applications.[172]

by means of Hall–Petch strengthening, e.g., to 230 kg/mm2 at an average grain size of 41 nm.[172] The development of undercut, i.e., the formation of a trapezoidal crosssection, and trace width limitation are directly related to the anisotropic dissolution of the copper foil during the lithographic etching process.[47] It is expected that the highly uniform corrosion characteristics observed for nanocrystalline copper[161] can be used to effectively reduce undercut and, at the same time, increase the wiring density, as shown schematically in Fig. 6.24. For this application, grain size optimization is required to find the best compromise between strength, etching behavior, and electrical resistivity, which increases due the electron scattering at grain boundaries as the grain size is decreased to below 100 nm.[172]

6.5.2 Nanocrystalline Metals in Microsystem Components In recent years, considerable progress has been made in the area of microsystem technology, which deals with intelligent miniaturized systems that combine sensing and/or actuating functions with processing functions. Most monolithic all-Si microsystems are constructed by various

6: Metals, Alloys, and Composites, Erb et al.

281

micromachining methods on a single-crystal substrate. Hybrid-type multifunctional microsystems, on the other hand, require the integration of several different materials that are produced by different methods. For the production of metallic microsystem components, several electrodeposition methods have been developed such as LIGA (German acronym for Lithographie [lithography], Galvanoformung [electroforming] and Abformung [molding]), HARMS (high-aspect-ratio moldings), or DEM (deep etching, electroforming and microreplication). The metals most commonly used in the electrodeposition of microcomponents are Ni, Cu, Au, Ag, and Ni—Fe alloys. However, in many cases the components produced by electrodeposition show considerable variations in their properties (see, e.g., Ref. 219), in particular mechanical properties. Large batch-to-batch variations and even local property variations in a single component are observed. Such variations are of great concern and can affect the overall performance and reliability of the entire microsystem. It has been shown that many of the nonuniform properties in electrodeposited microcomponents can be traced back to the microstructural evolution using conventional electrodeposition approaches.[214,215,220] As discussed in section 6.3, conventional electrodeposits normally exhibit a cross-sectional microstructure as shown in Fig. 6.6, starting with a finegrained structure at the substrate and developing a textured columnar grain structure with increasing deposit thickness. Baghbanan[221] and Baghbanan et al.[215] have analyzed the cross-sectional mechanical properties of Ni microcomponents with the microstructures shown in Fig. 6.6, using the nano-indentation technique. Figure 6.25 shows cross-sectional hardness and Young’s modulus for several traces parallel and perpendicular to the substrate. The hardness traces parallel to the substrate at distances of a few hundred micrometers from the substrate show more or less constant hardness of just under 2 GPa. However, the hardness trace perpendicular to the substrate shows an initial drop from about 3 GPa to about 2 GPa, which can be explained in terms of the grain size coarsening with increasing deposit thickness. Very large variations in the Young’s modulus, from as high as 280 GPa to as low as 130 GPa, were observed in traces parallel to the substrate in the region of columnar structure. On the basis of the relatively large anisotropy of the elastic properties in nickel, such variations in the Young’s modulus can be explained by the orientation changes in the columnar grains.[221] In contrast, microcomponents made with nanostructured electrodeposits such as shown in Fig. 6.7 exhibit the cross-sectional properties presented in Fig. 6.26.[221] First, the considerable grain refinement to about

282

Processing

Figure 6.25 Cross-sectional hardness (top) and Young’s modulus (bottom) for a microcomponent produced by conventional electrodeposition.[221]

Figure 6.26 Cross-sectional hardness (top) and Young’s modulus (bottom) for a microcomponent produced by nanoelectrodeposition.[221]

6: Metals, Alloys, and Composites, Erb et al.

283

15 nm results in an increase in hardness to more than 6 GPa with virtually no gradient in a direction perpendicular to the substrate. Second, the Young’s modulus in traces parallel to the substrate are relatively constant at a value of about 200 GPa. In addition to providing a much better uniformity in cross-sectional properties, it was shown that nanocrystalline microcomponents provide considerable enhancements of several performance indicators including elastic energy storage capacity (resilience), thermal shock resistance, tribological performance and reduced eddy currents for magnetic components operating at high frequencies.[214,215]

6.6 Summary Electrodeposition can be used to synthesize a large number of nanocrystalline metals, alloys, and composite materials. There is considerable opportunity to control their microstructure by process parameters such as bath composition, pH, temperature, current density, and current waveform. The properties of nanocrystalline electrodeposits have been shown to fall into two categories: properties exhibiting strong grain size dependence, and properties that, over large grain size ranges, are relatively unaffected by grain size. While some of the properties of these materials are fairly well understood, others need further research. In particular, further studies are required to develop a better understanding of deformation mechanisms and thermal stability. Nanoelectrodeposition can be considered a drop-in technology that can make use of existing electroplating infrastructure with only minor modifications. On the basis of their exceptional property combinations, a number of interesting applications have already been developed for these materials, ranging from large-scale structural application to components used in microsystems technology.

References 1. Schlesinger, M., and Paunovic, M (eds.), Modern Electroplating, 4th ed., Wiley, New York (2000). 2. Brenner, A., Electrodeposition of Alloys–Principles and Practice, Academic Press, New York (1963). 3. McMahon, G., and Erb, U., Microstruct. Sci., 17:447 (1989). 4. McMahon, G., and Erb, U., J. Mater. Sci. Lett., 8:865 (1989). 5. Palumbo, G., Thorpe, S.J., and Aust, K.T., Scripta Metall. Mater., 24:1347 (1990).

284

Processing

6. Aust, K.T., Erb, U., and Palumbo, G., in: Mechanical Properties and Deformation Behavior of Materials Having Ultrafine Microstructures (M. Nastasi, et al., eds.), p. 107, Kluwer Academic Publishers (1993). 7. Erb, U., Can. Metall. Q., 34:275 (1995). 8. Gleiter, H., in: Deformation of Polycrystals: Mechanisms and Microstructures, Proc. 2nd Risø Int. Symp. on Metallurgy and Materials Science, p. 15, Risø National Laboratory, Roskilde, Denmark (1981). 9. Rofagha, R., Langer, R., El-Sherik, A.M., Erb, U., Palumbo, G., and Aust, K.T., Scripta Metall. Mater., 25:2867 (1991). 10. Erb, U., El-Sherik, A.M., Palumbo, G., and Aust, K.T., Nanostruct. Mater., 2:383 (1993). 11. Erb, U., and El-Sherik, A.M., US Patent No. 5,352,266 (1994). 12. Bakonyi, I., Toth-Kadar, E., Toth, J., Tarnoczi, T., and Cziraki, A., in: Processing and Properties of Nanocrystalline Materials (C. Suryanarayana, et al., eds.), p. 465, TMS, Warrendale, PA (1996). 13. Bakonyi, I., Toth-Kadar, E., Tarnoczi, T., Varga, L.K., Cziraki, A., Gerocs, I., and Fogarassy, B., Nanostruct. Mater., 3:155 (1993). 14. Erb, U., El-Sherik, A.M., Cheung, C.K.S., and Aus, M.J., US Patent No. 5,433,797 (1995). 15. Würschum, R., Gruss, S., Gissibl, B., Natter, H., Hempelmann, R., and Schäfer, H.E., Nanostruct. Mater., 9:615 (1997). 16. Natter, H., and Hempelmann, R., J. Phys. Chem., 100:19525 (1996). 17. Saber, K.H., Koch, C.C., and Fedkiw, P.S., Mater. Sci. Eng. A, 341:174 (2003). 18. Grimmet, D.L., Schwarz, M., and Nobe, K., in: Electrochemical Technology in Electronics Symposium Proceedings (L.T. Romankiv, and T. Osaka, eds.), p. 561, The Electrochemical Society, New Jersey (1987). 19. Cheung, C., Djuanda, F., Erb, U., and Palumbo, G., Nanostruct. Mater., 5:513 (1995). 20. Alfantazi, A.M., El-Sherik, A.M., and Erb, U., Scripta Metall. Mater., 30:1245 (1994). 21. Alfantazi, A.M., and Erb, U., J. Mater. Sci. Lett., 15:1361 (1996). 22. Ganesan, P., Kumaraguru, S.P., and Popov, B.N., AESF SUR/FIN Proc., p. 1064 (2004). 23. Osmola, D., Renaud, E., Erb, U., Wong, L., Palumbo, G., and Aust, K.T., Mater. Res. Soc. Symp. Proc., 286:161 (1993). 24. Cheung, C., Synthesis and Microstructural Characterization of Electrodeposited Nanocrystalline Soft Magnets, Ph.D. Thesis, Queen’s University, Kingston, Ontario, Canada (2001). 25. Bryden, K.J., and Ying, J.Y., Nanostruct. Mater., 9:485 (1997). 26. Cheung, C., Erb, U., and Palumbo, G., Mater. Sci. Eng. A, 185:39 (1994). 27. Cheung, C., Nolan, P., and Erb, U., Mater. Lett., 20:135 (1994). 28. Cheung, C., Palumbo, G., and Erb, U., Scripta Metall. Mater., 31:735 (1994). 29. Zhang, Y., and Ivey, D.G., AESF SUR/FIN Proc., p. 821 (2004). 30. Kumaraguru, S.P., Ganesan, P., and Popov, B.N., AESF SUR/FIN Proc., (2004). 31. Veeraraghan, B., Kumaraguru, S.P., Haran, B., and Popov, B.N., AESF SUR/FIN Proc., p. 561 (2003).

6: Metals, Alloys, and Composites, Erb et al.

285

32. Palumbo, G., Erb, U., McCrea, J.L., Hibbard, D.G., Brooks, I., Gonzalez, F., and Panagiotopoulos, K., AESF SUR/FIN Proc., p. 204 (2002). 33. Oberle, R.R., Scanlon, M.R., Cammarata, R.C. and Searson, P.C., Appl. Phys. Lett., 66:19 (1995). 34. Qu, N.S., Chan, K.C., and Zhu, D., Scripta Mater., 50:1131 (2004). 35. Zimmerman, A.F., Clark, D.G., Aust, K.T., and Erb, U., Mater. Lett., 52:85 (2002). 36. Zimmerman, A.F., Palumbo, G., Aust, K.T., and Erb, U., Mater. Sci. Eng., A328:137 (2002). 37. Talbot, J.B., AESF SUR/FIN Proc., p. 701 (2003). 38. Palumbo, G., McCrea, J.L., and Erb, U., in: Encyclopedia of Nanoscience and Nanotechnology (H.S. Nalwa, ed.), vol. 1, p. 89, American Scientific Publishers, Stevenson Ranch, CA (2004). 39. Zhou, Y., Peng, X., and Wang, F., Scripta Mater., 50:1429 (2004). 40. Chen, X.H., Cheng, F.Q., Li, S.L., Zhou, L.P., and Li, D.Y., Surf. Coat. Tech., 155:274 (2002). 41. Bockris, J.O.M., in: Modern Aspects of Electrochemistry (J.O.M. Bockris, and B. Conway, eds.), Vol. 3, p. 224, Butterworths, London (1964). 42. Gorbunova, K.M., and Polukarov, Y.M., in: Advances in Electrochemistry and Electrochemical Engineering (W. Tobias, ed.), p. 249, Interscience, New York (1967). 43. Bockris, J.O.M., and Razumney, G.A., Fundamental Aspects of Electrocrystallization, p. 27, Plenum Press, New York (1967). 44. Choo, R.T.C., El-Sherik, A.M., Toguri, J., and Erb, U., J. Appl. Electrochem., 25:384 (1995). 45. Fischer, H., Angew. Chem. Int. Ed., 8:108 (1969). 46. Sheppard, K., Smith, D.A., Hentzell, H.T.G., and Ibrahim, A., in: Defect Structure, Morphology and Properties of Deposits (H. Merchant, ed.), p. 413, TMS, Warrendale, PA (1995). 47. Dini, J.W., Electrodeposition, Noyes Publications, Park Ridge, NJ (1993). 48. El-Sherik, A.M., and Erb, U., J. Mater. Sci., 30:5743 (1995). 49. Cziraki, A., Fogarassy, B., Gerocs, I., Toth-Kadar, E., and Bakonyi, I., J. Mater. Sci., 29:4771 (1994). 50. Alfantazi, A.M., Brehaut, G., and Erb, U., Surf. Coat. Tech., 89:239 (1997). 51. Cheung, C., Erb, U., and Palumbo, G., Mater. Sci. Eng., A185:39 (1994). 52. Zhou, Y., On the Interface Contribution to the Bulk Properties of Nanostructured Ni—P Alloys at Grain Sizes below 30 nm, Ph.D. Thesis, University of Toronto, Toronto, Ontario, Canada (2005). 53. Hansen, M., and Anderko, K., Constitution of Binary Alloys, p. 1027, McGraw-Hill, New York (1958). 54. Jeong, D., Abrasive Wear Behavior of Electrodeposited Nanocrystalline Materials, Ph.D. Thesis, University of Toronto, Ontario, Canada (2003). 55. Mehta, S.C., Smith, D.A., and Erb, U., Mater. Sci. Eng., A204:227 (1995). 56. Thomas, G.J., Siegel, R.W., and Eastman, J.A., Scripta Metall. Mater., 24:201 (1990). 57. Wunderlich, W., Ishida, Y., and Maurer, R., Scripta Metall. Mater., 24:403 (1990). 58. Van Petegem, S., Dalla Torre, F., Segers, D., and Van Swygenhoven, H., Scripta Mater., 48:17 (2003).

286

Processing

59. Haasz, T.R., Aust, K.T., Palumbo, G., El-Sherik, A.M., and Erb, U., Scripta Metall. Mater., 32:423 (1995). 60. Turi, T., Thermal and Thermodynamic Properties of Fully Dense Nanocrystalline Ni and Ni—Fe Alloys, Ph.D. Thesis, Queen’s University, Kingston, Ontario, Canada (1997). 61. Wang, N., Palumbo, G., Wang, Z., Erb, U., and Aust, K.T., Scripta Metall. Mater., 28:253 (1993). 62. Gill, K.S., Davis, B., Cooper, W.C., and Erb, U., Plat. Surf. Finish., 77(10):68 (1990). 63. Erb, U., Palumbo, G., and Aust, K.T. in: Nanostructured Films and Coatings (G.M. Chow, et al., eds.), NATO Sciences Series, p. 11, Kluwer Academic, Dordrecht, The Netherlands (2000). 64. Erb, U., Palumbo, G., Jeong, D., Kim, S., and Aust, K.T., in: Processing and Properties of Structural Nanomaterials (L.L. Shaw, et al., eds.), p. 109, TMS, Warrendale, PA (2004). 65. Gonzalez, F., Brennenstuhl, A.M., Palumbo, G., Erb, U., and Lichtenberger, P.C., Mater. Sci. Forum, 225–227:831 (1996). 66. Palumbo, G., Gonzalez, F., Brennenstuhl, A.M., Erb, U., Shmayda, W., and Lichtenberger, P.C., Nanostruct. Mater., 9:737 (1997). 67. Gleiter, H., Prog. Mater. Sci., 33:224 (1989). 68. Gong, W., Li, H., Zhao, Z., and Chen, J., J. Appl. Phys., 69:5119 (1991). 69. Gangopadhyay, S., Hadjipanayis, G.C., Dale, B., Sorensen, C.M., and Klabunde, K.J., Nanostruct. Mater., 1:77 (1992). 70. Schaefer, H.E., Kisker, H., Kronmüller, H., and Würschum, R., Nanostruct. Mater., 1:523 (1992). 71. Yao, Y.D., Chen, Y.Y., Hsu, C.M., Lin, H.M., Tung, C.Y., Tai, M.F., Wang, D.H., Wu, K.T., and Suo, C.T., Nanostruct. Mater., 6:933 (1995). 72. Krill, C.E., Merzoug, F., Krauss, W., and Birringer, R., Nanostruct. Mater., 9:455 (1997). 73. Aus, M.J., Szpunar, B., El-Sherik, A.M., Erb, U., Palumbo, G., and Aust, K.T., Scripta Metall. Mater., 27:1639 (1992). 74. Szpunar, B., Erb, U., Aust, K.T., Palumbo, G., and Lewis, L., Mater. Res. Soc. Symp. Proc., 318:477 (1994). 75. Szpunar, B., Zugic, R., Erb, U., and Lewis, L., Can. Metall. Q., 349:281 (1995). 76. Szpunar, B., Erb, U., Palumbo, G., Aust, K.T., and Lewis, L.J., Phys. Rev. B, 53:5547 (1996). 77. Daroczi, L., Beke, D.L., Posgay, G., Zhou, G.F., and Bakker, H., Nanostruct. Mater., 2:512 (1993). 78. Daroczi, L., Beke, D.L., Posgay, G., and Kis-Varga, M., Nanostruct. Mater., 6:981 (1995). 79. Weissmüller, J., McMichael, R.D., Barker, J., Brown, H.J., Erb, U., and Shull, R.D., Mater. Res. Soc. Symp. Proc., 457:231 (1997). 80. Kisker, H., Gessmann, T., Würschum, R., Kronmüller, H., and Schaefer, H.E., Nanostruct. Mater., 6:925 (1995). 81. Aus, M.J., Cheung, C., Szpunar, B., Erb, U., and Szpunar, J.A., J. Mater. Sci. Lett., 17:1949 (1998). 82. Szpunar, B., Aus, M.J., Cheung, C., Erb, U., Palumbo, G., and Szpunar, J.A., J. Magn. Magn. Mater., 187:325 (1998).

6: Metals, Alloys, and Composites, Erb et al.

287

83. Cheung, C., Aus, M.J., Erb, U., McCrea, J.L., and Palumbo, G., in: Proc. 6th Int. Conf. Nanostruct. Mater., (NANO 2002), CD-ROM, Nanotechnology Enterprises Inc., Rutgers University, NJ (2003). 84. Turi, T., and Erb, U., Mater. Sci. Eng. A, 204:34 (1995). 85. Rupp, J., and Birringer, R., Phys. Rev. B, 36:7888 (1987). 86. Lu, K., Wang, J.T., and Wei, W.D., J. Phys. D, 25:808 (1992). 87. Wagner, M., Phys. Rev. B, 45:376 (1992). 88. Fecht, H.J., Phys. Rev. Lett., 65:610 (1990). 89. Gleiter, H., presented at 2nd Int. Conf. on Nanostructured Materials, Stuttgart (Oct. 1994). 90. Szpunar, B., Lewis, L.J., Swainson, I., and Erb, U., Phys. Rev. B, 60:10107 (1999). 91. Nieman, G.W., Weertman, J.R., and Siegel, R.W., Scripta Metall. Mater., 24:145 (1990). 92. Korn, D., Morsch, A., Birringer, R., Arnold, W., and Gleiter, H., J. de Phys., 49:C5–769 (1988). 93. Nieman, G.W., Weertman, J.R., and Siegel, R.W., J. Mater. Res., 6:1012 (1991). 94. Krstic, V., Erb, U., and Palumbo, G., Scripta Metall. Mater., 29:1501 (1993). 95. Zugic, R., Szpunar, B., Krstic, V., and Erb, U., Phil. Mag. A, 75:1041 (1997). 96. Sanders, P.G., Youndahl, C.J., and Weertman, J.R., Mater. Sci. Eng., A234–236:77 (1997). 97. Zhou, Y., Erb, U., Aust, K.T., and Palumbo, G., Scripta Mater., 48:825 (2003). 98. Zhou, Y. Erb, U., Aust, K.T., and Palumbo, G., Z. Metallkd., 94:1157 (2003). 99. Shen, T.D., Koch, C.C., Tsui, T.Y., and Pharr, G.M., J. Mater. Res., 10:2892 (1995). 100. Somekawa, H., Nieh, T.G., and Higashi, K., Scripta Mater., 50:1361 (2004). 101. Safranek, W.H., The Properties of Electrodeposited Metals and Alloys, 2nd ed., AESF, Orlando, FL (1986). 102. El-Sherik, A.M., Erb, U., Palumbo, G., and Aust, K.T., Scripta Metall. Mater., 27:1185 (1992). 103. El-Sherik, A.M., and Erb, U., Nickel-Cobalt 97, in: Applications and Materials Performance (F.N. Smith, et al., eds.), Vol. IV, p. 257, The Metallurgical Society of CIM, Montreal (1997). 104. Wang, N., Wang, Z., Aust, K.T., and Erb, U., Mater. Sci. Eng. A, 237:150 (1997). 105. Jeong, D., Abrasive Wear Behavior of Electrodeposited Nanocrystalline Materials, Ph.D. Thesis, University of Toronto, Ontario, Canada (2003). 106. Aus, M.J., Szpunar, B., Erb, U., El-Sherik, A.M., Palumbo, G., and Aust, K.T., J. Appl. Phys., 75:3632 (1994). 107. McCrea, J.L., The Effect of Temperature on the Electrical Resistivity of Electrodeposited Nanocrystalline Materials, Ph.D. Thesis, University of Toronto, Ontario, Canada (2001). 108. Hall, E.O., Proc. Phys. Soc., London, B64:747 (1951). 109. Petch, N.J., Iron Steel Inst., 174:24 (1953).

288

Processing

110. Chokshi, A.H., Rosen, A.H., Karch, J., and Gleiter, H., Scripta Metall. Mater., 23:1679 (1989). 111. Palumbo, G., Erb, U., and Aust, K.T., Scripta Metall. Mater., 24:2347 (1990). 112. Rabukhin, V.B., Phys. Met. Metall., 61:149 (1985). 113. Lehockey, E.M., Palumbo, G., Aust, K.T., Erb, U., and Lin. P., Scripta Metall., 39:341 (1998). 114. Bollmann, W., Philos. Mag., A49:73 (1984). 115. Sadoc, J.F., and Mosseri, R., J. Physique, 45:1025 (1984). 116. Ovid’ko, I.A., Phys. Met. Metall., 67:18 (1989). 117. Brassard, J., Deformation Mechanisms During Creep and Cold Rolling of Nanocrystalline Nickel., M.A.Sc. Thesis, University of Toronto, Ontario, Canada (2003). 118. Markmann, J., Bunzel, P., Rösner, H., Liu, K.W., Padmanabhan, K.A., Birringer, R., Gleiter, H., and Weissmüller, J., Scripta Mater., 49:637 (2003). 119. Ebrahimi, F., Bourne, G.R., Kelly, M.S., and Matthews, T.E., Nanostruct. Mater., 11:343 (1999). 120. Dalla Torre, F., Van Swygenhoven, H., and Victoria, M., Acta Mater., 50:3957 (2002). 121. Kumar, K.S., Van Swygenhoven, H., and Suresh, S., Acta Mater., 51:5743 (2003). 122. Yin, W.M., and Whang, S.H., Scripta Mater., 44:569 (2001). 123. Hughes, G.D., Smith, S.D., Pande, C.S., Johnson, H.R., and Armstrong, R.W., Scripta Metall., 20:93 (1986). 124. Kumar, K.S., Suresh, S., Chisholm, M.F., Horton, J.A., and Want, P., Acta Mater., 51:387 (2003). 125. Lu, L., Li, S.X., and Lu, K., Scripta Mater., 45:1163 (2001). 126. Karimpoor, A.A., Mechanical Properties of Nanocrystalline Cobalt, Ph.D. Thesis, University of Toronto, Ontario, Canada (2005, in progress). 127. Karimpoor, A.A., Erb, U., Aust, K.T., Wang, Z., and Palumbo, G., J. Metastable Nanostruct. Mater., 13:415 (2002). 128. Karimpoor, A.A., Erb, U., Aust, K.T., and Palumbo, G., Scripta Mater., 49:651 (2003). 129. McCrea, J.L., Palumbo, G., Hibbard, G.D., and Erb, U., Rev. Adv. Mater. Sci., 5:252 (2003). 130. Wei, H.H., Tensile Properties of Nanocrystalline Nickel-Iron Alloys, M.A.Sc. Thesis, University of Toronto, Ontario, Canada, (2006, in progress). 131. Shen, Y.F., Lu, L., Lu, Q.H., Jin, Z.H., and Lu, K., Scripta Mater., 49:651 (2003). 132. Chan, K.C., Wang, G.F., Wang, C.L., and Zhang, K.F., Scripta Mater., 53:1285 (2005). 133. Schwaiger, R., Moser, B., Dao, M., Chollacoop, N., and Suresh, S., Acta Mater., 51:5159 (2003). 134. McFadden, S.X., Zhilyaev, A.P., Mishra, R.S., and Mukherjee, K., Mater. Lett., 45:345 (2000). 135. Derlet, P.M., Hasnaoui, A., and Van Swygenhoven, H., Scripta Mater., 49:629 (2003).

6: Metals, Alloys, and Composites, Erb et al.

289

136. Hanlon, T., Kwon, Y.N., and Suresh, S., Scripta Mater., 49:675 (2003). 137. McCrea, J.L., Palumbo, G., Gonzalez, F., Robertson, A., Panagiotopoulos, K., and Erb, U., AESF SUR/FIN Proc., CD-ROM, Section D, Orlando, FL (2001). 138. Koch, C.C., and Shen, T.D., in: Processing and Properties of Nanocrystalline Materials (C. Suryanarayana, et al., eds.), p. 333, TMS, Warrendale, PA (1996). 139. Schuh, C.A., Nieh, T.G., and Iwasaki, H., Acta Mater., 51:431 (2003). 140. Jeong, D.H., Gonzalez, F., Palumbo, G., Aust, K.T., and Erb, U., Scripta Mater., 44:493 (2001). 141. Jeong, D.H., Erb, U., Aust, K.T., Robertson, A., McCrea, J.L., and Palumbo, G., AESF SUR/FIN Proc., CD-ROM, Section D, Orlando, FL (2001). 142. Jeong, D.H., Erb, U., Aust, K.T., and Palumbo, G., AESF SUR/FIN Proc., p. 235, Orlando, FL (2002). 143. Jeong, D.H., Erb, U., Aust, K.T., and Palumbo, G., Mater. Sci. Forum, 408–412:925 (2002). 144. Schuh, C.A., Nieh, T.G., and Yamasaki, T., Scripta Mater., 46:735 (2002). 145. Jeong, D.H., Erb, U., Aust, K.T., and Palumbo, G., Scripta Mater., 48:1067 (2003). 146. Jeong, D.H., Erb, U., Aust, K.T., and Palumbo, G., Metastable Nanostruct. Mater., 15–16:635 (2003). 147. Jeong, D.H., Erb, U., McCrea, J.L., and Palumbo, G., AESF SUR/FIN Proc., p. 697, Orlando, FL (2003). 148. Archard, J.F., Appl. Phys., 44:981 (1953). 149. Palumbo, G., and Aust, K.T., in: Materials Interfaces (D. Wolf, and S. Yip, eds.), Chapman and Hall, New York (1991). 150. Rofagha, R., Langer, R., El-Sherik, A.M., Erb, U., Palumbo, G., and Aust, K.T., Mater. Res. Soc. Symp. Proc., 238:751 (1992). 151. Rofagha, R., Erb, U., Ostrander, D., Palumbo, G., and Aust, K.T., Nanostruct. Mater., 2:1 (1993). 152. Rofagha, R., Splinter, S.J., Erb, U., and McIntyre, N.S., Nanostruct. Mater., 4:69 (1994). 153. Wang, S., Rofagha, R., Roberge, P.R., and Erb, U., Electrochem. Soc. Proc., 95–98:224 (1995). 154. Tang, P.T., Watanabe, T., Andersen, J.E.T., and Bech-Nielsen, G., J. Appl. Electrochem., 25:347 (1995). 155. Kim, S.H., Aust, K.T., Erb, U., and Gonzalez, F., AESF SUR/FIN Proc., p. 225, Orlando, FL (2002). 156. Kim, S.H., Aust, K.T., Erb, U., Gonzalez, F., and Palumbo, G., Scripta Mater., 48:1379 (2003). 157. Kim, S.H., Franken, T., Hibbard, G.D., Erb, U., Aust, K.T., and Palumbo, G., J. Metastable Nanostruct. Mater., 15–16:643 (2003). 158. Kim, S.H., Corrosion Properties of Nanocrystalline Ni and Co Electrodeposits, Ph.D. Thesis, University of Toronto, Ontario, Canada (2004). 159. Saito, M., Yamada, K., Ohashi, K., Yasue, Y., Sogawa, Y., and Osaka, T., J. Electrochem. Soc., 146:2845 (1999). 160. Benea, L., Bonora, P.L., Borelli, A., and Martelli, S., Wear, 249:995 (2002). 161. Yu, B., Corrosion Behavior of Nanocrystalline Copper, M.A.Sc. Thesis, University of Toronto, Ontario, Canada (2006, in progress).

290

Processing

162. El-Sherik, A.M., and Erb, U., Plat. Surf. Finish., 82(9):85 (1995). 163. Splinter, S.J., Rofagha, R., McIntyre, N.S., and Erb, U., Surf. Interface Anal., 24:181 (1996). 164. Youssef, Kh.M.S., Koch, C.C., and Fedkiw, P.S., Corrosion Science, 46:51 (2004). 165. Aledresse, A., and Alfantazi, A., J. Mater. Sci., 39:1523 (2004). 166. Palumbo, G., and Erb, U., MRS Bulletin, 24:27 (November 1999). 167. Seeger, A., and Schottky, G., Acta Metall., 7:495 (1959). 168. Andrews, P.V., West, W.B., and Robenson, C.R., Philos. Mag., 19:887 (1969). 169. Kasen, M.B., Philos. Mag., 21:599 (1970). 170. Gangulee, A., J. Appl. Phys., 43:867 (1972). 171. Nakamichi, I., J. Sci. Hiroshima Univ., 54A:49 (1990). 172. Woo, P., Synthesis and Characterization of Nanocrystalline Copper for Electronic Applications, University of Toronto, Ontario, Canada, (2006, in progress). 173. McCrea, J.L., Aust, K.T., Palumbo, G. and Erb, U., Mater. Res. Soc. Symp. Proc., 581:461 (2000). 174. Qian, L.H., Lu, Q.H., Kong, W.J., and Lu, K., Scripta Mater., 50:1407 (2004). 175. El-Sherik, A.M., Boylan, K., Erb, U., Palumbo, G., and Aust, K.T., Mater. Res. Soc. Symp. Proc., 238:727 (1992). 176. Cziraki, A., Tonkovics, Z., Gerocs, I., Fogarassy, B., Groma, I., Toth-Kadar, E., Tarnoczi, T., and Bakonyi, I., Mater. Sci. Eng., A179/180:531 (1994). 177. Klement, U., Erb, U., El-Sherik, A.M., and Aust, K.T., Mater. Sci. Eng. A, 203:177 (1995). 178. Klement, U., Erb, U., and Aust, K.T., Nanostruct. Mater., 6:581 (1995). 179. Wang, N., Wang, Z., Aust, K.T., and Erb, U., Acta Mater., 45:1655 (1997). 180. Natter, H., Schmelzer, M., and Hempelmann, R., J. Mater. Res., 15:1186 (1998). 181. Iordache, M.C., Whang, S.H., Jiao, Z., and Wang, Z.M., Nanostruct. Mater., 11:1343 (1999). 182. Hibbard, G.D., Erb, U., Aust, K.T., and Palumbo, G., Mater. Res. Soc. Symp. Proc., 580:183 (2000). 183. Warren, P.J., Thuvander, M., Abraham, M., Lane, H., Gerezo, A., and Smith, G.D.W., Mater. Sci. Forum, 343–346:701 (2000). 184. Xiao, C., Mishrams, R.A., Whang, S.H., and Yin, W.M., Mater. Sci. Eng., A301:35 (2001). 185. Hibbard, G.D., Microstructural Evolution During Annealing of Nanostructured Electrodeposits, Ph.D. Thesis, University of Toronto, Ontario, Canada (2002). 186. Hibbard, G.D., McCrea J.L., Palumbo, G., Aust, K.T., and Erb, U., Scripta Mater., 47:83 (2002). 187. Boylan, K., Ostrander, D., Erb, U., Palumbo, G., and Aust, K.T., Scripta Metall. Mater., 25:2711 (1991). 188. Osmola, D., Nolan, P., Erb, U., Palumbo, G., and Aust, K.T., Phys. Status Solidi, A131:569 (1992). 189. Abraham, M., Thuvander, M., Lane, H.M., Cerezo, A., and Smith, G.D.W., Mater. Res. Soc. Symp. Proc., 581:517 (2002).

6: Metals, Alloys, and Composites, Erb et al.

291

190. Hentschel, T.H., Isheim, D., Kirchheim, R., Muller, F., and Kreye, H., Acta Mater., 48:933 (2000). 191. Czerwinski, F., Li, H., Megret, M., and Szpunar, J.A., Scripta Mater., 37:1967 (1997). 192. Czerwinski, F., Li, H., Megret, F., Szpunar, J.A., Clark, D.G., and Erb, U., Mater. Res. Soc. Symp. Proc., 451:501 (1997). 193. Bryden, K.J., and Ying, J.Y., Electrochem. Soc., 145:3339 (1998). 194. Hibbard, D.G., Aust, K.T., Palumbo, G., and Erb, U., Scripta Mater., 44:513 (2001). 195. Hibbard, G.D., Aust, K.T., Palumbo, G., and Erb, U., Phil. Mag., 86:125 (2006). 196. Choi, P., da Silva, M., Klement, U., Al-Kassab, T., and Kirchheim, R., Acta Mater., 53:4473 (2005). 197. Hibbard, G.D., Aust, K.T., and Erb, U., Mater. Sci. Eng., submitted (2005). 198. Choi, P., Al-Kassab, T., Gartner, F., Kreye, H., and Kirchheim, R., Mater. Sci. Eng., A353:1967 (1997). 199. Hibbard, G.D., Erb, U., Aust, K.T., Klement, U., and Palumbo, G., Mater. Sci. Forum, 386–388:387 (2002). 200. Aust, K.T., Hibbard, G.D., Palumbo, G., and Erb, U., Z. Metallk., 94:10 (2003). 201. Aust, K.T., Hibbard, G.D., and Palumbo, G., in: Encyclopedia of Nanoscience and Nanotechnology (H.S. Nalwa, ed.), p. 267, American Scientific Publ., Stevenson Ranch, CA (2004). 202. Chen, L.C., and Spaepen, F., J. Appl. Phys., 69:679 (1991). 203. Kaur, I., Gust, W., and Kosma, L., (eds.), Handbook of Grain and Interface Boundary Diffusion Data, Ziegler Press, Stuttgart (1989). 204. Turi, T., and Erb, U., unpublished data. 205. Soong, C., Synthesis and Thermal Stability of Sulphur-Free Nanocrystalline Nickel, M.A.Sc. Thesis, University of Toronto, Ontario, Canada, (2006, in progress). 206. Palumbo, G., Gonzalez, F., Tomantschger, K., Erb, U., and Aust K.T., Plat. Surf. Finish, 90(2):36 (2003). 207. Rofagha, R., Wood, D., and Erb, U., AESF SUR/FIN Proc., p. 197. (1995). 208. Erb, U., Brooks, I., Alfantazi, A., and Palumbo, G., AESF SUR/FIN Proc., Section D (CD-ROM) (2001). 209. Palumbo, G., Erb, U., McCrea, J.L., Hibbard, G.D., Brooks, I., and Gonzalez, F., AESF SUR/FIN Proc., p. 204 (2002). 210. Kim, S.H., Aust, K.T., Erb, U., Ogundale, G., and Gonzalez, F., AESF SUR/FIN Proc., p. 225 (2002). 211. McCrea, J.L., Palumbo, G., Gonzalez, F., Robertson, A., Panagiotopoulos, K., and Erb, U., AESF SUR/FIN Proc., Section D (CD-ROM) (2001). 212. McCrea, J.L., Palumbo, G., Lee, D.E., and Psaila-Dombrowski, M.J., AESF/EPA Proc., p. 186 (2003). 213. Palumbo, G., Gonzalez, F., Tomantschger, K., Erb, U., and Aust, K.T., AESF SUR/FIN Proc., p. 528 (2003). 214. Erb, U., Baghbanan, M.R., Cheung, C., and Palumbo, G., in: Processing and Fabrication of Advanced Materials XII (T.S. Srivatsan, and R.A. Varin, eds.), p. 301, ASM Materials Park, OH, (2004).

292

Processing

215. Baghbanan, M.R., Erb, U., and Palumbo, G., in: Nanostructured Metals for Enhanced Performance of LIGA Components, Materials Processing and Manufacturing Fifth Global Symposium (S.M. Mukhopadhyay, et al., eds.), p. 307, TMS, Warrendale, PA (2004). 216. Coombs, C.F., (ed.), Printed Circuit Handbook, 5th ed., McGraw-Hill, New York (2001). 217. Merchant, H.D., Khatibi, G., and Weiss, B. J. Mater. Sci., 39:4157 (2004). 218. Woo, P., and Erb, U., Proc. IPC Printed Circuits Expo, APEX and Designers Summit, Section S37-02 (CD-ROM), Anaheim, CA (2006). 219. Buchheit, T.E., Christenson, T.R., Schmale, D.T., and Lavan, D.A., Mater. Res. Soc. Symp. Proc., 546:121 (1999). 220. Prasad, S.V., Michel, J.R., and Christenson, T.R., Scripta Mater., 48:255 (2003). 221. Baghbanan, M.R., Application of Nanostructured Materials in Microsystem Components, University of Toronto, Ontario, Canada (2006, in progress).

7 Computer Modeling of Nanostructured Materials Donald W. Brenner Department of Materials Science and Engineering, North Carolina State University, Raleigh, North Carolina, USA

7.1 Introduction Over the last decade, computer modeling has evolved to become a third pillar of research alongside experiment and theory. There are many reasons for this emergence, the most prominent of which include an exponential increase in processing speed, relatively inexpensive platforms for parallel computing and data storage, new visualization capabilities, and the development of powerful algorithms that take full advantage of these advances in hardware. In addition to hardware and software capabilities, advances in simulation methodologies have made the results of atomiclevel computer modeling reliable to the extent that in many cases the simulations can replace expensive and difficult experiments. Methods that enable atomistic dynamics using full electronic energies now allow processes involving up to several thousand atoms to be accurately modeled. Similarly, formalisms like the moments expansion have produced relatively simple analytic potential energy expressions that can capture quantum-mechanical bonding properties for systems involving well over a billion atoms.[1] Computer modeling has played an especially important role in developing our current understanding of nanometer-scale structures and processes.[2] Indeed, among the many scientific and technological advances provided by the current emphasis on nanotechnology is the ability for computer modeling and experiment to characterize phenomena on a common scale. Atomic-level computer modeling is now commonly used to explore and predict new phenomena that can be probed experimentally, to suggest new materials and structures with unique and desirable properties, to provide insight into the results of experiments, to generate data for larger-scale analysis, and to test scaling laws and analytic Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 293–328 © 2007 William Andrew, Inc.

293

294

Processing

theories. In the case of nanostructured solids, computer modeling is allowing researchers to “see” and analyze at the atomic level the structure, deformation mechanisms, and thermal-mechanical properties of these new materials with unprecedented detail.[3,4] Computer modeling of nanostructured solids has been a very active and fruitful field that continues to expand. Because of this, reviewing all of the work in this field in a single chapter is an impossible task. Rather than providing a comprehensive review of the literature, the intent of this chapter is to highlight some of the insights and results that computer modeling has provided in the field of nanostructured materials, and to point out common themes regarding properties of materials induced by structure at this scale. There are also several on-going debates among the leading modelers in this area that are related to the fundamental grain boundary structure and the origins of the mechanical behavior of these materials. The intent of this chapter is not to “weigh-in” on these issues, but rather to point out discrepancies where they exist, with references to other sources where appropriate arguments have been made. The chapter starts with a brief discussion of the atomic-level modeling methods that have played the largest role in this field. Included in this section is a discussion of some of the recent multiscale modeling methodologies that have been developed for materials modeling. Although these methodologies have yet to play a large role in the simulation of nanostructured materials, they offer the potential for directly bridging the atomic-level results with more macroscopic measurements of mechanical properties. As such, they will play an increasing role in this area over the next decade. In the remaining sections, modeling efforts have been broken down into the phenomena being studied, starting with a brief discussion of the properties of nanometer-scale clusters. As mentioned above, no attempt has been made to include all of the literature in a given area, but rather particular studies have been chosen to illustrate some of the contributions that atomic-level computer modeling has made to the field of nanostructured solids.

7.2 Modeling Methods 7.2.1 Molecular Dynamics and Monte Carlo Modeling The two standard methods for modeling nanometer-scale systems are molecular dynamics and Monte Carlo simulation. In the former technique,

7: Computer Modeling, Brenner

295

classical equations of motion for each atom are integrated step-wise in time. Time steps typically range from one to tens of femtoseconds, and simulations are typically carried out for picoseconds to tens of nanoseconds depending on system size, the phenomena to be studied, and the method in which interatomic forces are calculated. In a typical equilibrium Monte Carlo simulation, atomic configurations are generated with a probability that is proportional to their Boltzmann factor. Thermodynamic quantities are then obtained by averaging over the properties of each configuration. Alternatively, in kinetic Monte Carlo simulation, time-ordered configurations are generated, typically using some rate expression. Monte Carlo simulations that use a Boltzmann factor to determine configuration probabilities generate thermodynamics properties that correspond to a canonical ensemble. Other ensemble averages can be generated in a straightforward manner by using standard partition functions to define probabilities that take into account volume, pressure, or the number and type of atoms in a simulation. Integrating classical equations of motion in a molecular dynamics simulation and averaging over system properties generates thermodynamic averages that correspond to a microcanonical ensemble provided that the system adequately samples the appropriate phase space. Controlling the temperature and pressure in a molecular dynamics simulation so that other ensemble averages are generated is not as straightforward as it is for Monte Carlo simulations, and many different schemes have been proposed over the last two decades. Two of these schemes have been of particular use in atomic simulations, including studies of nanostructured solids. The first is the control of stress using the Parrinello–Raman method.[5] In this approach, a fictitious Lagrangian is written that involves atomic degrees of freedom as well as a set of degrees of freedom that correspond to strain components of a bounding box that defines periodic boundaries. System stresses play the role of interatomic forces for the strain components, and coupled equations of motion are simultaneously integrated for the strain components and the atom properties. This allows the bounding box size and shape to change in response to the combination of stresses within the atomic system and any externally applied stresses. The second scheme that has profoundly impacted simulation capabilities was introduced by Nosè,[6] and was based on the concept of constrained dynamics developed by Hoover, Evans, and others. In this approach, constraint forces are added to the equations of motion in such a way that not only are appropriate time-averaged properties produced, but fluctuations in these properties are also maintained that are appropriate for a finite-sized system. These temperature–pressure

296

Processing

constraints are commonly referred to as the Nosè–Hoover thermostat. Further details regarding these constraint forces are given in an excellent review by Hoover.[7]

7.2.2 Atomic Potential Energies and Forces Equilibrium Monte Carlo modeling requires specifying a potential energy as a function of atomic positions to calculate Boltzmann factors, while molecular dynamics simulations require interatomic forces, which can be obtained as partial derivatives of the potential energy. Two approaches are typically used to calculate interatomic energies and forces. In the least computationally intensive approach, the interactions between atoms due to the electrons are replaced with effective interactions that are typically described with an analytic potential energy function. At present there is no definitive mathematical form for the potential energy function, and forms that range from relatively simple pair-additive expressions to complicated many-body forms are used depending on the system, the type of bonding, and the phenomena to be studied. For metals, widely-used formalisms include the embedded-atom method,[8,9] Finnis-Sinclair,[10] and effective medium theory potentials.[11] For covalent materials, bondorder[12–17] and Stillinger–Weber[18] forms have found widespread use, although there are many other potential forms used for each class of material, including forms that model charge transfer. A more in-depth discussion of these potentials, including the quantum-mechanical origins of some of the many-body potentials, is given in Ref. 1. In the second approach to calculating potential energies, explicit electronic degrees of freedom are retained, the energy for which is calculated either using first-principles methods or through a simplified semiempirical Hamiltonian. This approach has several important advantages over the analytic bonding expressions, including energies that are often more transferable between bonding environments, understanding relations between electronic states and bonding energies and structures, and the ability to fully model chemical reactivity. The calculation of total energies from first principles is also well defined in terms of basis sets, electron correlation for ab initio methods, or choice of density functional expression and pseudopotential for density functional theory calculations. This is in contrast to analytic potentials, where a set of parameters (and often entirely new functional forms) must be developed for each system. Retaining the electrons, however, is computationally expensive relative to analytic potentials, and therefore first-principles studies are typically

7: Computer Modeling, Brenner

297

restricted to systems of from tens to up to a few thousands of atoms depending on factors such as the level of electronic structure used and available computing resources. Because of this, much more has been done in modeling nanostructured solid with analytic potentials than has been done with first-principles methods.

7.2.3 Multiscale Modeling Due to current limitations in computational resources, potential energy functions for molecular dynamics simulations are typically restricted to system sizes of ∼103–108 atoms and timescales up to about 10−9 seconds. For studying nanometer-scale phenomena, the size limitations are not especially critical because they match much of the processes of interest, because boundary conditions can be applied that mimic larger systems, and because system sizes can be increased relatively cheaply with massively parallel computing platforms. On the other hand, the time limitations can be a severe problem. Equations of motion are solved sequentially in time using a time step that is necessarily on the scale of the fastest process in a system, typically the highest-frequency vibrational period. To gain significant speed-up in time without some other numerical approximation requires faster processors, which is more challenging from a technology viewpoint than is adding more processors. Increasing system temperature can speed up activated processes, and similarly conditions such as working at very high strain rates can induce dynamics on the simulation timescale. Unfortunately, both can also produce dynamics that may be nonphysical, unpredictable, or at least not typical of the longertimescale processes that one may be trying to model. Monte Carlo modeling has similar limitations. In thermodynamic Monte Carlo simulation, energies but not forces are needed, and therefore somewhat larger systems can be modeled. However, only modest gains are possible that are not sufficient to reach significantly larger systems. Thermodynamic Monte Carlo simulation chooses systems from an appropriate ensemble independent of their time sequence, and therefore dynamic processes are not modeled. In kinetic Monte Carlo modeling, competing processes are chosen based on their rate, with higher-rate processes having a higher probability of occurring. The timescale is determined by the processes with the highest rate, and therefore for very slow processes experimental timescales are accessible. For a mix of slow and fast processes, it can often be assumed that the fast processes lead to equilibrium configurations, and therefore thermodynamic and kinetic

298

Processing

Monte Carlo modeling can in principle be combined in these situations to approach long times. A major drawback in kinetic Monte Carlo modeling is that the dynamic processes and their rates must be known, and often not all of the relevant processes are well characterized. For example, for many years it was assumed that diffusion of surface atoms took place strictly by an activated “hopping” of adatoms from binding site to binding site. Reaction paths and associated barrier heights were therefore calculated for these types of trajectories. In the 1980s and 1990s, however, simulations on metals[19] and covalent surfaces[2] demonstrated that processes in which adatoms exchange with atoms in the first few atomic layers of the solid can have lower barriers and therefore contribute significantly to net surface diffusion. A number of multiscale modeling approaches have been developed in attempts to overcome the spatial and temporal limitations inherent in molecular dynamics and Monte Carlo modeling. In serial multiscale modeling, information from one scale is passed to another scale. Examples of this information include binding energies, elastic properties, dynamic mechanisms, rate constants, activation energies, scaling relations, or in some cases constitutive relations. In concurrent multiscale modeling, multiple length or time scales are included in a single simulation, usually by some hybrid modeling approach. Examples include coupling atomic regions to a finite element mesh,[21,22] replacing constitutive relations in a fine mesh with properties from an atomic simulation,[23] and Lagrangians that couple finite-element nodes and atomic degrees of freedom.[24] Schemes for extending the temporal scale of simulations have also been introduced. Voter, for example, introduced a method for modeling rare event processes in which an effective potential energy hypersurface is defined such that energy wells between regions corresponding to rare events are “filled in.”[25] A molecular dynamics simulation is carried out on the hypersurface, with the true time of the simulation taken as the time step on the hypersurface multiplied by a scaling factor consisting of the ratio of the Boltzmann factors of the true and hypersurface potential energies. The key to applying this method is to define an appropriate hypersurface. For well-defined processes, such as surface diffusion on an ordered surface from one binding site to another, the hypersurface is straightforward to define. For less well-defined dynamics, the choice of hypersurface may not be straightforward, and different methodologies for determining these surfaces “on-the-fly” have been introduced. While multiscale modeling has been applied to many problems in solid mechanics and materials properties, it has not yet played a major role in modeling nanostructured solids. As this field continues to mature in terms

7: Computer Modeling, Brenner

299

of the synthesis, characterization and applications of structural materials at the macroscopic scale, and as our understanding of the origins of the mechanical properties on the nanometer scale improves, these types of approaches will undoubtedly play a critical role in this field.

7.3 Nanostructured Materials As discussed in detail elsewhere in this book, nanostructured solids have a unique combination of properties that make them potentially useful for a range of applications. However, much of the detail of the underlying processes and structures that are responsible for these properties remains poorly understood. With increasing computing power, more reliable analytic potentials, and better analysis tools, computer modeling is now able to provide critical details about these processes and structures that often cannot be obtained in any other way. These details are in turn leading to new and important insights into the relationship between structure and properties of these systems.

7.3.1 Nanoparticle Properties Understanding the unique properties of nanometer-scale clusters has been a longstanding goal of the scientific community. It is well established experimentally, for example, that there can be “magic number” structures that show an anomalous stability within a family of clusters of different sizes.[26] This effect has been attributed to both the filling of mesoelectronic shells and the completion of high-symmetry structures. A related area in which computer modeling has made seminal contributions is in understanding the phase stability of nanoclusters relative to their bulk counterpart. Other examples include simulations of the size dependence of vibrational states, mode broadening and thermal transport, and the dynamics of compression-induced phase changes and surface reactions. Although not directly related to nanostructured solids, these studies do provide some perspective and background on how structure and dynamics that are facilitated by the nanometer scale may help drive processes such as nanoparticle sintering and amorphous grain boundary formation, as described below. These studies also help demonstrate the complexity of these systems, and how details of the simulation, for example, the potential energy function, can influence a simulation result. A brief discussion of some of the recent modeling highlights in nanocluster research is therefore included in this section.

300

Processing

The relative stability of different phases can change in dramatic and often nonintuitive ways as particles are reduced to nanometer-scale dimensions. Landman and co-workers, for example, used molecular dynamics simulations to analyze the structure and dynamics of gold clusters containing 75, 146, or 459 atoms as a function of temperature.[27] Interatomic forces and energies were obtained from the embedded-atom method; therefore, any effects from electronic states were not included in the simulations. At low temperatures all of the clusters were crystalline. The two smaller clusters had high-symmetry truncated Mackay decahedral structures, while the larger cluster was a face-centered cubic crystallite with a truncated octahedral morphology. As the temperature was increased from 300 K, all three clusters showed diffusionless transitions between different solid shapes, with the initial temperatures for these transitions increasing with increasing cluster size. Further heating resulted in the formation of surface liquid layers at temperatures well below the bulk melting point. In contrast to the solid transitions, where the entire cluster would change shape, upon premelting the liquid remained at the surface while the interior of the clusters remained crystalline. Further heating resulted in the full melting of the clusters. Similar results have recently been reported for copper nanoparticles.[28] There remains some uncertainty with respect to the temperature dependence of gold clusters, with some more recent simulations apparently confirming the general conclusions reached by Landman, while other simulations produced conflicting results. Simulations by Wang et al. using a slightly different potential function indicate that clusters nucleated from a melt do form Mackay decahedral structures, but these clusters do not show a surface liquid below the bulk melting point.[29] On the other hand, Barnard et al. used first-principles data and a thermodynamic model to calculate the equilibrium shape of gold clusters in the size range from 3 to 100 nm.[30] The results of the calculations by Barnard et al. suggest that a truncated octahedron is the energetically favorable structure for clusters in this size range, in contrast to the two simulations that used analytic potentials. Barnard’s results do, however, support the concept that solid-solid transitions occur between different cluster morphologies prior to melting. High-resolution electron microscopy studies by Koga et al. of gold clusters in the range 3–14 nm do indeed indicate a structural transformation from an icosahedral to a decahedral morphology just below the melting point.[31] Recent simulation results for a nickel cluster by Schebarchov and Hendy have suggested an interesting variation on the structural transitions of metal clusters prior to melting.[32] They simulated a 1415-atom nickel

7: Computer Modeling, Brenner

301

cluster up to the melting temperature using an embedded-atom method potential. They report a two-phase system below the bulk melting point consisting of a crystalline core within a surface liquid. In these nickel simulations, however, there is a transition of the solid core from a partial icosahedral to a truncated decahedral structure that is almost fully wet by the melt. In this case the transition appears to be driven by a preference for the melt to wet the decahedral structure. Meyer et al. characterized the vibrational states of copper, nickel, and silver clusters, and compared these states to their bulk counterparts, as well as to nanostructured solids with different degrees of porosity.[33] Because of the size of the systems, rather than diagonalizing a force matrix to obtain vibrational frequencies, the vibrational states were calculated by Fourier-transforming the velocity autocorrelation function for atoms identified as belonging either to the core or to the surface of the cluster. Examination of the vibrational states for a 791-atom copper cluster with an underlying face-centered cubic lattice revealed striking differences in the vibrational states for the two types of atoms. For the core atoms, two separated but otherwise broad peaks were calculated, a low-frequency peak that corresponds to the transverse modes of the bulk crystal, and a higherfrequency peak associated with the longitudinal bulk vibrational modes. The longitudinal cluster mode was shifted to a slightly higher frequency compared to the bulk mode. The shift was attributed to a capillary pressure in the interior of the cluster from the large surface stress that results from the nanometer size scale. The shifted vibrational mode compares well with the vibrational states for a bulk crystal under a similar compressive stress, a result that essentially confirms the stress origin of the vibrational shift for the cluster. The surface atoms in the cluster have a much different set of vibrational states, with the low-frequency transverse states shifted to lower energies, and the longitudinal mode reduced to a weak shoulder in the density of states. The combined result is to increase the number of low-frequency modes in the vibrational density of states in the cluster relative to the bulk. Worth noting is that the loss of resolution of the vibrational states for the surface atoms compared to the core atoms correlates well with the decrease in structural order leading to premelting of cluster surfaces as discussed above. To probe a different set of properties induced by the nanometer scale, Vashishta and co-workers simulated the pressure-induced transformation of gallium arsenide nanocrystals from a zinc blende to a rock salt structure as a function cluster size and shape.[34] The transformation pressure was found to be very dependent on the size of the cluster, with the smaller clusters showing a lower transition pressure. The final structure of the

302

Processing

transformed clusters was also found to be dependent on cluster size. For clusters smaller than about 4.4 nm, only a single orientation of the rock salt lattice was observed. At larger cluster sizes, however, grain boundaries were observed between regions with different crystal orientations. Apparently, the transformation pressures and associated strains are not uniform throughout these larger clusters during the structural transition. Recent work by Vashishta and co-workers examined the detailed dynamics of the oxidation of aluminum nanoparticles.[35,36] They used a variable-charge interatomic interaction model that includes both ionic and covalent bonding contributions to model rapid oxidation of aluminum clusters that were 20 nm in diameter. With no mechanism for removing kinetic energy, the oxidizing reactions were explosive due to the enormous energy release associated with aluminum–oxygen bonding, and local stresses in the oxide caused rapid diffusion of aluminum and oxygen atoms (Fig. 7.1, top). In contrast, when using a thermostat to model a canonical ensemble, the simulation yields an oxide depth that initially grows linearly in time and ends at a saturation depth of about 40 Å formed after 466 ps (Fig. 7.1, bottom). By detailed analyses of the trajectories of oxygen atoms during reaction, a three-step process of oxidative percolation was proposed that explains deceleration of oxide growth in the canonical ensemble. Carbon nanostructures provide another excellent example of a system in which phase stability is strongly coupled to system size.[37] Bonding associated with sp2 and sp3 atomic hybridizations are very close in energy. At the macroscopic scale, where the system has translational symmetry, the roughly equal bonding energies result in the diamond and graphite phases being very close in stability. At the nanometer scale where threedimensional translational symmetry is not required, however, there are many structures with competing energies, including fullerenes, nanotubes, nanodiamond clusters, bucky onions, etc. Because the binding energies of each structure are quite close to one another, and because the ratio of surface to bulk atoms is very high, relatively subtle effects such as inequivalent surface stresses can lead to unanticipated changes in phase stability with system size. There have been several attempts to organize these configurations into classes of structures in terms of bonding and stability, and nanometer-scale equivalents to phase diagrams for these systems have been proposed.[37]

7.3.2 Microstructure Modeling The mechanisms associated with mechanical deformation gleaned from computer modeling are inherently dependent on the nanostructure used in

7: Computer Modeling, Brenner

303

Figure 7.1 Snapshots (at 50 ps) of variable-charge molecular dynamics simulations of the oxidation of an aluminum nanoparticle of radius 100 Å, where small and large spheres are aluminum and oxygen atoms, respectively. The top panel shows a result in the microcanonical ensemble, where color represents temperature. The bottom panel shows a result in the canonical ensemble at temperature 400 K, where color represents atomic charge. Only 1/8 of the nanoparticle is shown. Results are from Refs. 35 and 36. (Figure courtesy of R. Kalia, A. Nakano, and P. Vashishta.)

a simulation. At present, however, there is some disagreement over the appropriate nanostructures for these materials, as well as several different approaches to defining atomic positions with a chosen nanostructure. Some of the methods for generating atomic positions are based on geometrical constraints imposed by the simulation conditions, for example, ensuring that active slip systems are appropriately oriented with respect

304

Processing

to periodic boundaries. Other methods use random grain centers, with grain boundaries chosen based on a Voronoi construction. Variations on this method include picking grain orientations to produce a particular range of tilt angle (e.g., low-angle grains), or picking grain centers to produce a log-normal distribution of grain sizes. Crystallization dynamics have also been used to generate nanostructures. These methods can be based on a Johnson–Mehl or a Potts model construction, both of which produce a log-normal grain size distribution, or by using a molecular dynamics simulation to model crystallization from a melt. Still other researchers have generated nanostructures by simulating compaction and sintering of nanoclusters. Wolf and co-workers have performed extensive studies generating and characterizing grain boundary structures using molecular dynamics simulations.[3,38,39] Simulations of both covalent and metallic materials were carried out in which grains were grown from crystalline seeds that were embedded into a supercooled melt. The grain boundaries produced in this way are more disordered than is typically seen for more traditional grain sizes, and apparently have more in common with clusters described above that form a disordered (liquid) overlayer prior to melting. Based on annealing studies using molecular dynamics simulations, Wolf and coworkers have argued that for randomly oriented nanometer-sized grains these structures are the thermodynamically stable configurations and are not metastable products of the short simulation time.[3] They have also argued based on structural analysis that the grain boundary configurations are very similar to those found in the bulk homogeneous amorphous structures of these materials.[3] Van Swygenhoven,[4] Jacobsen,[40] and others have generated initial structures by first choosing grain centers and crystalline orientations, much like the seeds in the molecular dynamics simulations. However, rather than simulating grain growth, a Voronoi construction is used to define the positions of the grain boundaries. The grain boundary structures are then refined through energy considerations using some assumed interatomic potential. This is typically accomplished by removing or rearranging atoms with high potential energies, by annealing the samples, or both. This procedure tends to produce more ordered grain boundary structures than the dynamic simulations that appear closer to the established structures in conventional grain boundaries (Fig. 7.2, left). While convenient for creating initial structures, it is well known that Voronoi constructions produce microstructures that are inconsistent with experimental observations for materials with conventional grain sizes.[41] For example, for a small number of grains Voronoi cells satisfy a Poisson

7: Computer Modeling, Brenner

305

Figure 7.2 Illustrations of a columnar nanostructure (left) and a fully-dense three-dimensional nanostructure (right).

distribution, while grains created by nucleation satisfy a log-normal distribution. In addition, comparisons of grain shapes produced by Voronoi constructions with experimental structures in conventional materials suggest that the density of small facets defining the Voronoi cells is too large compared to experiment. While annealing may create interface shapes that are more consistent with experiment, at present it is unclear whether this construction produces grain shapes consistent with experimental nanostructured materials, or even whether these issues play a major role in the mechanical properties of these structures. To overcome limitations associated with Voronoi constructions, Zheng et al. used a Monte Carlo solution to a Pott’s model for total energy to create an initial microstructure.[41] Input into the energy function includes grain orientations, interface and volumetric energy functions, and anisotropies from impurities and second phases. Ignoring the latter (i.e., assuming a single-phase material) and assuming a constant interfacial and volumetric energy, this method produces a microstructure that has a lognormal grain distribution. Furthermore, more complicated functions in the Pott’s model are capable of creating more complex structures. To better understand the results of deformation simulations, methods for generating nanostructures have been used to produce specific types and

306

Processing

orientations of interfaces. Van Swygenhoven and co-workers, for example, have produced initial nanostructures with primarily either low- or highangle grain boundaries by carefully choosing the crystal orientations within the Voronoi cells.[42] Similarly, Wolf and co-workers have explored the creation of low-energy twins from their crystallization simulations by carefully orienting the lattices of two seed grains.[39] They report that while the correct angle was produced, the lateral distance between grains needed to produce the correct twin structure did not occur in the simulations. Columnar nanostructures have also been created such that the slip planes and directions are properly oriented with respect to the direction of deformation and the periodic boundaries so that the restricted system sizes do not induce size- and orientation-dependent artifacts in the simulations.[43] Creation of nanostructured solids via molecular dynamics simulation of particle compaction has also been carried out. Zhang et al., for example, modeled the creation of fully dense nanocrystalline copper by compaction of copper nanoparticles.[44] Meyer et al. created nanocrystalline structures of copper, nickel, or silver with different pore fractions, starting with randomly positioned nanoclusters compacted under various combinations of temperature and applied pressure.[33] These structures were used to probe the relative roles played by core, surface, and grain boundary atoms in determining the vibrational states in the solid, as discussed below. Vashishta and co-workers have generated nanostructures by simulating the compaction and annealing of gallium arsenide and silicon carbide nanocrystallites.[45,46] In the latter, the simulations produced disordered grain boundary structures similar to those in Wolf’s studies mentioned above. The dynamics of the compaction and annealing simulations are discussed further in the next section. A critical assessment regarding which of the proposed methods for creating atomic positions results in the “best” nanostructure is difficult for several reasons. First, there are multiple experimental methods for creating nanostructured materials, each of which can produce a different type of structure. Hence discussions regarding which configurations are the true equilibrium structures may be irrelevant when trying to compare with experiment. Similarly, nanostructured materials can undergo relaxation at multiple timescales, including changes in structure and properties that can take place over days (or longer). None of the computational methods described above can reproduce long-time relaxation. Finally, idealized structures, for example, with grains at particular angles or slip systems aligned in particular directions, may be inaccessible experimentally but are nonetheless important because they greatly facilitate identifying underlying mechanisms for mechanical deformation that would otherwise

7: Computer Modeling, Brenner

307

be very difficult to assess in more complicated nanostructures. There is clearly room within this field for multiple “correct” structures.

7.3.3 Sintering and Grain Growth Dynamics Much of the simulation work done on sintering and grain growth dynamics–and indeed many of the simulation studies in general on nanostructured materials–have focused on understanding how processes and rates at the nanometer scale differ from their counterparts in macroscopicscale systems. In macroscopic-scale sintering theory, six distinct mechanisms contribute to the sintering dynamics of crystalline particles: surface diffusion; lattice diffusion from the surface, from grain boundaries, and through dislocations; vapor transport; and grain boundary diffusion. At the nanometer scale, there is a much larger degree of surface curvature and a much higher ratio of interface to bulk atoms (and hence of interface energy to bulk energy), both of which may lead to very different sintering mechanisms. The details regarding these differences, and some new and unexpected effects at the nanometer scale, are just beginning to be understood from computer simulations. In a fairly early set of studies, Zeng et al. used molecular dynamics simulations to study surface energies, grain boundary mobility, and sintering of copper and gold nanoparticle arrays at different temperatures.[47] The results suggested that of the macroscopic-scale mechanisms associated with sintering listed above, only surface and grain boundary diffusion contribute significantly to nanometer-scale sintering dynamics, and that these two processes are accelerated due to the large interfacial forces. In addition, the simulations showed three unconventional mechanisms that contribute to the early stages of nanometer-scale sintering: mechanical rotation, plastic deformation via dislocation generation and transmission, and amorphization of subcritical grains. Taken together, these results suggest that the kinetics of sintering (at least in ductile metals) should be over six orders of magnitude more rapid than the rate estimated by standard sintering theory, and could even proceed fairly rapidly at room temperature in many materials. The compaction and sintering of arrays of nanometer-scale copper particles were modeled by Zhang et al. using molecular dynamics simulations and an embedded-atom potential.[44] The initial particles ranged in size from 3.25 to 8.40 nm. Each particle was assigned a random rotation angle and then placed on either a face-centered cubic, body-centered cubic, or simple cubic lattice, or arranged randomly within a cubic

308

Processing

supercell. The particles were then compacted to the volume associated with a bulk crystal at temperatures of 300, 700 or 1200 K. After sintering, the pressure was reduced and the sample was cooled to 0 K. Several interesting effects were characterized from the simulations. First, it was observed that smaller particles were more deformed than larger particles for the same level of compaction. For the larger particles, the grain interior remained crystalline, while the surface atoms became amorphous. For smaller particles, the ratio of surface to bulk atoms is higher, and hence the surface disorder represents a larger fraction of the particle. Second, the degree of crystallinity was lowest when the particles were initially on a simple cubic lattice, and increased from the body-centered to the facecentered cubic initial arrangement of particles. The simple-cubic arrangement had the largest initial amount of free space, and therefore during compaction those nanoparticles must deform to fill the space to a greater degree than the higher-density initial packings. Third, it was observed that for the same particle sizes and arrangement, compaction at the higher temperatures resulted in more disorder in the final system. Finally, it was observed that larger particles required higher applied stresses to compact the structure to the bulk density than did systems with smaller particles. Further discussion of the differences in mechanical properties of these simulated sintered systems is given in the next section. Meyer et al. simulated the creation of nanostructured copper, silver, and nickel with different degrees of porosity by using different combinations of sintering temperature and pressure to compact nanoclusters that were initially placed in random positions.[33] After compaction without an applied pressure, both the copper and nickel produced a relative density of 77%, while the silver had a lower density of 70%. In contrast, after applying and releasing a sintering pressure of 1.7 GPa, silver had the highest density (98% of the bulk density), while copper and nickel had lower relative densities of 0.96 and 0.92, respectively. These results can be understood by comparing the surface and stacking fault energies for the three elements. The surface energy is higher for copper and nickel than it is for silver, and therefore in the initial stages of sintering, which is driven by the reduction of free surfaces, silver forms a more porous structure. Compaction, however, involves plastic deformation that includes dislocation motion and formation of stacking faults. Of the three elements, silver has the smallest stacking fault energy, followed by copper and then nickel. Silver grains can therefore more easily deform to fill the voids between grains, resulting in a more dense material. Similarly, plastic deformation of copper grains occurs more readily than does deformation of nickel, resulting in a higher density for copper than for nickel.

7: Computer Modeling, Brenner

309

Simulations of the sintering and compaction of nanometer-scale ceramics have also been carried out. In a joint simulation and experimental neutron scattering study, Chatterjee et al. characterized the sintering properties of silicon carbide nanoparticles that were about 3 nm in size.[46] Both the neutron scattering data and the atomic simulations indicated onset of sintering at about 1500 K, which is considerably lower than the value of ∼2200 K for a coarse-grained material. The simulations showed disordered interfaces due to the high-temperature sintering, and a much higher diffusion rate for atoms at the grains compared to those in the interior of the crystals. Hence, surface diffusion rather than bulk diffusion dominates the sintering process. This is in contrast to the metallic systems discussed above, where dislocation generation and transmission as well as grain rotation contributed to the sintering process. Koparde and Cummings have recently used molecular dynamics to model the sintering dynamics of two colliding titanium dioxide nanoparticles as a function of phase, size, and temperature.[48] Titanium dioxide nanoparticles have several applications, including as a white dye pigment, a raw material for electronic and structural ceramics, and as a catalyst under ultraviolet light. In nature, titanium dioxide is found in several distinct structures–anatase and rutile, which are both tetragonal, and brookite, which has an orthorhombic structure. Only the tetragonal structures were considered in the simulations. In the initial stage of sintering a “neck” was formed between the two spheres upon first contact within the first few picoseconds. As the simulation progressed, the neck increased in width and further particle interpenetration occurred. However, full coalescence of the particles was not observed in the simulations, although the authors speculate that this would occur at longer times in a second stage of sintering. Accompanying the sintering dynamics was an increase in temperature due to the formation of new bonds within the neck, with a larger increase in temperature seen for smaller particles. For particles in the anatase structure, higher initial temperatures resulted in larger neck sizes and closer particle distances. This temperature effect was not seen if the two particles were initially in the rutile structure. At all temperatures and for all structures the titanium ions in the neck region were more mobile than those in the crystalline interior of the particles, consistent with the studies discussed above. The authors also report a large dependence of neck size and interparticle distance after sintering on the initial relative orientation of the particles. This is apparently a result of a dipole interaction between the two particles that results from their nanometer-scale dimensions and crystallinity (as well as possibly the fixed atom-centered partial charges assumed in the potential energy function).

310

Processing

In a related study, Wolf and co-workers simulated the dynamics of grain growth in nanocrystalline fcc metals assuming a columnar structure and grain sizes of about 15 nm.[49] In agreement with the conclusions of Zeng et al.,[47] these simulations indicate that grain rotation can play a role in grain growth that is as equally important as grain boundary migration. The simulations predict that necessary changes in the grain shape during grain rotation in the nanostructure can be accommodated by diffusion either through the grain boundaries or through the grain interior. Based on this result, it was suggested that both mechanisms, which can be coupled,[50] should be accounted for in mesoscopic models of grain growth. Moldovan et al. have recently reported the existence of a critical length scale in the system that enables the growth process to be characterized by two regimes.[51] If the average grain size is smaller than the critical length, as in the case of nanocrystals, grain growth is dominated by the grain-rotation coalescence mechanism. For average grain sizes exceeding the critical size, the growth mechanism is due to grain boundary migration. Building on the two-dimensional simulation results, Moldovan et al. developed a new model for grain growth at the nanometer scale that involves exclusively a grain rotation coalescence mechanism.[52] An analytic analysis of the model, together with extensive mesoscopic scale simulations, demonstrated that this grain growth mechanism leads to a power-law growth rate, similar to traditional grain growth dynamics but having a universal scaling exponent that depends only on the manner in which grain rotations are accommodated. They also show that the grains obey a log-normal distribution as grain growth occurs. This model has in turn been used as the basis for a mesoscopic scale model for grain growth as grain sizes increase.[53] This bridging from dynamics gleaned from atomic simulations to mesoscopic-scale grain growth simulations is an excellent example of serial multiscale modeling. To help confirm the role of grain rotation in fully three-dimensional grain growth, Xiao and Hu recently simulated grain growth in nanocrystalline aluminum using molecular dynamics simulations.[54] The microstructure was generated using a Voronoi construction, and the interatomic forces were modeled with a variation of the embedded-atom method. The simulations gave an initial power-law dependence for grain growth, followed by a linear relaxation process. Analysis of the simulations suggested that the initial power-law growth was a result of grain rotation and grain boundary migration, accompanied by an increase in dislocation density. In the linear relaxation regime, the dislocations (and associated stacking faults) are gradually eliminated.

7: Computer Modeling, Brenner

311

Two general conclusions emerge from these atomic-level studies of the compaction and sintering of nanoparticles that appear to be a result of particle size. First, the effective sintering temperature for nanoparticles appears to be much lower than for larger particles. This difference can be traced directly to the scale of the system, where there is a large interface region that results in more bond formation and hence energy release during sintering, and a decrease in the size of the crystalline interior, which reduces the degrees of freedom that can absorb this energy. Second, mobilities of the atoms in the interface regions are much greater than those in the bulk crystal, which also contributes to a much faster sintering rate for nanoparticles. Finally, particle rotation appears to play a much larger role in the sintering of nanoparticles than it does for larger-dimension particles.

7.3.4 Mechanical Deformation and Fracture In conventional metals, plastic deformation results from the motion of dislocations. This motion can be inhibited by grain boundaries, which leads to the Hall–Petch relation that the yield strength is proportional to the inverse square root of the grain size. There is a threshold grain size, however, where materials begin to get softer with decreasing grain size. This so-called “inverse Hall–Petch” behavior has been attributed to a transition from dislocation-mediated plastic deformation to grain boundary sliding for some critical grain size. This transition has been supported by several sets of molecular dynamics simulations that predict a transition grain size of about 10–15 nm, in good agreement with experimental results. At the same time, the simulations have also revealed a rich and unanticipated set of dynamics near the threshold region that can be related to the fundamental properties of the bulk materials. These unique dynamics include an enhanced role of grain rotation (as in the sintering studies discussed above), cooperative intergrain dynamics, and formation of stacking faults via motion of partial dislocations across grains. Jacobsen and co-workers simulated the deformation of strained nanocrystalline copper with grain sizes that average about 5 nm.[40] These simulations showed material softening for small grain sizes, in agreement with experimental measurements. The simulations confirm that plastic deformation in the inverse Hall–Petch region occurs mainly by grain boundary sliding, with a minimal influence of dislocation motion on the deformation. In related simulations, Van Swygenhoven and co-workers performed a series of large-scale molecular dynamics simulations of the

312

Processing

deformation of nanostructured nickel and copper with grain sizes ranging from 3.5 nm to 12 nm.[55] For grain sizes less than about 10 nm, deformation occurred primarily by grain boundary sliding, with the rate of deformation increasing with decreasing grain size. For the larger grain sizes, deformation occurred by a combination of dislocation motion and grain boundary sliding. In subsequent simulations, detailed mechanisms of strain accommodation were characterized that included both single-atom motion and correlated motion of several atoms, as well as stress-assisted free volume migration.[56] Wolf and co-workers carried out detailed studies of the deformation of columnar structures of aluminum, where emission of partial dislocations that were formed at grain boundaries and triple junctions was observed during deformation.[57] The simulations also showed that these structures can be reabsorbed upon removal of the applied stress, which the authors suggest may contribute to the fact that dislocations are not normally observed experimentally in systems of this type after external stresses are released. Near the grain size at which plastic deformation transitions from dislocation-mediated plasticity to grain boundary sliding, the motion of single partial dislocations across nanograins during tensile loading is commonly observed in simulations. Without emission of a trailing partial dislocation, an intrinsic stacking fault is formed along the width of the nanograin. These dynamics are illustrated in Fig. 7.3 for a columnar aluminum nanostructure, where the arrows indicate the formation of the stacking faults. Also illustrated in Fig. 7.3 is the emission of full dislocations from the grain boundaries (this occurs in the region between the arrows). Apparently the formation of a low-energy ordered grain boundary drives the emission of a full dislocation and the resulting absence of a stacking fault. Van Swygenhoven and co-workers have argued that nucleation of the initial partial dislocation and the atomic rearrangement at the grain boundary associated with its emission sufficiently lowers the grain boundary energy such that emission of the trailing partial dislocation is not always needed to further relax the system.[58] Based on simulations of aluminum with a columnar nanostructure, Yamakov et al. suggested that the stacking fault width, and hence the intrinsic stacking fault energy, as defined by the distance between two partial dislocations is the central quantity that defines the transition from full to partial dislocation emission as grain sizes approach the critical size for the onset of inverse Hall–Petch behavior.[59] In subsequent work Van Swygenhoven and co-workers pointed out that the relation between the emission of partial dislocations does not correlate well with calculated stacking fault energies for nickel, copper, and

7: Computer Modeling, Brenner

313

Figure 7.3 Illustrative snapshots from a simulation of strained nanocrystalline aluminum with a columnar nanostructure. Only atoms with local bonding symmetry different from the bulk structure are shown. Strain increases from the top to the bottom frame. Arrows indicate emission of partial dislocations and resulting formation of stacking faults.

314

Processing

aluminum.[58]58 Instead, they have argued that full dynamics associated with the nucleation of a partial dislocation from a grain boundary should be considered, and therefore that the full planar fault energy, which includes the stable and unstable stacking fault energy as well as twin fault energies, must be used to understand and ultimately predict the relation between mechanical deformation and grain size. More recently Farkas and Curtin have examined detailed mechanisms for dislocation emission from a columnar nickel nanostructure with grain sizes ranging from 4 to 20 nm.[60] They report that dislocations are emitted from preexisting dislocation-like structures that are inherent to the grain boundaries, and that the number of emitted dislocations per unit length of grain boundary saturates as grain sizes increase. By assuming that dislocations cannot be emitted some small distance from triple junctions, they are able to reproduce the dislocation densities observed in the simulations. It should be noted that further complicating all of these observations in simulations and related models is the timescale of the mechanisms needed to nucleate dislocation emission relative to the short timescale of the simulations. Clearly much more work is needed to fully understand this transition region between normal and inverse Hall–Petch behavior and its relationship to material properties, the structure of the grain boundaries, the size of the grains, and the strain rate. Recent experiments have revealed the formation of “dimples” on the fracture surfaces of nanocrystalline metals with grain sizes less than about 100 nm.[61] These dimples appear for fully compact materials and therefore are not the result of material artifacts such as the presence of voids. The scale of the dimples is typically of the size of multiple grains, which suggests a fracture mechanism involving some form of cooperative motion between grains. While the deformation mechanisms of nanostructured metals are becoming better understood in terms of the competition between grain boundary sliding and the emission of partial and full dislocations as discussed above, the nature of cooperative motion leading to the dimple formation is not clear. Using large-scale molecular dynamics simulations, Hasnaoui et al. studied the deformation of nanostructured nickel containing 125 randomly oriented grains with a narrow grain size distribution and mean grain size of 6 nm.[62] To enhance the dynamics of grain boundary sliding and diffusion at the limited timescale of the simulation, the simulation was carried out at 0.45 of the melting temperature of pure nickel as given by the interatomic potential used in the simulation. A tensile stress of 1.5 GPa was applied to the sample for a total of 350 ps. Cooperative intergranular motion was observed in the form of local shear bands that extend through multiple grains. Three mechanisms

7: Computer Modeling, Brenner

315

were reported that contributed to the formation of these shear bands: grain migration via boundary sliding that lined up complementary shear planes; reorientation of neighboring grains separated by low-angle grain boundaries; and intragranular slip. Distributed within the simulated microstructure were special grain boundary structures such as twins that are particularly resistant to sliding. These structures create pinning points around which surrounding grains deform and form shear bands. It was proposed that it is this cooperative motion between grains that leads to the dimples observed experimentally. Farkas and co-workers have carried out extensive simulations of crack propagation in nanostructured metals to better understand their fracture, fatigue, and toughness properties.[63–65] These simulations have revealed crack propagation mechanisms that have features similar to the plastic response of fully dense samples as discussed above, as well as some key differences that result from the presence of the crack tip. For example, in simulations of nanocrystalline nickel with a mean grain size of 10 nm, mode I crack propagation occurred by intergrain decohesion via a mechanism involving coalescence of nanovoids that form in front of the crack tip.[63] In addition, plastic deformations leading to both partial and full dislocations were observed in the neighboring grains. As discussed above, full dislocations are not generally observed in simulations of the tensile loading of nanostructured solids that do not contain cracks, a result that has been attributed to a sufficient stress release by the leading partial dislocation that the emission of the trailing dislocation is not needed to relax the system. In the case of crack propagation, the formation of the nanovoids in front of the crack tip appears to alter the stress balance in the sample such that trailing dislocations are emitted. Similarly, twinning of the sample associated with crack propagation was also observed that is apparently not a major deformation mode for tensile loading crack-free nanostructured metals. While a majority of studies have been on nanostructured metals, there have been a few studies on ceramics. In work by Szlufarska, Nakano, and Vashishta, for example, molecular dynamics simulations were used to examine the dependence of deformation mechanisms on the grain size of nanostructured silicon carbide during nano-indentation.[66] In these studies the grain structure was created by simulated sintering of silicon carbide nanocrystallites. The resulting structure was composed of crystalline nanograins surrounded by amorphous grain boundaries. This structure reportedly reproduces neutron scattering data of experimentally sintered nanocrystalline silicon carbide. This nanostructure is similar to a two-phase system, where the soft amorphous phase at the grain

316

Processing

boundaries plays a large role in determining the toughness of the material, while the harder crystalline phase determines the yield strength. Correlating the atomic dynamics with the load versus displacement relations from the simulations revealed four characteristic regimes. The dynamics at initial contact is purely elastic. At higher loads, plastic flow of the amorphous phase produces a small change in the slope of the loading curve and a small hysteresis between the loading and unloading curves. Because the amorphous phase shows only a small amount of yielding, the grains move in a concerted manner, with an elastic deformation zone that extends beyond the region directly below the indenter. This observation of a cooperative motion of the grains is related to the model described above for forming dimples in fractured nanocrystalline metals. At still higher loading, the amorphous grain boundaries in the silicon carbide show significant plastic yielding, and the motions of the grains become decoupled from one another. In the final loading regime, plastic damage occurs within the grains directly below the indenter, and individual events such as dislocation glide within a grain are apparent as a fine structure in the simulated load versus displacement curves. This simulation is an excellent example of how grains within a nanostructured solid can respond both in a cooperative manner as well as on an individual basis depending on the structure and degree of mechanical deformation.

7.3.5 Shock Loading Compared to their tensile and indentation mechanical properties, at present much less is understood about the influence of grain size on the shock loading properties of nanostructured solids. The simulations that have been carried out, however, suggest a strong coupling between nanostructure and shock loading dynamics, as well as unique and very important properties of shocked-loaded nanostructured metals. Kadau et al. have used large-scale molecular dynamics simulations to characterize phase transformations in shock-loaded single-crystal iron as a function of pressure and crystal orientation.[67] Upon increasing temperature or pressure, iron will undergo a phase transition from a bodycentered cubic structure to a more close-packed structure. The transition between the two structures involves collective motion by which atoms typically travel less than a near-neighbor spacing. There are distinct crystallographic relations between the two phases that can lead to the formation of defects such as twins and stacking faults depending on the

7: Computer Modeling, Brenner

317

orientation of the shock loading, the rate of the pressure and temperature rise, and any preexisting defects such as grain boundaries. Under shock loading, both the temperature and pressure are increased at an extremely high rate, and therefore the transition dynamics and resulting structures near the shock front are difficult to probe experimentally. Fortunately, the timescales of the pressure and temperature rise are the same as in typical molecular dynamics simulations, and therefore this modeling method can provide important insight into the dynamics of shock-loaded solids. In the simulations of Kadau et al. at shock velocities below that needed to induce the transition, an elastic transformation between the unshocked and compressed body-centered cubic structure was observed. At shock velocities that produce a front pressure just above the transition pressure, the simulations show a split wave structure in which the first wave induces an elastic response in the initial structure and a second, slower, wave results from the phase transformation without any intermediate elastic-plastic deformation. The authors note that the lack of an intermediate plastic wave may be due to the initially perfect structure of the sample compared to experimental systems that likely contain defects. The shear stress at the phase transformation is relieved by the homogeneous formation of closepacked nanometer-scale grains. Because the transition front is established by the formation of only a few grain nucleation centers, the transformation front is rough compared to similar shock loading conditions in solids without the transition. Hence there is a strong relation between formation of transient nanometer-scale grains and the structure of the shock front. At higher shock pressures (induced by a faster driving piston velocity), the driving force for the structural transition increases, which results in a smoother transition front. At very high shock strengths, the elastic precursor and transition fronts merge, leading to a single narrow shock front at which the transition occurs. Bringa et al. have noted that the mechanisms associated with the mechanical deformation of nanostructured metals depend strongly on temperature, pressure, and strain rate, and therefore these materials may show ultrahigh strength under shock loading depending on the system and shock loading conditions.[68] The fast temperature and pressure rises associated with shock fronts freezes out deformation mechanisms that require diffusion. Similarly, production of dislocations that requires nucleating events is inhibited. In the case of grain boundary accommodation, increasing the pressure results in an increase in the threshold stress for sliding plasticity. The threshold for dislocation plasticity increases with increasing pressure because of an increase in the shear modulus with increasing pressure. Taking these effects into account, and assuming that the maximum in

318

Processing

hardness as a function of grain size occurs where stress for sliding and dislocation plasticity are equal, Bringa et al. were able to show that ultrahigh hardness can be achieved by shock loading of nanostructured solids. To validate their analysis, and to characterize the details of these competing effects, Bringa et al used large-scale molecular dynamics simulations to model the shock loading of nanostructured copper with different grain sizes. At relatively low shock velocities, and hence low stresses, grain boundary sliding is observed, which results in a relatively low hardness value that increases with increasing grain size. At intermediate stresses, the hardness of the copper increases with increasing shock strength for all grain sizes, with a shift in the maximum hardness toward lower grain sizes compared to deformation at lower strain rates. This leads to an overall increase in the maximum hardness of the material. At higher stresses, however, the simulations predict a drop in strength due to an increase in temperature and an associated increase in dislocation nucleation and motion. At the highest shock pressures studied, 150–220 GPa, shock-induced melting was observed with an associated small value of shear stress and a further reduction in hardness. At the intermediate shock pressures, the authors note that dislocation emission occurs even for very small grain sizes, with dislocation emission occurring from both triple junctions and grain boundaries. Emission of dislocations from triple junctions under shock loading is contrary to the conclusions drawn by Farkas and Curtin based on their studies of the deformation of a columnar microstructure at low loading rates as discussed above. The shock studies by Bringa et al. also report the formation of defects such as stacking faults, which indicate emission of partial dislocations, nanotwins, and defect clusters with sizes of the order of 1 nm.

7.3.6 Vibrational Properties Understanding the influence of grain sizes, pore structure, and grain boundaries on the vibrational properties of nanocrystalline solids compared to those of bulk materials is important because the vibrational states contribute to thermodynamic properties such as the heat capacity. Similarly, it is well established that grain boundaries induce phonon scattering and therefore reduce thermal transport in conventional materials. In principle, the high density of grains in nanostructured solids could result in very low thermal conductivities compared to materials with conventional grain sizes, although relatively little has been done in this area to fully understand such effects.

7: Computer Modeling, Brenner

319

An enhancement of both the low- and high-frequency vibrational modes in nanostructured metals compared to their bulk counterparts has been measured with neutron scattering.[69] To identify the origin of this result, as well as to better understand the vibrational (and hence thermodynamic) properties of nanostructured metals in general, several groups have used molecular dynamics simulations to characterize the vibrational states of these systems. In work by Derlet et al.,[70] as well as by Meyer et al.,[33] vibrational states of nanostructured copper, nickel, and silver were characterized by Fourier-transforming the velocity autocorrelation function for sets of atoms identified as belonging to grains or grain boundaries. In all cases the vibrational states within the core of the grains were almost identical to bulk states for the same potential function. Two peaks were apparent, a low-frequency peak that corresponds to the transverse crystalline modes, and a higher frequency peak that corresponds to the longitudinal modes. In contrast to the grain interiors, the vibrational states of the atoms associated with the grain boundaries showed a broad and largely structureless distribution that extends to both lower and higher frequencies compared to the grain interior vibrations. Hence, the simulations suggest that the enhancements in the low- and highfrequency modes observed experimentally are both due to the grain boundaries. In the simulations by Meyer et al. two different packing densities were used for copper, 89% and 96% relative compaction. Both samples had almost identical vibrational states, further suggesting that surface states associated with small pores do not play a large role in determining the vibrational properties of these systems. An interesting observation made by Meyer et al. is that the vibrational states within the grains are not shifted as they are in isolated clusters (see above) despite a slight compressive stress within the grains. The authors suggest that this may be due to the extension of the vibrational states between grains that apparently cancels the frequency shift.

7.3.7 Nanoalloys Although very much in the initial stages of understanding, recent simulation results on nanostructured alloys have suggested rich and potentially important dynamics in nanostructured alloys. At present there are two issues that are of particular importance. The first is the influence of impurity atoms on the mechanisms and properties of dislocation emission from grain boundaries in nanostructured metals. As discussed in more

320

Processing

detail below, there are preliminary results that lead impurities in nanostructured aluminum with grain sizes larger than the transition to inverseHall–Petch behavior may raise the applied stress needed for partial dislocation emission into the nanograins and therefore strengthen the material relative to pure nanostructured metal. The other issue is the influence of impurities at the grain boundaries on grain growth, in particular whether these impurities can suppress grain growth and thereby stabilize the nanostructure. The high curvature of grain boundaries together with the additional grain growth mechanism of grain rotation discussed above for nanostructured systems makes them especially susceptible to grain growth. However, if the specific grain boundary energy could be reduced to zero (or less), grain growth could be suppressed and nanostructures thermodynamically stabilized.[71] This can in principle be accomplished if the impurity atoms have a large positive heat of segregation provided that the formation of a second phase involving the impurities can be suppressed. Millett et al. have used molecular dynamics simulations to characterize the stabilization of an idealized nanocrystalline metal intended to mimic copper against grain growth by the addition of impurities as a function of impurity concentration and the degree of radius mismatch between the metal and impurity atoms.[71] Because this was a parametric study, relatively simple pair-additive Lennard–Jones interactions were used to model the interatomic forces. Although much less accurate for describing material properties than many-body potentials such as the embedded-atom method, they have the advantage that the properties of the impurities can be varied continuously in a straightforward manner. For these simulations, the segregation energy was changed by changing the radius of the impurities. For the material annealed at 800 K with no impurities and an average grain size less than 10 nm, grain growth by both rotationcoalescence (as described above) and grain boundary diffusion was initially observed. As the grain sizes started to grow larger than 10 nm, rotation was no longer observed, and grain growth occurred via grain boundary motion. To study the influence of impurities, similar simulations were carried out in which impurities with concentrations between 0.1 and 2 at% were initially placed evenly along the grains. For the smallest concentration, the grain sizes at the end of the simulation were similar to those for the pure system. For the highest concentrations, however, the grain sizes did not change, and therefore the impurities inhibited grain growth. For the intermediate concentrations, the simulations showed different degrees of coarsening for the same total simulation time. By examining the variation of the enthalpy of the alloy with impurity size and concen-

7: Computer Modeling, Brenner

321

tration, they were able to confirm that the system was thermodynamically stabilized against grain growth. Preliminary simulations are being carried out of lead impurities in aluminum to better understand their potential influence on the mechanical and grain growth properties of this system.[72] Lead alloying is often used to enhance the machinability of bulk steels and brass. In bulk aluminum, however, lead is largely immiscible, and it is well established that lead impurities in aluminum segregate to form clusters. Recent experimental studies have indicated that lead impurities segregate to grain boundaries in nanocrystalline aluminum, and that a very small amount of lead, less than 1 at%, can have a strong influence on the mechanical properties of nanostructured aluminum.[73] Whether lead stabilizes this system against grain growth has not been firmly established, nor are the specific details of how lead alters the mechanical properties of this system well understood at this time. The preliminary simulations have used a modified embedded-atom method potential to describe the interatomic interactions within the lead–aluminum alloys.[74] Because of the immiscibility of lead in aluminum, little experimental data is available on alloy structures and their properties. Because of this, the potential function was fit to lattice constants, cohesive energies, and elastic constants for several hypothetical alloys that were calculated using first-principles methods. To characterize the segregation properties of lead in aluminum, Monte Carlo simulations were carried out in which moves consisted of both small displacements of individual atoms, and the “switching” of atom identities between lead and aluminum. This model predicts clustering of lead impurities in a bulk sample, as seen experimentally (Fig. 7.4). In the presence of the grain boundary, however, the lead disperses along the boundary, with no indication of clustering (Fig. 7.4). Analysis of the stresses along the grain boundary suggests that the dispersion of lead is stress driven. Because the lead atoms are larger than the aluminum atoms, they tend to segregate to sites that are under tensile hydrostatic stress. Similar simulations using columnar and three-dimensional fully-dense nanostructures have shown the same result (Fig. 7.4, bottom panels) that lead impurities are welldispersed to the grain boundaries, and that dispersion reduces the grain boundary energies, suggesting that lead might inhibit grain growth in this system. Simulations of grain growth that are similar in spirit to those done by Millett et al. on model Lennard–Jones systems discussed above are currently being carried out.[71] The effect of these impurities on grain boundary sliding and dislocation emission in strained nanocrystalline aluminum is also being explored with molecular simulations.

322

Processing

Figure 7.4 Illustrations from a Monte Carlo simulation of lead impurities in aluminum. The lead atoms are shaded lighter. Top left: 1 at% lead in bulk aluminum. A tendency for clustering of lead impurities is apparent from the figure. Top right: Bulk system containing a tilt grain boundary. Bottom left: The lead impurities in a fully-dense three-dimensional aluminum nanocrystal. Bottom right: The lead impurities in an aluminum nanocrystal with a columnar nanostructure. In all cases with grain boundaries the lead impurities segregate to and wet the interfaces.

7.4 Prospects for Future Modeling It is clear from the discussion in the preceding sections that atomic simulations are providing new and exciting insights into the unique properties of nanosystems in general and nanostructured solids in particular. Despite this success, however, there are several areas where improved modeling methods are needed to increase the depth of the understanding and the quality of the results coming from these simulations. A first obvious issue is the timescale of the molecular dynamics simulations. It is clear from the modeling studies that plastic deformation via

7: Computer Modeling, Brenner

323

emission of dislocations may require a nucleating event, and that grain boundary accommodation processes may require diffusion. Therefore, these two processes are potentially very sensitive to strain rate. Furthermore, these processes may in some instances require timescales that are far longer than is practical for a simple convergence study of results as a function of strain rate using molecular dynamics simulations. Approaches such as the hyperdynamics method described above hold promise in bridging the time gap. However, the current scaling between the number of possible rare events that can be modeled and the time boost factor is such that for multiple possible diffusion and nucleation events in a given nanostructure the boost factor is negligible. Just as the problem of producing thermodynamic quantities from molecular dynamics that correspond to ensembles other than the microcanonical ensemble was solved by Nosè after many approaches were explored by others, a more practical and general solution to the timescale problem for complex events will likely evolve over the next decade. A second issue whose solution seems more straightforward is the use of more accurate and transferable interatomic interactions in the simulations. It is apparent from the discussion above that, while some results are qualitative in nature and are therefore largely independent of the details of the interatomic interactions, other results require accurate energies for a range of materials properties. The stacking fault energies and the critical grain size for emission of partial dislocations is one obvious relation, but other more subtle relations likely exist that are not yet apparent. While the quality of analytic potential energy functions for describing energies over a wide range of configurations has risen significantly over the last two decades, the transferability of these functions to configurations well outside of those in the fitting database remains a cause for concern. An obvious solution is to derive forces from first principles as in the Car– Parrinello method,[75] but these methods are computationally very expensive, and the results are still prone to approximations used in the basis set, the pseudopotential, and the form of the density functional that can still lead to unpredictable results. The good news is that unlike the timescale problem, which in a straightforward solution requires fast processors, the calculation of first-principles forces can be made much more practicable by increasing the number of processors. Indeed, massively parallel computers now exist in several national laboratories and are beginning to making complex nanostructures accessible to first-principles force calculations. As the experimental capability to produce sample sizes of nanocrystalline materials large enough for bulk applications is developed, better

324

Processing

multiscale models of these materials will be needed. The mesoscopicscale analyses of grain rotation-coalescence by Wolf and co-workers, as well as related models, are an excellent first step in this direction. Constitutive relations that capture the unique properties of nanostructured materials are needed as input into larger-scale modeling. Fortunately, the simulations described (plus many more in the literature not specifically discussed) are laying the foundation needed to derive these relations.

References 1. Brenner, D.W., Shenderova, O.A., and Areshkin, D.A., “Quantum-Based Analytic Interatomic Forces and Materials Simulation,” in: Reviews in Computational Chemistry (K.B. Lipkowitz, and D.B. Boyd, eds.), pp. 213–245, VCH Publishers, New York (1998). 2. Brenner, D.W., Shenderova, O.A., Schall, J.D., Areshkin, D.A., Adiga, S., Harrison, J.A., and Stuart, S.J., “Contributions of Molecular Modeling to Nanometer-Scale Science and Technology,” in: Nanoscience, Engineering and Technology Handbook (W. Goddard, D. Brenner, S. Lyshevski, and G. Iafrate, eds.), Chapter 24, CRC Press, Boca Raton, FL (2002). 3. Wolf, D., Yamakov, V., Phillpot, S.R., Mukherjee, A., and Gleiter, H., “Deformation of Nanocrystalline Materials by Molecular Dynamics Simulations: Relation to Experiment?,” Acta. Mater. 53:1 (2005). 4. Kumar, K.S., Van Swygenhoven, H., and Suresh, S. “Mechanical Behavior of Nanocrystalline Metals and Alloys,” Acta. Mater. 51:5743 (2003). 5. Parinello, M., and Rahman, A., “Crystal Structure and Pair Potentials: A Molecular-Dynamics Study,” Phys. Rev. Lett. 45:1196 (1980). 6. Nosè, S., “A Unified Formulation of the Constant Temperature Molecular Dynamics Methods,” J. Chem. Phys. 81:511 (1984). 7. Hoover, W.G., Molecular Dynamics (Lectures Notes in Physics vol. 258), Springer-Verlag, Berlin (1986). 8. Daw, M.S., “Embedded-Atom Method: Derivation and Application to Impurities, Surface and Other Defects in Metals,” Phys. Rev. B 29:6443 (1984). 9. Foiles, S.M., “Embedded-Atom and Related Methods for Modeling Metallic Systems,” Mater. Res. Soc. Bulletin 21:24 (1996). 10. Finnis, M.W., and Sinclair, J.E., “A Simple Empirical N-body Potential for Transition Metals,” Philos. Mag. A50:45 (1984). 11. Jacobsen, K.W., “Bonding in Metallic Systems: An Effective Medium Approach,” Comments Condens Matter Phys. 14:129 (1988). 12. Tersoff, J., “New Empirical Model for the Structural Properties of Silicon,” Phys. Rev. Lett. 56:632 (1986). 13. Tersoff, J., “Modeling Solid-State Chem-istry: Interatomic Potentials for Multicomponent Systems,” Phys. Rev. B 39:5566 (1989). 14. D.W.Brenner, D.W., “Empirical Potential for Hydrocarbons for Use in Simulating the Chemical Vapor Deposition of Diamond Films,” Phys. Rev. B 42:9458 (1990).

7: Computer Modeling, Brenner

325

15. Brenner, D.W., Shenderova, O.A., Harrison, J.A., Stuart, S.J., Ni, B., and Sinnott, S.B., “Second Generation Reactive Empirical Bond Order (REBO) Potential Energy Expression for Hydrocarbons,” J. Phys. C: Condensed Matter 14:783 (2002). 16. Brenner, D.W., “The Art and Science of an Analytic Potential,” Phys. Status Solidi. B. 217:23 (2000). 17. Pettifor, D.G., and Oleinik, I.I., “Bounded Analytic Bond-Order Potentials for Sigma and Pi Bonds,” Phys. Rev. Lett. 84:4124 (2000). 18. Stillinger, F., and Weber, T.A., “Computer Simulation of Local Order in Condensed Phases of Silicon,” Phys. Rev. B 31:5262 (1985). 19. Feibelman, P.J., “Diffusion Path for an Al Adatom on Al(001),” Phys. Rev. Lett. 65:729 (1990). 20. Srivastava, D., Garrison, B.J., and Brenner, D.W., “Anisotropic Spread of Surface Dimer Openings During the Initial Stages of the Epitaxial Growth of Si on Si [100],” Phys. Rev. Lett. 63:302 (1989). 21. Abraham, F.F., Broughton, J.Q., Bernstein, N., and Kaxiras, E., “Spanning the Length Scales in Dynamic Simulation,” Comput. Phys. 12:538 (1998). 22. Broughton, J.Q., “Concurrent Coupling of Length Scales: Methodology and Application,” Phys. Rev. B 60:2391 (1999). 23. Tadmor, E., Ortiz, M., and Phillips, R., “Quasicontinuum Analysis of Defects in Solids,” Philos. Mag. A 73:1529 (1996). 24. Wagner, G.J., and Liu, W.K., “Coupling of Atomistic and Continuum Simulations using a Bridging Scale Decomposition Approach,” J. Comp. Phys. 190:249 (2003). 25. Voter, A.F., Montalenti, F., and Germann, T.C., “Extending the Time Scale in Atomistic Simulation of Materials,” Annu. Rev. Mater. Res. 32:321 (2002). 26. Castleman, A.W., Jr., and Khanna, S.N., “Cluster and Nanoscale Science: Overview and Perspective,” in: Quantum Phenomena in Clusters and Nanostructures (Springer-Verlag Series in Cluster Physics; S.N. Khanna, and A. W. Castleman, Jr., eds.), Springer-Verlag, New York (2003). 27. Cleveland, C.L., Luedtke, W.D., and Landman, U. “Melting of Gold Clusters,” Phy. Rev. B 60:5065 (1999). 28. Delogu, F., “Structural and Energetic Properties of Unsupported Cu Nanoparticles from Room Temperature to the Melting Point: Molecular Dynamics Simulations,” Phys. Rev. B 72:205418 (2005). 29. Wang, Y., Teitel, S., and Dellago, C., “Melting of Icosahedral Gold Nanoclusters from Molecular Dynamics Simulations,” J. Chem. Phys. 122:214722 (2005). 30. Barnard, A.S., Lin, X.M., and Curtiss, L.A., “Equilibrium Morphology of Face-Centered Cubic Gold Nanoparticles >3 nm and the Shape Changes Induced by Temperature,” J. Phys. Chem. B 109:24465 (2005). 31. Koga, K., Ikeshoji, T., and Sugawara, K.-I., “Size and TemperatureDependent Structural Transitions in Gold Nanoparticles,” Phys. Rev. Lett. 92:115507 (2004). 32. Schebarchov, D., and Hendy, S.C., “Static, Transient and Dynamic Phase Coexistence in Metal Nanoclusters,” J. Chem. Phys. 123:104701 (2005). 33. Meyer, R., Lewis, L.J., Prakash, S., and Entel, P. “Vibrational Properties of Nanoscale Materials: From Nanoparticles to Nanocrystalline Materials,” Phys. Rev. B 68:104303 (2003).

326

Processing

34. Kodiyalam, S., Kalia, R.K., Nakano, A., Vashishta, P., “Multiple Grains in Nanocrystals: Effect of Initial Size and Shape on Transformed Structures Under Pressure,” Phys. Rev. Lett. 93:203401 (2004). 35. Campbell, T.J., Kalia, R.K., Nakano, A., Vashishta, P., Ogata, S., and Rodgers, S., “Dynamics of Oxidation of Aluminum Nanoclusters using Variable Charge Molecular-Dynamics Simulations on Parallel Computers,” Phys. Rev. Lett. 82:4866 (1999). 36. Campbell, T.J., Aral, G., Ogata, S., Kalia, R.K., Nakano, A., and Vashishta, P., “Oxidation of Aluminum Nanoclusters,” Phys. Rev. B 71:205413 (2005). 37. Shenderova, O.A., Zhirnov, V., and Brenner, D.W., “Carbon Materials and Nanostructures,” Crit. Rev. Solid State Mater. Sci. 32:347 (2002). 38. Phillpot, S.R., Wolf, D., and Gleiter, H., “Molecular-Dynamics Study of the Synthesis and Characterization of a fully Dense, Three-Dimensional Nanocrystalline Material,” J. Appl. Phys. 78:847 (1995). 39. Phillpot, S.R., Wolf, D., and Gleiter, H., “A Structural Model for Grain Boundaries in Nanocrystalline Materials,” Scripta Metall. Mater. 33:1245 (1995). 40. Schiotz, J., Di Tolla, F.D., and Jacobsen, K.W., “Softening of Nanocrystalline Metals at Very Small Grain Sizes,” Nature 391:1223 (1998). 41. Zheng, G.P., Wang, Y.M., and Li, M., “Atomistic Simulation Studies on Deformation Mechanism of Nanocrystalline Cobalt,” Acta. Mater. 53:3893 (2005). 42. Van Swygenhoven, H., Spaczer, M., Car, A., and Farkas, D., “Competing Plastic Deformation Mechanisms in Nanophase Metals,” Phys. Rev. B 60:22 (1999). 43. Yanakov, V., Wolf, D., Salazar, M., Phillpot, S.R., and Gleiter, H., “LengthScale Effects in the Nucleation of Extended Dislocations in Nanocrystalline Al by Molecular-Dynamics Simulation,” Acta. Mater. 49:2713 (2001). 44. Zhang, Y.W., Liu, P., and Lu, C., “Molecular Dynamics Simulations of the Preparation and Deformation of Nanocrystalline Copper,” Acta. Mater. 52:5105 (2004). 45. Kodiyalam, S., Kalia, R., Nakano, A., and Vashashta, P., “Multiple Grains in Nanocrystals: Effect of Initial Shape and Size on Transformed Structures under Pressure,” Phys. Rev. Lett. 93:203401 (2004). 46. Chatterjee, A., Kalia, R.K., Nakano, A., Omeltchenko, A., Tsuruta, K., Vashishta, P., Loong, C.-K., Winterer, M., and Klein, S., “Sintering, Structure, and Mechanical Properties of Nanophase SiC: A Molecular-Dynamics and Neutron Scattering Study,” Appl. Phys. Lett. 77:1132 (2000). 47. Zeng, P., Zajac, S., Clapp, P.C., and Rifkin, J.A., “Nanoparticle Sintering Simulations,” Mater. Sci. Eng. A. 252:301 (1998). 48. Koparde, V.N., and Cummings, P.T., “Molecular Dynamics Simulation of Titanium Dioxide Nanoparticle Sintering,” J. Phys. Chem. B 109:24280 (2005). 49. Haslam, A.F., Phillpot, S.R., Wolf, D., Moldovan, D., and Gleiter, H., “Mechanisms of Grain Growth in Nanocrystalline fcc Metals by MolecularDynamics Simulation,” Mater. Sci. Eng. A 318:293 (2001). 50. Bachurin, D.V., Nazarov, A.A., Shenderova, O.A., and Brenner, D.W., “Diffusion-Accomodated Rigid Body Translations Along Grain Boundaries in Nanostructured Materials,” Mater. Sci. Eng. A 359:247 (2003).

7: Computer Modeling, Brenner

327

51. Moldovan, D., Wolf, D., and Philpott, S.R., “Theory of DiffusionAccommodated Grain Rotation in Columnar Polycrystalline Microstructures,” Acta. Mater. 49:3521 (2001). 52. Moldovan, D., Yamakiv, V., Wolf, D., and Phillpot, S.R., “Scaling Behavior of Grain-Rotation-Induced Grain Growth,” Phys. Rev. Lett. 89:206101 (2002). 53. Moldovan, D., Wolf, D., and Phillpot, S., “Linking Atomistic and Mesoscale Simulations of Nanocrystalline Materials: Quantitative Validation for the Case of Grain Growth,” Philos. Mag. 83:3643 (2003). 54. Xiao, S., and Hu, W., “Molecular Dynamics Simulations of Grain Growth in Nanocrystalline Ag,” J. Crystal Growth 286:512 (2006). 55. Van Swygenhoven, H., Spazcer, M., Caro, A., and Farkas, D. “Competing Plastic Deformation Mechanisms in Nanophase Metals,” Phys. Rev. B 60:22 (1999). 56. Van Swygenhoven, H., and Derlet, P.M., “Grain Boundary Sliding in Nanocrystalline fcc Metals,” Phys. Rev. B 64:224105 (2001). 57. Yamakov, V., Wolf, D., Salazar, M., Phillpot, S.R., and Gleiter, H., “Length Scale Effects in the Nucleation of Extended Dislocations in Nanocrysalline Al by Molecular Dynamics Simulation,” Acta. Mater. 49:2713 (2001). 58. Van Swygenhoven, H., Derlet, P.M., and Froseth, A.G., “Stacking Fault Energies and Slip in Nanocrystalline Metals,” Nat. Mat. 3:399 (2004). 59. Yamakov, V., Wolf, D., Phillpot, S.R., Mukherjee, A.K., and Gleiter, H., “Deformation Mechanism Map for Nanocrystalline metals by Molecular Dynamics Simulation,” Nat. Mater. 3:43 (2004). 60. Farkas, D., and Curtin, W., “Plastic Deformation Mechanisms in Nanocrystalline Columnar Grain Structures,” Mater. Science Eng. A 412:316 (2005). 61. Wang, Y.M., Ma, E., and Chen, M.W., Enhanced tensile ductility and toughness in nanostructured Cu. Appl. Phys. Lett. 80:2395 (2002). 62. Hasnaoui, A., Van Swygenhoven, H., and Derlet, P.M., “Dimples on Nanocrystalline Fracture Surfaces as Evidenced for Shear Plane Formation,” Science 300:1550 (2003). 63. Latapie, A., and Farkas, D., “Molecular Dynamics Investigation of the Fracture Behavior of Nanocrystalline α-Fe,” Phys. Rev. B 69:134110 (2004). 64. Farkas, D., Van Petegem, S., Derlet, P.M., and Van Swygenhoven, H., “Dislocation Activity in Nano-Void Formation Near Crack Tips in Nanocrystalline Ni,” Acta. Mater. 53:3115 (2005). 65. Farkas, D., Sillemann, M., and Hyde, B. “Atomistic Mechanisms of Fatigue in Nanocrystalline Metals,” Phys. Rev. Lett. 94:165502 (2005). 66. Szlufarska, I., Nakano, A., and Vashishta, P., “A Crossover in the Mechanical Response of Nanocrystalline Ceramics,” Science 309:911 (2005). 67. Kadua, K., Germann, T.C., Lomdahl, P.S., and Holian, B.L., “Microscopic View of Structural Phase Transitions Induced by Shock Waves,” Science 296:1681 (2002). 68. Bringa, E.M., Caro, A., Wang, Y., Victoria, M., McNaney, J.M., Remington, B.A., Smith, R.F., Torralva, B.R., and Van Swygenhoven, H., “Ultrahigh Strength in Nanocrystalline Materials Under Shock Loading,” Science 309:1838 (2005). 69. Pasquini, L., Barla, A., Chumakov, A.I., Leupold, O., Ruffer, R., Deriu, A., and Bonetti, E., “Size and Oxidation Effects on the Vibrational Properties of Nanocrystalline α-Fe,” Phys. Rev. B 66:073410 (2002).

328

Processing

70. Derlet, P.M., Meyer, R., Lewis, L.J., Stuhr, U., and Van Swygenhoven, H., “Low-Frequency Vibrational Properties of Nanocrystalline Materials,” Phys. Rev. Lett. 87:205501 (2001). 71. Millett, P.C., Selvam, R.P., and Saxene, A., “Molecular Dynamics Simulations of Grain Size Stabilization in Nanocrystalline Materials by Addition of Dopants,” Acta. Mater. 54:297 (2006). 72. Jang, S., Purohit, Y., Irving, D., Padgett, C., Scattergood, R., and Brenner, D., unpublished. 73. Rajulapati, K.V., Scattergood, R.O., Murty, K.L., Duscher, G., and Koch, C.C., “Effect of Pb on the Mechanical Properties of Nanocrystalline Al,” unpublished. 74. Baskes, M.I., “Modified Embedded-Atom Potentials for Cubic Materials and Impurities,” Phys. Rev. B 46:2727 (1992). 75. Car, R., and Parrinello, M., “Unified Approach for Molecular Dynamics and Density-Functional Theory,” Phys. Rev. Lett. 55:2471 (1985).

PART 2 PROPERTIES

8 Diffusion in Nanocrystalline Materials Wolfgang Sprengel University of Stuttgart, Stuttgart, Germany

8.1 Introduction Nanocrystalline (n-) materials are polycrystals with an ultrafine grain size (diameter 5–100 nm) and a high volume fraction of atoms located in interfaces. Since the pioneering research work of Gleiter and co-workers in 1984,[1] nanocrystalline materials have gained considerable interest due to high prospects for their attractive potential applications arising from improved mechanical and magnetic properties compared to their coarsegrained counterparts.[2] With respect to both the structure and the physical properties of nanocrystalline solids, the understanding of the atomic transport properties in these materials represents a key research issue. In general, atomic transport in nanocrystalline materials differs substantially from that in coarse-grained or single-crystalline materials. This is due to the fact that, in nanocrystalline solids, the crystallite interfaces provide paths of high diffusivity, whereas in more coarse-grained crystals, volume self-diffusion or substitutional diffusion dominates at least at temperatures higher than approximately half of the melting temperature. Interface diffusion, in combination with a high fraction of atoms in interfaces, gives rise to modified physical properties of nanocrystalline solids. For instance, enhanced ductility of nanocrystalline ceramics[3,4] and intermetallic compounds[5] has been analyzed in the framework of models of mesoscopic sliding[3] or grain switching according to AshbyVerrall,[5] both of which are controlled by diffusion in interfaces. With respect to nanocrystalline magnetic materials, atomic diffusion enables, for example, a controlled stress-induced adjustment of magnetic anisotropies in soft-magnetic alloys[6] or texturing of hard-magnetic Nd2Fe14B-nanocomposites.[7] Furthermore, diffusion processes may control the formation of nanocrystalline materials, for example, by means of crystallization of amorphous precursors,[8] as well as the Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 331–364 © 2007 William Andrew, Inc.

331

332

Properties

stability of nanocrystalline materials (relaxation, crystallite growth), their reactivity, corrosion behavior, or interaction with gases. The relevance of diffusion-controlled processes, as outlined above, demands a comprehensive understanding of atomic diffusion in nanocrystalline materials. After the first study in this field reported 1987 by Horvath et al.[9] there is the key question to what extent the interface diffusion in nanocrystalline materials differs from that in conventional grain boundaries. Due to the correlation between the interface structure and the diffusion behavior (see below), an answer to this question enables conclusions on the basic structure of interfaces in nanocrystalline materials. The present article is an up-to-date review of atomic diffusion in nanocrystalline materials.[10] Since the earlier reviews on this topic,[11–13] diffusion data on highly-dense nanocrystalline metals,[14,15] and oxides[16] after gas-phase condensation, as well as on nanocrystalline materials prepared by other routes, e.g., crystallization[17] and severe plastic deformation,[18] have become available. Most importantly, diffusion studies of conventional grain boundaries have been extended to low temperatures,[19–21] thus allowing a conclusive comparison with the diffusion characteristics of nanocrystalline solids studied in a similar temperature regime. After a brief summary of basic models for interface diffusion (section 8.2) and of the diffusion characteristics in conventional grain boundaries (section 8.3), results of diffusion studies on nanocrystalline metals are presented and discussed (section 8.4). The correlation between the diffusion and rapid grain growth in highly-dense nanocrystalline metals and attempts to stabilize the nanocrystalline structure are considered. As particular examples of diffusion in a structurally stable nanocrystalline system, results obtained on the prototype soft-magnetic alloy Fe73.5Si13.5B9Nb3Cu1 (FINEMET) (section 8.4.4) and hard-magnetic Fe2Nd14B (section 8.4.5) are presented. The discussion includes the phenomena of an amorphous intergranular phase and of intergranular phase melting. In section 8.4.6 a brief summary is given of hydrogen diffusion in nanocrystalline materials, which is used as a probe for free volumes in interfaces. The final section (section 8.5), gives an overview of recent diffusion studies in nanocrystalline ceramics with a focus on n-ZrO2-base materials.

8.2 Modeling of Interface Diffusion Diffusion coefficients in solids are most precisely determined from the analysis of concentration–penetration profiles obtained from long-range

8: Diffusion in Nanocrystalline Materials, Sprengel

333

diffusion experiments with radioactive tracer atoms. In the most common case of diffusion in polycrystals, two simultaneous diffusion processes have to be considered, i.e., rapid diffusion in the crystallite interfaces (diffusion coefficient DB) accompanied by diffusion from the interfaces or the specimen surface into the volume of the crystallites (diffusion coefficient DV). The combination of these two processes can conveniently be classified according to Harrison,[22] and Levine and MacCallum,[23] by three different diffusion regimes, denoted as types A, B, and C. The regimes are defined by the different ratios of the diffusion length in the crystallites (∝ ( Dv t ) ), t = diffusion time) to the crystallite diameter, d, or the interface thickness, δ. The evaluation of the diffusion profiles is relatively straight forward.[24] In the practical case of diffusion experiments in nanocrystals the diffusivity inside the crystallites is negligibly small ( ( Dv t ) << δ) as these experiments are preformed at low temperature to ensure a stable nanocrystalline structure. In this type-C diffusion regime the concentration profiles C(x) show a Gaussian shape and the diffusion coefficient DB in the interfaces is directly determined according to C( x , t ) ∝

2 1 ⎛ x ⎞ exp − ⎝ 4 DBt ⎠ DBt

Eq. (8-1)

which is the solution of the diffusion equation for a thin layer of radiotracer atoms on the specimen surface at t = 0. Here x and t denote the diffusion depth and the diffusion time.[24] In cases where the condition δ << ( Dv t ) << d holds, i.e., when nonnegligible diffusion of atoms from the interfaces into the crystallites occurs, the type-B diffusion regime is valid from which the triple product sδDB = 1.322

DV ⎛ ∂ ln C( x, t ) ⎞ − t ⎝ ∂x 6 5 ⎠

−5 3

Eq. (8-2)

can be derived, where s denotes the grain-boundary segregation factor. In this case of a superposition of interface and volume diffusion, data on the volume diffusivity DV inside the crystallites are required for a determination of the grain-boundary diffusivity DB.[24] Of particular importance, considering nanocrystalline metals, are cases where diffusion is accompanied by interface migration, which is often due to crystallite growth (see section 8.4). Then tracer atoms are excluded from the fast interfacial diffusion paths and are immobilized inside the crystallites where the volume diffusivity is negligible. Due to this process, an evaluation of the diffusion profiles according to type-C kinetics (Eq. 8-1) results in an apparent diffusivity that is lower than the actual atomic

Properties

334

diffusivity in interfaces. If for the interface migration a constant interface velocity, v, is assumed, a simple model yields the following relationship for the concentration profile:[25] ν ⎞ ⎛ C( x ) ∝ exp − x ⎝ DBδ ⎠

Eq. (8-3)

On the other hand, different types of interfaces may be present in nanocrystalline materials. Klinger et al.,[26] took into account diffusion in triple-line intersections of interfaces in addition to diffusion in the interfaces and the crystallites, which leads to a refinement of the diffusion profile analysis. Furthermore, if nanocrystalline materials are prepared from powders consisting of agglomerates of nanocrystallites, the interfaces between the agglomerates (inter-agglomerate boundaries) and residual porosity between the agglomerates provide pathways of even higher diffusivity than in the interfaces between the nanocrystallites which are located inside the agglomerates (intra-agglomerate boundaries). A description of the diffusion behavior taking these two types of interfaces into account was first reported by Bokstein et al. on diffusion in clusterassembled nanocrystalline Ni[27] and later refined by Divinski et al.[28] for diffusion in nanocrystalline γ-Fe61.2Ni38.8. In both cases the diffusion profiles were analyzed using relationships analogous to those of type-B grain boundary diffusion kinetics for conventional, coarse-grained polycrystals, where now the inter-agglomerate boundaries are equivalent to the conventional grain boundaries in polycrystals and the intra-agglomerate boundaries take into account the diffusion fluxes from the inter-agglomerate boundaries into the intra-agglomerate boundaries. For both diffusion in nanocrystalline Ni[27] and in γ-Fe61.2Ni38.8,[28] the diffusion coefficients of the intra-agglomerate boundaries were found to be characteristic of conventional grain boundary diffusivities, whereas diffusivities several orders of magnitude higher were deduced for the inter-agglomerate boundaries.

8.3 Diffusion in Grain Boundaries of Metals Grain boundary diffusion in conventional polycrystalline materials or in bicrystals represents a subject of comprehensive research, particularly due to its technical relevance for a variety of processes, e.g., solid-state transformation, creep, or corrosion. (see reviews by Kaur et al.[24] and Sutton and Balluffi[29]) The first quantitative description of grain boundary diffusion dates back to Langmuir in 1934.[30]

8: Diffusion in Nanocrystalline Materials, Sprengel

335

Strong indications of a vacancy-mediated self-diffusion process in grain boundaries of metals are derived from diffusion studies under hydrostatic pressure (Ag[31]), studies of the isotope effect (Ag[32]), and computer simulations (Fe,[33] Ag,[34] Cu[35]). The extension of diffusion studies of conventional grain boundaries toward lower temperatures[19–21] is important with respect to an assessment of the diffusion characteristics of nanocrystalline metals. According to these studies, the diffusivities in conventional grain boundaries at low temperatures are higher and the diffusion enthalpies, HD, are lower than expected from extrapolations from high temperatures (see Fig. 8.2). Atomistic simulations indicate that the variation of HD with temperature is due to an interstitial-type diffusion mechanism[34] that adds, at low temperatures, to the vacancy mechanism that is believed to prevail in conventional grain boundaries of metals (see above). Correlations between the grain boundary structure and diffusion are derived from diffusion studies on [001]-tilt boundaries in noble metals.[21,36,37] These studies show that, with increasing tilt angle, the diffusivity parallel to the tilt axis increases up to an order of magnitude,[21,36] with local minima near-coincidence orientations of the adjacent crystals (Σ5).[37] A similar correlation exists between the tilt- or twist-angle and the specific energy of grain boundaries.[38] On the other hand, the specific energy increases with the excess volume of grain boundaries.[38,39] These results, in conclusion, may indicate an increase of the grain boundary diffusivity with the excess volume.

8.4 Diffusion in Nanocrystalline Metals 8.4.1 Specific Aspects The first diffusion studies on nanocrystalline materials, i.e., Cu selfdiffusion[9] and Ag diffusion[40] in nanocrystalline Cu, revealed extraordinary high diffusivities compared to conventional grain boundaries (see Table 8.1). However, it turned out later that effects such as structural relaxation, grain growth, and purity or porosity of the specimens have to be taken into account to obtain unambiguous and reproducible results. This is especially true if the various routes for the synthesis of nanocrystalline materials are taken into account. In the meantime, the research on and consequently the experience with nanocrystalline materials has expanded significantly and high-quality nanocrystalline specimens that are stable

Cu

ED

B

Cu

SPD

CC

Ni

Sb

CC

CC

Au

Bi

CC

CC

Cu

Ag

CC

CC

Fe

Fe

CC

Fe

SPD

SPD

Fe

CC

Diffusor 423–523

a

452–499 293–393

–

– 293–383

423

398–448

293–473

773(b)

398

323–373

293–413

373

303–373

293

293

293

293

293

527

527

a

403

371–623

293–673

373

Td (K)

Ta (K)

1 × 10

(0.27TM)

(0.27TM)

(0.245TM)

(4.5 × 10−18)

−17

3.8 × 10

9.6 × 10−15c (0.245TM)

1.6 × 10 −19

5.9 × 10−18

1.3 × 10−22

−21

2.4 × 10

3 × 10−20c

−22

4.8 × 10

1.6 × 10−18

−19

9.2 × 10

3 × 10−19c (0.29TM)

−21c

3 × 10−20

2.8 × 10 −20

D (0.25TM) (m2s−1)

(7.1 × 10 )

1.2 × 10−16 (0.29TM) (2.7 × 10−19)f

bcc, general

Ag

Σ5-tilt-GB Ag/Au

−23 f

(1.7 × 10−21)f

fcc, general

Grain boundary (GB) diffusion in coarse-grained metals and bicrystal:

Fe

Ni

Cu

Fe

Pd

Synthesis

Pure Metals

Table 8.1 Overview of Diffusion Studies on Nanocrystalline Materials

e

C

C

C

2 IFe

2 IF

C

C

C

C

C

C

C

C

Model

31 8

7

30

300d

70

50

10

10

8

d

NS

b

80b

50

b

d (nm)

SIMS

SIMS

SIMS

RT

RBS

RBS

AES

EPMA

RT

RT

RT

RT

RT

Method

[65]

[64]

[62]

[62]

[63]

[45]

[45]

[27]

[56]

[44]

[44]

[40]

[9]

[43]

[43]

[18]

[15]

Ref.

336 Properties

MS/HC

MA/HC

MA/HC

SPD

Nd14.2Fe80.8B5

γ−Fe61.2Ni38.8

Al91.9Ti7.8Fe0.3

Al96.8Mg3Sc0.2/ Al99.8Sc0.2

CC

(ZrO2)93.1(Y2O3)6.9

1223–1323

O

O

18

1223–1353

Ta (K)

18

293

673

Diffusor

Cu

1123

1048

Nd

Fe

1048

Fe

873

817

Ge

Fe

810–818

Ta (K)

Fe

Diffusor −8

3 −1

473–773

723–1223

Td (K)

523–723

371–571

636–1013

623–950

2.0 × 10

m s

0.91

1.95 2 −1

−5

3.3 × 10 m s

2 −1

−5

Q (eV)

1.02 g

dD0 or D0

–

0.364

1.53

2 −1

2.4 × 10−12 m2 s−1

−3

3.4 × 10 m s

2.65 1.93

2 −1

1.74

1.64

1.70

2.91

1.9

Q (ev)

4.2 × 10−3 m2 s−1

4

1.7 × 10 m s

1.5 × 10−11 m3 s−1

m s

6.8 × 10

689–951

3 −1

623–741

−14

2.8 × 10−7 m2 s−1

1.1× 10 m s

–

d Do or Do

593–773

735–783

628–773

Td (K)

e

A

A, B

Model

BM

C

B, 2 IF

A, AB

C

B

2 IFe

B

BL

Model

b

60–100

80

d (nm)

200

22

80–100d

100–500

100d

18d

(d)

13

13

d

d (nm)

SIMS

SIMS

Method

EPMA

SIMS

RT

RT

RT

RT

RT

RT

Method

[110]

[109]

Ref.

[67]

[66]

[28]

[92]

[42,52]

[48]

[50]

[17]

Ref.

Pure Metals: Diffusivities D at 1/4 of the melting temperature TM obtained by linear interpolation of the experimental data. D-values obtained by linear extrapolation are given in parentheses. Grain boundary (GB) diffusivities in coarse-grained metals and bicrystals are given for comparison. Alloys, Ceramics: Activation energy Q and preexponential factors δD0 or D0 of diffusion.

CC

ZrO2

Synthesis

MS/CRY

Fe90Zr7B3

Ceramics

MS/CRY

Synthesis

Fe73.5Si13.5B9-Nb3Cu1

Alloys

8: Diffusion in Nanocrystalline Materials, Sprengel 337

Model used for analysis: GB diffusion kinetics of type A, B, or C; models with two different types of interfaces (2 IF); brick-layer model (BL); Boltzmann–Matano technique (BM). Synthesis route: Crystallite condensation and compaction (CC), severe plastic deformation (SPD), electrodeposition (ED), melt-spinning and crystallization (MS/CRY), melt-spinning and hot compaction (MS/HC), mechanical alloying and hot compaction (MA/HC). Experimental method: radiotracer (RT), electron-probe microanalysis (EPMA), Auger electron spectroscopy (AES); secondary-ion mass spectroscopy (SIMS), Rutherford backscattering (RBS). Ta: pre-annealing temperature. Td: temperature rang of diffusion. d: crystallite size prior to diffusion. a Temperature of crystallite compaction. b Crystallite growth during diffusion annealing observed. c Variation with temperature of preanneal or with diffusion time observed. d Crystallite size constant upon diffusion annealing. e 2 IF diffusion model (the two values in the 1st and 2nd lines refer to diffusion along two types of interfaces). f Related to grain-boundary thickness δ = 1 nm. g Interdiffusion.

Table 8.1 Footnote (cont’d)

338 Properties

8: Diffusion in Nanocrystalline Materials, Sprengel

339

over a wide temperature range can be prepared for various metals and alloys and are the basis for profound diffusion studies.[14,15,17,18,28,41–43] For the measurement of the diffusivities in nanocrystalline materials, nearly all experimental techniques already established for diffusion measurements in solids[24] are applicable, such as the high-precision radiotracer method with sputter- or mechanical-sectioning, electron probe microanalysis (EPMA), Auger electron spectroscopy (AES) secondary ion-mass spectroscopy (SIMS) with depth profiling, Rutherford backscattering (RBS), or nuclear magnetic resonance (NMR). In the following, specific aspects of diffusion in nanocrystalline materials such as structural relaxation, grain growth, porosity, different types of interfaces, intergranular amorphous phases, and intergranular phase melting, which are often directly related to the preparation method, are briefly discussed. For nanocrystalline metals prepared by crystallite condensation in an inert-gas atmosphere with subsequent crystallite compaction, extensive experimental evidence exists that structural relaxation occurs at slightly elevated temperatures (see Ref. 24). This can be inferred from the decrease of the diffusivity with increasing diffusion time, as measured, for instance, in nanocrystalline Fe prepared by the cluster condensation and compaction route and by severe plastic deformation.[43] Similar observations were made by Kolobov et al.[45] on nanostructured Ni prepared by severe plastic deformation. In that case, the diffusivity of Cu at 423 K was found to decrease by more than three orders of magnitude upon preannealing the sample at 523 K prior to the onset of grain growth.[45] Both in n-Fe[43] and in n-Ni[45] the interfacial diffusion coefficients in the relaxed state appear to be similar to or only slightly higher than the values expected for conventional grain boundaries in coarse-grained materials from an extrapolation of high-temperature data. Grain growth during the diffusion experiments reduces the number of interfaces and the migrating interfaces slow down the tracer diffusion as a proportion of the tracer atoms become immobile on lattice sites inside the crystallites. The complications introduced by grain growth are even more serious when attempting to ascribe a physical interpretation to experimental values for the activation energies of diffusion, because grain growth seriously limits the temperature range of such studies, especially in the case of pure metals. Therefore, only characteristic diffusion coefficients are quoted in the first part of Table 8.1. An enhanced stability of nanocrystalline metals with respect to grain growth may be obtained in the case of thin films, as studied, e.g., in n-Cu thin films by Erdelyi et al.[46]

340

Properties

Despite a stable microstructure, the diffusion characteristics of some nanocrystalline alloys may be quite complex, owing to the presence of different types of interfaces, particularly in those cases in which bulk samples are prepared from powders consisting of agglomerates of nanocrystallites. The inter-agglomerate interfaces as well as residual porosity between the agglomerates are paths of high diffusivity.[47] Lower diffusivities are expected along the interfaces between the nanocrystallites inside the agglomerates. The first diffusion studies taking these two types of interfaces into account were performed by Bokstein et al. on clusterassembled nanocrystalline Ni with a density of 92–93% of the bulk density,[27] and more recently by Divinski et al.[28] on 98%-dense γFe61.2Ni38.8 prepared by mechanical alloying and hot compaction. For nanocrystalline, soft-magnetic Fe-based alloys, e.g., Fe73.5Si13.5B9Nb3Cu1 or Fe90Zr7B3, prepared by crystallization of an amorphous precursor material, Fe-tracer diffusivities have been reported that are substantially lower than those for grain boundaries in α-Fe.[41,48,49] This particular feature differs strongly from the diffusion behavior observed in other nanocrystalline metals and alloys. The reduced interface diffusivities are similar to or even lower than those of the initial amorphous phases, which leads to the conclusion that an intergranular amorphous phase provides the prevailing diffusion paths in these specific nanocrystalline alloys. For nanocrystalline alloys based on Nd2Fe14B, an intergranular melting transition of Nd-enriched grain boundaries has been observed.[51] The diffusivity in these alloys increases several orders of magnitude upon passing the transition temperature toward diffusivity values measured for bulk melts. Below this intergranular melting transition, the grain boundary diffusivities in n-Nd2Fe14B are found to be similar to those of α-Fe.[42,52] Specific diffusion studies are discussed in detail in a separate section below.

8.4.2 Nanocrystalline Pure Metals Selected data on the diffusion of metals and metalloids in nanocrystalline pure metals and alloys are given in Table 8.1 in comparison with diffusion data obtained on crystals and conventional grain boundaries. So far, diffusion studies on nanocrystalline pure metals have focused primarily on the cubic face-centered metals Pd, Cu, and Ni, of which the results are summarized in Fig. 8.1. In the following, the results of Fe-tracer diffusion studies on nanocrystalline Pd are discussed in detail.[10,14,15,18] The studies were aimed at

8: Diffusion in Nanocrystalline Materials, Sprengel

341

Figure 8.1 Arrhenius plots of tracer diffusivities D in face-centered cubic nanocrystalline metals (the diffusion temperatures are scaled to the melting temperatures TM; the tracer atoms are quoted in brackets): Diffusion of Fe ()[14] and Pt ()[14] in cluster-compacted n-Pd without preannealing; diffusion of Fe in submicrocrystalline (smc) Pd prepared by severe plastic deformation[18] without pre-annealing () and after preannealing at Ta = 453 K (), 553 K () and 673 K () for 2400 s. The data for the 59Fe diffusion in smc-Pd and n-Pd were derived from the deep-penetration tails of the diffusion profiles assuming Gaussian-type solutions (type-C kinetics[24]). (): Diffusion of Cu,[9] Ag,[40] Au,[57,58] Bi,[57,58] and Sb[56] in n-Cu; diffusion of Ag, Cu, and Au in n-Pd;[57] Ni diffusion in n-Ni[27]. Analysis of n-Ni(Ni) and n-Cu(Sb) according to a model of atomic diffusion in a porous solid (see Ref. 27): diffusion on cluster surfaces (– – –), diffusion in interfaces of crystallite clusters [—, n-Cu(Sb)]. Coarsegrained metals (dotted lines denote extrapolated values): self-diffusion on Ni (110) surfaces parallel to <110>;[52] self-diffusion in crystalline (c-) Ni,[60] Cu,[53] and Pd;[61] Fe diffusion in Pd after annealing of deformation-prepared Pd at 977 K ();[18] grain-boundary diffusivity in fcc metals[62] extrapolated from high temperatures (assuming a grain-boundary thickness δ = 1 nm). Grain-boundary diffusion (g) in the type-C regime: Au[19,21] and Se[20] in grain boundaries of Cu; Ag in Σ5(310)-tilt grain boundaries of Au/Ag[21] (δ = 1 nm); Ag in evaporated Pd layers (Δ).[40]

diffusion measurements on highly-dense nanocrystalline metals where the influence of porosity can be excluded or is negligible. By means of crystallite condensation and compaction, the nearly theoretical mass density of nanocrystalline Pd is achieved if a high compaction pressure (4 GPa) at slightly elevated temperatures (380 K) under ultrahigh vacuum conditions is applied.[10,15] In addition, the diffusion studies are extended to porosity-free submicrocrystalline (smc) Pd prepared by severe plastic

342

Properties

deformation.[10,18] The diffusion measurements were performed by means of radiotracer techniques, where the profiles of the specific activity corresponding to the radiotracer concentration was determined by ion-beam sectioning.[53] Both for nanocrystalline Pd prepared by cluster condensation and compaction (CC) and for submicrocrystalline Pd prepared by severe plastic deformation (SPD),[18] fast diffusion of Fe was detected at temperatures slightly higher than ambient temperature. This is considered as evidence that the atomic transport occurs in the crystallite interfaces since the Fediffusion in crystalline Pd, which is similar to the Pd self-diffusivity, can be neglected at these temperatures (Fig. 8.1). The 59Fe diffusivities in n-Pd are considered to be characteristic of selfdiffusion in interfaces since initial studies with 103Pd indicate a rather similar diffusion behavior. Moreover, the diffusion in cluster-compacted n-Pd is probably not affected by porosity as indicated by the high relative mass density (ρ/ρ0 > 0.97), the rapid grain growth, and the similar 59Fediffusivities compared with deformation-prepared smc-Pd (Fig. 8.1). The interfacial diffusivity in nanocrystalline Pd appears to be similar to conventional grain boundaries, which can be concluded from a comparison with the diffusion studies of bicrystals or coarse-grained polycrystals (see Fig. 8.1). According to the correlation between the structure and diffusion, the similar diffusion behavior suggests an equilibrium-like structure of the interfaces in nanocrystalline metals. For obvious reasons this may be related to interface migration. During this process, interfaces with an initially nonequilibrium structure due to the preparation can easily relax within a short migration distance. Strong evidence for interfacial relaxation in cluster-compacted nanocrystalline metals, at slightly elevated temperatures, is obtained from a decrease of the excess free volume[54] and of internal strains.[55] Diffusivities in cluster-condensed n-Pd, similar to conventional grain boundaries, can also be estimated from a study of the Pt diffusion by means of Rutherford backscattering (Fig. 8.2),[15] taking into account interface migration during the diffusion annealing. In combination with the high fraction of interfaces, the interface diffusion, much faster than lattice self-diffusion (Fig. 8.1), strongly favors diffusion-controlled processes in nanocrystalline solids. Data according to which, in contrast to the experiments reported above, diffusivities in cluster-compacted nanocrystalline metals are substantially enhanced as compared to grain boundaries (Fig. 8.1) might have been affected by fast diffusion paths due to residual porosity as signified by the

8: Diffusion in Nanocrystalline Materials, Sprengel

343

reduced mass density in the earlier experiments (ρ/ρ0 ≈ 0.9). Taking into account porosity in the framework of the aforementioned cluster model, Bokstein et al.[27] estimated a self-diffusivity similar to conventional grain boundaries for the crystallite interfaces in cluster-compacted n-Ni. Diffusivity values of, for example, Sb in n-Cu, which are considerably lower than the self-diffusivities in conventional grain boundaries (Fig. 8.1) might presumably arise from the effect of chemical diffusion and the formation of intermetallic compounds due to relatively high concentrations of the diffusing atoms necessary for the application of the SIMS technique.[56] In addition, the diffusion may be hampered by oxides that are formed in the interfaces as a consequence of oxygen penetration in the case of insufficient powder densification.[56] Diffusion studies on nanocrystalline cluster-condensed metals have been extended to body-centered cubic Fe.[43] In that case, direct evidence for the influence of structural relaxation on the interface diffusion is derived from the isothermal time-dependence of the self-diffusion coefficient. In the relaxed state of n-Fe, interface diffusivities characteristic of conventional grain boundaries occur similarly to nanocrystalline fcc metals.

8.4.3 Correlation between Diffusion and Crystallite Growth The interface diffusion in highly-dense n-metals after cluster synthesis or severe plastic deformation is accompanied by strong crystallite growth (Fig. 8.2). A quantitative analysis shows that the rapid crystallite growth can be attributed to the high driving forces due to the small crystallite size, taking into account the dynamics of equilibrium-like grain boundaries (Fig. 8.3).[14] The analysis is based on the reasonable assumption that the activation enthalpy HG of crystallite growth is given by the activation enthalpy HM of grain boundary migration, which—according to Sutton and Balluffi[29]—is lower than the grain boundary diffusion enthalpy HD. The reduced value of HM, compared to HD, is considered one indication that the migration of large-angle boundaries occurs by a local shuffling of atoms across the interface of the adjacent crystals rather than by a transport of atoms in the grain boundary plane.[29,68] In conclusion, both the mobility of interfaces (Fig. 8.3) and the interfacial diffusivities (Fig. 8.1) in highly-dense nanocrystalline metals

Figure 8.2 Variation of the crystallite size d with the annealing temperature Ta in submicrocrystalline (smc) Cu[80] and Pd[18] prepared by severe plastic deformation (SPD), in n-Pd (I, ρ/ρ0 > 97%;[15] II, ρ/ρ0 = 86%[81]) and n-Ni (ρ/ρ0 = 79%[10]) prepared by cluster condensation and compaction, in n-Pd prepared by electrochemical deposition (ED) (III),[73] and in crystallized Fe73.5Si13.5B9Nb3Cu1 (FINEMET[82]). ρ/ρ0 denotes the relative mass density of the cluster-compacted metals after preparation.

Figure 8.3 Arrhenius-representation of crystallite sizes d in submicrocrystalline and nanocrystalline metals and in FINEMET (cf. Fig. 8.2) according to a parabolic growth behavior (d 20 − d 2)/t ∝ exp(−H G/kBT) (for details see Ref. 14; d0 = initial crystallite size). The dotted or dashed lines refer to a parabolic growth behavior with the activation enthalpy of crystallite growth HG as given by the activation enthalpy H M of the grain boundary mobility [H M = 0.83 eV (Cu), 1.12 eV (Pd); see Ref. 14] or the grain boundary self-diffusion [H D = 1.43 eV (Pd)[62]], respectively.

8: Diffusion in Nanocrystalline Materials, Sprengel

345

appear to follow the same characteristics as those of conventional grain boundaries. A retardation of the normal crystallite growth, which is frequently observed in nanocrystalline metals (Fig. 8.2) (for a review see Ref. 69) and which is combined with abnormal crystallite growth, presumably arises from interface pinning due to solute impurities, impurity-related second phases, or residual porosity. (See Ref. 70 for a review on drag mechanisms.) Porosity-retarded crystallite growth[71] is discussed for nanocrystalline ceramics.[72] It may also prevail in nanocrystalline metals[73] with relative mass densities, ρ/ρ0, below the theoretical limit. This is supported, for example, by the correlation found between the mass density and (abnormal) grain growth at ambient temperature in cluster-compacted n-Cu.[74] Similarly, combined studies of crystallite growth (Fig. 8.3) and positron lifetime in n-Pd prepared by pulsed electrodeposition suggest a correlation between the thermal stability and the presence of nanopores.[73] A promising strategy for the controlled stabilization of nanocrystalline metals, in addition, for example, to interface pinning by oxide or carbide nanoparticles,[75,76] consists of decreasing the driving force for crystallite growth by means of a doping-induced decrease of the specific grain boundary energy.[77] The diffusion characteristics of Bi in nanocrystalline Cu, as reported by Höfler et al.,[58] are of particular interest in this respect. Since Bi is extremely insoluble in crystalline Cu, it exhibits a strong tendency for segregation at interfaces. Similarly to that observed for conventional grain boundaries, e.g., in Fe-Sn alloys,[78] the segregation gives rise to a decrease of the Bi diffusivity in the interfaces of n-Cu with increasing Bi concentration,[58] which is attributed to a segregationinduced decrease of the specific grain boundary energy. Probably as a consequence of that, a retardation of the crystallite growth occurs in Bi-doped n-Cu in comparison to undoped n-Cu.[79] An even stronger stabilization reported for Zr-doped Pd is also considered to arise from a decrease of the grain boundary energy.[77]

8.4.4 The Nanocrystalline Soft-magnetic Alloys Fe73.5Si13.5B9Nb3Cu1 and Fe90Zr7B3 A way of stabilizing the nanocrystalline structure, similar to that described at the end of the preceding section, is opened up by the use of intergranular amorphous phases. This is accomplished in the case of the soft-magnetic nanocomposite Fe73.5Si13.5B9Nb3Cu1 (FINEMET),[83] which

346

Properties

consists of ordered intermetallic nanocrystallites (D03-Fe3Si, crystallite size d = 12 nm) embedded in an Nb- and B-enriched amorphous matrix.[83,84] This nanocomposite is structurally stable until well above the temperature at which the nanocrystalline structure is formed by crystallization of the initially amorphous alloy (Ta = 810 K) (Fig. 8.2). The thermal stability makes this alloy highly attractive for technical applications as well as for diffusion studies, which are summarized in the following.[17] High-temperature studies of positron lifetime (Fig. 8.5) and 59Fe-tracer diffusion of the FINEMET alloy (Fig. 8.4) show thermal-vacancy formation and rapid self-diffusion in the D03-Fe80Si20-nanocrystallites similar to that observed in single crystals of this intermetallic compound.[86,87] This is concluded from the increase of the diffusivity upon crystallization (Fig. 8.4) and from the high-temperature increase of the positron lifetime, which can be analyzed in terms of competitive positron trapping and annihilation at thermal vacancies in the nanocrystallites and at free volumes in the crystallite interfaces (Fig. 8.5).[17] The rapid Fe-diffusion in the nanocrystallites can be taken into account in a diffusion model in which the crystallites and the interfaces with the

Figure 8.4 Arrhenius plots of 59Fe-tracer diffusivities in nanocrystalline Fe73.5Si13.5B9Nb3Cu1 (FINEMET) crystallized at Tcrys = 793 K ( ), 810 K (Δ), 813 K (), 818 K () and in the relaxed amorphous state (, preannealed at 723 K) prior to crystallization.[17] Literature data are shown for comparison (extrapolations are dotted): Fe diffusion in cluster-compacted nanocrystalline Pd (,[14]), in the ferromagnetic phase of crystalline α-Fe (c-Fe,[85]), in grain boundaries of Fe (g-Fe[78]) as well as in the intermetallic compounds[86] D03-Fe79Si21 and D03-Fe82Si18.[86]

8: Diffusion in Nanocrystalline Materials, Sprengel

347

Figure 8.5 Thermal-vacancy formation in Fe73.5Si13.5B9Nb3Cu1 (FINEMET).[17] Mean positron lifetime, τ¯, in the nanocrystalline state after annealing at 813 K for 1 h () or 21 h () and in the microcrystalline state after annealing at 993 K for 31 h (). The solid curve is a numerical fit taking into account competitive positron trapping and annihilation at thermal vacancies in the nanocrystallites and at free volumes in the crystallite interfaces (for details see Ref. 17). The dotted straight lines show the mean positron lifetime in the absence of thermal vacancy formation assuming a linear temperature variation like that in the free state of pure metals.[17]

amorphous intergranular phase are treated as alternating layers perpendicular to the direction of diffusion (see Ref. 17). This model leads to the conclusion that the diffusivity, DIF, in the amorphous interfaces is similar to that measured in the relaxed amorphous state prior to crystallization (Fig. 8.1), which is characteristic of Fe—B—Si amorphous alloys.[88] The interfacial diffusivity, which is substantially lower than in grain boundaries of coarse-grained or nanocrystalline pure metals (Fig. 8.4), reflects a high packing density of both the amorphous intergranular phase and the interfaces between these interlayers and the D03-Fe80Si20 nanocrystallites of the FINEMET alloy. Metallic amorphous phases exhibit densities higher than those of large-angle boundaries.[89] A dense interface structure in the FINEMET alloy is also concluded from positron lifetime measurements at ambient temperature, which show the existence of the same type of free volumes as in the amorphous state, i.e., free volumes of smaller size than lattice vacancies.[17] These direct indications of a densely packed structure of the crystallite–amorphous interfaces are in accordance with the low specific energy of this type of interface.[90]

348

Properties

Interface diffusivities lower than those of n-metals have been found in other n-alloys prepared by crystallization of melt-spun amorphous ribbons.[91] The combination of a dense interface structure, due to an intergranular amorphous phase, and a high thermal-vacancy concentration in the nanocrystallites of the FINEMET nanocomposite gives rise to the unique behavior in which self-diffusion in the interfacial regions is slower than in the crystallites. The variations of the Fe diffusivity (Fig. 8.4) and the thermal-vacancy formation (Fig. 8.5) with temperature and time of crystallization indicate atomic ordering processes in the nanocrystallites.[17] This is concluded from the fact that, in the coarse-grained intermetallic compound Fe3Si, the Fe diffusivity is observed to increase with increasing Si content or degree of order[86] due to an increase of the thermal vacancy concentration CV.[87] The timescale and temperature range suggest that these atomic ordering processes in FINEMET are controlled by the slow diffusivity on the Si sublattice of the nanocrystallites.[17] It is worthy of note that the annealing behavior of the stress-induced magnetic anisotropy, which is of interest with respect to tailoring the soft-magnetic properties of FINEMET,[6] can be quantitatively understood on the basis of the slow ordering processes mentioned above. The crystallite growth during crystallization of FINEMET, on the other hand, is controlled by the diffusion of large atoms, i.e., Nb, in the amorphous matrix.[8] This can be concluded from the high values of the activation enthalpy and preexponential factor of crystallization, a correlation that is similar to that existing between the activation enthalpy and the preexponential factor of tracer diffusion in amorphous alloys (see Ref. 8). Iron tracer diffusion was studied in soft-magnetic nanocrystalline Fe90Zr7B3 without any influence of porosity, relaxation, or grain growth as the specimens were highly dense and pore-free. The interfacial diffusion characteristics differ substantially from grain boundaries in metals due to the residual intergranular amorphous phase remaining after crystallization, which has a significant impact on the interface diffusion characteristics. The diffusion in this type of interface is characteristic of amorphous alloys and is strongly reduced in comparison to conventional grain boundaries. An indication of a second fast diffusion path similar to that in grain boundaries was found and analyzed within the framework of a two-type interface model. The diffusion results are summarized in Table 8.1. The studies of n-Fe73.5Si13.5B9Nb3Cu1[17] and Fe90Zr7B3[48] described above have led, for the first time, to the detection of vacancy-mediated

8: Diffusion in Nanocrystalline Materials, Sprengel

349

rapid self-diffusion in nanocrystallites. These combined high-temperature studies of positron lifetime and tracer diffusion may be extended to structurally stabilized nanocrystalline metals[75,77] in order to assess the relation between thermal-vacancy formation and diffusion in grain boundaries of metals[33] (cf. section 8.3). Furthermore, diffusion studies on stabilized nanocrystalline metals without concomitant crystallite growth might enable distinction of the various diffusion paths available in nanocrystalline metals, e.g., interfaces and triple-lines between interfaces.

8.4.5 Nanocrystalline Hard-magnetic Nd2Fe14B Compounds Diffusion studies of nanocrystalline alloys and compounds have recently been extended to Nd2Fe14B-based systems, which manifest an intergranular melting transition of Nd-enriched grain boundaries.[51] These studies are motivated by the important role played by intergranular melting in the technological application of this permanent magnetic material. The powder-metallurgical processing of Nd2Fe14B takes advantage of intergranular melting in order to induce crystallographic and thus magnetic texture.[51] The diffusion studies performed on n-Nd2Fe14B shed light on a novel type of diffusion behavior in nanocrystalline materials. Radiotracer measurements carried out with the isotopes 59Fe and 147Nd show a substantial increase of the interface diffusion coefficient DB for temperatures above the intergranular melting transition[42,52,92] (Fig. 8.6). The increase of DB, reaching high values characteristic for diffusivities in bulk melts, occurs over a distinct temperature range, rather than abruptly. Such a gradual change could result either from confinement effects or from initially local melting. Both effects give rise to reduced long-range diffusivities compared to bulk melts, as long as the dimensions of the molten region are small or an interconnected network of liquid regions has not yet been formed during the initial stage of melting. Below the intergranular melting transition, the grain boundary diffusivities in n-Nd2Fe14B are found to be similar to those of α-Fe, within a framework of grain boundary diffusion kinetics of type B with an assumed volume self-diffusivity as in α-Fe. In the case of Nd diffusion experiments the diffusivities were directly determined from experiments in the kinetic regime of type C and are similar to the ones reported for Fe diffusion (see Fig. 6).[92] It turned out that neither the Nd nor the Fe diffusion is rate controlling for the thermodynamically induced generation of the texture observed for magnetic

350

Properties

Figure 8.6 Arrhenius plots of 59Fe-tracer diffusivities in nanocrystalline Fe[43] and the Fe-rich nanocrystalline alloys Fe73.5Si13.5B9Nb3Cu1[50], Fe90Zr7B3,[48] and Nd12.2Fe81.8B6[42] (interface thickness δ = 1 nm). For Fe90Zr7B3 the diffusivities in two types of interfaces ( and ) as well as in the initial amorphous state are shown. The data of n-Fe refer to relaxed gain boundaries. Literature data of Fe diffusion in the ferromagnetic phase of crystalline α-Fe (c-Fe[45a]) and in grain boundaries of Fe (g-Fe[46a]) are shown for comparison (extrapolations are dotted).

Nd2Fe14B compounds. For a more detailed analysis of the diffusion characteristics of ultrafine-grained Nd2Fe14B, measurements of bulk diffusion coefficients in single-crystalline or coarse-grained Nd2Fe14B would be desirable.

8.4.6 Diffusion of Hydrogen in Nanocrystalline Metals The diffusion of hydrogen in nanocrystalline metals and alloys is being intensively studied to explore the application potentials in the field of

8: Diffusion in Nanocrystalline Materials, Sprengel

351

hydrogen storage. The dependence of diffusion or local atomic jumps of hydrogen on the hydrogen concentration CH is used as a local probe technique that yields information on the spectra of available trapping sites in the crystallite interfaces. This information is deduced from measurements of the chemical potential (n-Pd[95]), hydrogen permeation (n-Ni[96]), thermal desorption (n-Fe90Zr10[97,98]), magnetic after-effect (e.g., nFe90Zr10,[98] n-Pd3Fe[99]), internal friction (n-CoZr2,[100] n-Pd[101]), or inelastic neutron scattering (n-Pd[102,103]). Studies of hydrogen spectroscopy on n-Fe90Zr10 by Hirscher et al.[98,104] support the results of positron lifetime spectroscopy and Fe-tracer diffusion according to which the interfaces in nanocrystalline alloys crystallized from amorphous precursors are densely packed similarly to the ¯ , of initial amorphous state. A decrease of the mean activation energy, Q reorientation jumps with increasing hydrogen concentration is observed by measurements of the magnetic after-effect.[98,104] Following a model on ¯ is ascribed to the H diffusion in amorphous alloys,[95] this decrease of Q a successive filling of energetically favorable sites (trapping centers) in the interfaces with hydrogen. The trapping centers are characterized by a continuous spectrum of binding energies with a width similar to that in ¯ −CH relation of nthe initial amorphous state.[104] The shift of the Q Fe90Zr10 to lower hydrogen concentrations in comparison to amorphous Fe90Zr10[104] is due to the fact that hydrogen in the nanocrystalline state is primarily located in the interfaces, the volume fraction of which is reduced compared to the initial amorphous phase.[10] In CoZr2, a slight increase (∼6%) of the mean activation energy of hydrogen jumps is observed upon the amorphous-to-nanocrystalline transition by means of internal friction.[100] In contrast to crystallization-prepared nanocrystalline alloys, the spectrum of hydrogen sites in the interfaces of cluster-compacted metals (nPd[95]) appears to differ from that in amorphous alloys, e.g., a-Pd83Si17.[95] On the one hand, the energy distribution of the H sites as deduced from the variation of the H diffusivity with concentration is enhanced by about 30% compared to a-Pd83Si17.[95] On the other hand, the H diffusivity in nPd at low H concentrations with respect to that in the crystalline state is reduced more strongly than in a-Pd83Si17,[95] which indicates traps in the interfaces of cluster-compacted n-Pd with higher binding energies than in amorphous alloys. This fits the different interface structures that are found for cluster-compacted n-metals and crystallized alloys by means of positron lifetime spectroscopy[105,106] and Fe tracer diffusion (see section 8.5). In the case of n-Pd prepared by pulsed electrodeposition, initial studies by means of quasielastic neutron scattering indicate deep traps in the interfaces.[103]

352

Properties

8.5 Diffusion in Nanocrystalline Ceramics For nanocrystalline ceramics, i.e., transition metal oxides ZrO2[16] and TiO2,[57,107] comprehensive interface diffusion studies have been reported. They are of particular importance since interface diffusion in nanocrystalline ceramics gives rise to enhanced sintering rates and improved deformation characteristics in comparison to coarse-grained ceramics.[4,108] Furthermore, knowledge of interface diffusion in nanocrystalline zirconia, ZrO2,[109] and nanocrystalline yttria-stabilized zirconia, ZrO2·Y2O3 (YSZ),[110] is essential for the application of these materials as solid electrolytes in solid oxide fuel cells (SOF). To achieve high oxygen ion conductivity, fuel cells have to be operated at elevated temperatures up to 1173–1373 K, which, however, makes the electrolytes prone to degradation. Therefore, it is highly desirable to reduce the operating temperature by simultaneously maintaining or even increasing the oxygen ion conductivity, which could be achieved by utilization of interfaces as paths of enhanced diffusivity. Ceramics with stoichiometric compositions, in general, exhibit high activation energies for bulk self-diffusion due to the strong binding of the constituents.[111] Deviations from stoichiometry may provide mechanisms of fast diffusion with a reduced activation energy and, therefore, exert a strong influence on the diffusion behavior. This particularly applies to zirconium dioxide.[112] In this interesting model system both the oxygen diffusivity and the crystallite structure[113] are determined by the presence of cationic dopants with a lower valency, which induce extrinsic vacancies on the oxygen sublattice. Diffusion studies on nanocrystalline ZrO2 are aimed at assessing the effect of crystallite interfaces on the diffusion behavior. The oxygen diffusion experiments reported in the following for nanocrystalline ZrO2-based compounds were all performed using 18O as a tracer and secondary-ion mass spectroscopy (SIMS) for measuring the concentration-penetration profiles (for experimental details see Ref. 16) The undoped nanocrystalline ZrO2 (n-ZrO2) samples were prepared by inert-gas condensation, post-oxidation, and in-situ consolidation at ambient temperature with an axial pressure of 1.8 GPa.[114] A relative mass density of about 97% and an average grain size, d, of 80 nm were obtained by subsequent pressureless sintering at Tsinter = 1223–1243 K for 2–3 hours.[16] Further sintering at Ta = 1323 K for 12 hours led to nearly full density and an increase of the grain size to 300 nm. The shape of the 18O diffusion profiles indicates two concomitant diffusion processes, i.e., volume diffusion in the crystallites (DV) and interface diffusion (DB). Analyses of the diffusion profiles according to type-B

8: Diffusion in Nanocrystalline Materials, Sprengel

353

diffusion kinetics (see section 8.2) reveal interface diffusivities, DB, 3 to 4 orders of magnitude higher than the volume diffusivities, DV, within the entire temperature range of 723–1223 K studied[16] (Fig. 8.7). The volume diffusion coefficients, DV = (2.5 ± 1.5) × 10−7 exp[−(2.29 ± 0.1) eV/kBT] m2 s−1 were determined from the n-ZrO2 samples with the larger grain size (Tsinter = 1323 K) where the surface-near parts of the 18Odiffusion profiles up to depths of several 100 nm are dominated by the diffusion from the surface into the volume of the crystallites. The interface diffusion coefficients, DB = (3.3 ± 1.5) × 10−5 exp[−(1.95 ± 0.05) eV/kBT] m2 s−1, determined from the slopes at deeper penetration are almost identical for both sintering temperatures (Fig. 8.7) and are independent of grain size. They show that the diffusion of oxygen in interfaces is 103 to 104 times faster than in the bulk of the crystallites.[109]

Figure 8.7 Arrhenius plot of the 18O tracer diffusion coefficients DV and DB measured in undoped, monoclinic n-ZrO2 with a grain size of 80–300 nm.[16] Oxygen diffusion in undoped ZrO2 (– – –) obtained by gas exchange techniques (I, Madeyski et al.;[115] II, Keneshea et al.,[116]) or thermogravimetry (III, Ikuma et al.[120]), in Ca- (CSZ, – · · –) or Y-stabilized zirconia (– · –) as measured by SIMS profiling (IV, Simpson et al.;[117] V, Tannhauser et al.[118]) or spectroscopy of the exchanged gas (VI, Kim et al.[119]) and in n-TiO2 and bulk (c-)TiO2 (· · ·, Hoefler et al.[57,105]). Values for the cation self-diffusion in YSZ as deduced from the shrinkage of dislocation loops (Heuer et al.[121]) are shown for comparison.

354

Properties

These directly determined oxygen diffusivities in undoped n-ZrO2 support earlier results where diffusivities in coarse-grained undoped monoclinic ZrO2 were indirectly obtained by using gas-exchange techniques[115,116] (Fig. 8.7). The interface diffusion coefficients in n-ZrO2 are lower than the oxygen diffusivities in Ca- (CSZ) or Y-stabilized zirconia (YSZ),[117,119] where high volume diffusion coefficients with a low activation energy, QV = 1.2–1.3 eV, occur due to a high concentration of structural oxygen vacancies (Fig. 8.7). Nevertheless, on the basis of the results on undoped n-ZrO2 it may be anticipated that the oxygen diffusivity in YSZ and CSZ may be further enhanced by the introduction of a high number of crystallite interfaces. The very high oxygen interface diffusivities in undoped monoclinic nanocrystalline ZrO2 raise the question whether the already high oxygen diffusivities in bulk ZrO2·Y2O3 can be further enhanced by the introduction of a large number of crystallite interfaces, as the reason for enhanced oxygen diffusion could be a loosely packed grain boundary structure, which is well established for pure metals[122] or easy vacancy formation within the grain boundaries.[123] The large volume fraction of grain boundaries in nanostructured ZrO2·Y2O3 with grain sizes below 100 nm could result in an increased effective diffusion coefficient. Nanocrystalline ZrO2·Y2O3 specimens with crystallite sizes of 4–5 nm were prepared by DC-sputtering and crystallite condensation and compaction (CC). From the sputter targets Zr92Y8 and Zr84Y16, specimens with the final compositions (ZrO2)96.6(Y2O3)3.4 and (ZrO2)93.1(Y2O3)6.9 were prepared with a negligible contamination of iron and carbon. By X-ray diffraction, the crystal structure after sintering was determined to be tetragonal or cubic for the specimens containing 3.4 or 6.9 mol% Y2O3. Sintering at 1223–1353 K resulted in a densification of the specimens at crystallite sizes of 60–100 nm. Using a geometrical method and the Archimedes method for density determination together with microporosity measurements, it was shown that by sintering at atmospheric pressure, nanoscaled pores are annealed out, resulting in specimens composed of dense nanocrystalline agglomerates with a macroscopic open pore structure in between. This macroporosity can be eliminated by high pressure at elevated temperatures. The 18O tracer concentration at the surface as revealed by SIMS analysis was limited by the gas exchange coefficient. The diffusivities DB in nanocrystalline (ZrO2)93.1(Y2O3)6.9 are higher by three orders of magnitude than the values of the volume diffusivity DV in single-crystalline (ZrO2)90.5(Y2O3)9.5. For the temperature variation of the 18O diffusion coefficient in (ZrO2)90.5(Y2O3)9.5 single crystals. a relationship DV = 8.0 × 10−7 exp(−1.11 eV/kBT) m2 s−1 was derived, whereas for nanocrystalline

8: Diffusion in Nanocrystalline Materials, Sprengel

355

(ZrO2)93.1(Y2O3)6.9 the relationship DB = 2.0 × 10−5 exp(−0.91 eV/kBT) m2 s−1 holds. For a high oxygen diffusion flux, a small crystallite size is desirable in addition to the high value of DB. For both the nanocrystalline specimens and the single crystalline specimens, the temperature variation of the gas exchange coefficient, k, is given by k = 1.4 × 10−2 exp(−1.13 eV/kBT) m2 s−1. An enhancement of the gas exchange coefficient, achievable by a higher hydrostatic oxygen pressure, is highly desirable for making use of the high oxygen diffusivities in n-ZrO2·Y2O3 in future applications. A substantial reduction in the grain size of nanocrystalline ZrO2·Y2O3 can be achieved by admixture of 0.34 mol% Al2O3, which inhibits the grain growth.[124] The reduction in grain size leads to an increase in the volume fraction of grain boundaries and diffusional interfaces. However, the 18O diffusivity in nanocrystalline (ZrO2)93.1(Y2O3)6.56(Al2O3)0.34 is by a factor of 2 smaller than that of nanocrystalline (ZrO2)93.1(Y2O3)6.9 without Al2O3 admixture and larger grain size.[124] The fact that in this case no enhancement of the oxygen diffusion mediated by interface boundaries is observed is attributed to the blocking effect of Al3+ ions in the grain boundary. The Al3+ ions segregate at the grain boundary during the sintering process, inhibiting grain growth and also blocking the flow of oxygen in the diffusion process.[124] A comparison of the results for ZrO2 with the chemically related TiO2 is of particular interest. The diffusion of oxygen was studied in rutile n-TiO2[57,107] with relative densities above 95% and crystallite sizes d ≈ 30 nm after sinter forging at 843 K (pressure of 1.5 GPa). In n-TiO2 the oxygen diffusivity is much faster (105 times) and, accordingly, the diffusion activation energy, QB ≈ 1.5 eV,[57,107] lower than in bulk-TiO2 (QV = 2.58 eV,[125] Fig. 8.7). Although the differences between the bulk and interface O-diffusion are similar in TiO2 and ZrO2, the interface O-diffusion in n-ZrO2 appears to be much faster than in TiO2 when these diffusivities are related to the melting temperatures (Fig. 8.7). Regarding the cation self-diffusion in nanocrystalline ceramics, some information is available from studies of Hf-diffusion in n-TiO2,[57,107] since Hf as a substitutional cation is considered to reflect the Ti self-diffusion. The Hf diffusivities in n-TiO2 are similar to extrapolated values for the Ti volume diffusion in TiO2 based on data of Hoshino et al.,[126] The cationic self-diffusion in rutile is considered to occur via an interstitial mechanism.[111,126] This view is based on the tendency of TiO2 to exhibit an excess of the metallic constituent[111] and on the observation that the Ti selfdiffusivity in rutile increases under reducing conditions due to the formation of Ti interstitials.[126] The results on the Hf-diffusion in n-TiO2 indicate that self-diffusion via self-interstitials in interfaces is not

356

Properties

enhanced in comparison to volume diffusion in crystals. This situation in TiO2 appears to differ from metals, where diffusion via self-interstitials may contribute to the fast grain boundary diffusion.[34] (see section 8.3) Compared to TiO2, the cationic self-diffusion in ZrO2 is characterized by slower bulk diffusivities with a higher activation energy (4–5 eV) as deduced from radiotracer experiments (CSZ[127]) and the shrinking of dislocation rings (YSZ[121]). Some indirect information on the cation selfdiffusion may also be deduced from sintering processes where both the cation and the anion diffusion are involved. The onset temperatures of full densification and massive grain growth are substantially higher for nZrO2[16] than for n-TiO2.[57,104] This again indicates that cation diffusivities in ZrO2 are lower than in n-TiO2. On the other hand, the fact that n-ZrO2 can be sintered to full density at temperatures below 0.45Tm has to be regarded as evidence that in ZrO2, in contrast to n-TiO2, the cation diffusivity in the interfaces is enhanced compared to that in the crystalline state. This item could be further clarified by studies of the Zr self-diffusion in fully sintered n-ZrO2 where both the grain boundary and volume diffusion should be accessible. For a complete understanding of the atomic processes involved in diffusion in nanocrystalline ceramics, i.e., ionic conductors, of which the structures are more complex than for metals, the above described results might ideally be accompanied by impedance spectroscopy and short-range diffusion techniques monitoring local atomic hopping processes. Comprehensive studies on diffusion in nanocrystalline Li ion conductors based on Li2O employing conductivity and nuclear magnetic resonance (NMR) techniques have been reported by Heitjans and coworkers and are summarized in Ref. 128.

Acknowledgments The author is indebted to R. Würschum, TU-Graz, Austria, for fruitful collaboration and discussion. Part of the work was financially supported by Deutsche Forschungsgemeinschaft (Wu 191/4-1,2).

References 1. Birringer, R., Gleiter, H., Klein, H.P., and Marquardt, P., “Nanocrystalline Materials: An Approach to a Novel Solid Structure With Gas-Like Disorder?” Phys. Lett. A, 102:365–369 (1984). 2. Edelstein, A.S., and Cammarata, R.C., (eds.), Nanomaterials: Synthesis, Properties, and Applications, Institute of Physics, Bristol (1996).

8: Diffusion in Nanocrystalline Materials, Sprengel

357

3. Padmanabhan, K.A., Mondal, P., and Hahn, H., “Plastic Deformation in Nanocrystalline Materials,” Nanostruct. Mater., 9:603–606 (1997). 4. Mayo, M.J., “Processing of Nanocrystalline Ceramics from Ultrafine Particles,” Int. Mater. Rev., 41:85–115 (1996). 5. Altstetter, C., “Superplasticity of Nanophase Intermetallics?” in: Mechanical Properties and Deformation Behavior of Materials Having Ultrafine Microstructure, (M.A. Nastasi, D.M. Parker, and H. Gleiter, eds.), E233:381–399, NATO ASI Series, Kluwer Academic Press, Dordrecht (1993). 6. Hofmann, B., and Kronmüller, H., “Stress-Induced Magnetic Anisotropy in Nanocrystalline FeCuNbSi Alloy,” J. Magn. Magn. Mater., 152:91–98 (1996). 7. Grünberger, W., Bürger, G., Hinz, D., Müller, K.H., and Schultz, L., “Warmumformung Nanokristalliner NdFeB-Legierungen,” Verhandl. DPG (VI), 31:AMV (1996). Cao, L., Müller, K.H., Handstein, A., Grünberger, W., Neu, V., and Schultz, L., “High Performance Sm2Fe15Ga2C2 Permanent Magnets Made by Mechanical Alloying and Hot Pressing,” J. Phys. D: Appl. Phys., 29:271–272 (1996). 8. Damson, B., and Würschum, R., “Correlation Between the Kinetics of the Amorphous-to-Nanocrystalline Transformation and the Diffusion in Alloys,” J. Appl. Phys., 80:747–751 (1996). 9. Horvath, J., Birringer, R., and Gleiter, H., “Diffusion in Nanocrystalline Material,” Solid State Commun., 62:319–322 (1987). 10. Würschum, R., Struktur und Diffusionseigenschaften Nanokristalliner Metalle und Legierungen, Habilitation Thesis, University of Stuttgart, Germany (1997). 11. Gleiter, H., “Diffusion in Nanostructured Metals,” Phys. Status Solidi B, 172:41–51 (1992). 12. Schaefer, H.E., “Nanostructured Solids—Diffusional Properties and Nanocrystalline Metal Oxides,” in: Defects in Insulating Materials (O. Kanert, and J.M. Spaeth, eds.), pp. 122–136, World Scientific, Singapore (1993). 13. Larikov, L.N., “Diffusion in Nanocrystalline Materials,” Met. Phys. Adv. Techn., 15:1–37 (1995). 14. Würschum, R., Reimann, K., and Farber, P., “Correlation Between Interfacial Structure, Tracer Diffusion, and Crystal Growth in Nanocrystalline Metals and Alloys,” Defect and Diffusion Forum, 143–147:1463–1468 (1997). 15. Würschum, R., Reimann, K., Gruß, S., Kübler, A., Scharwaechter, P., Frank, W., Kruse, O., Carstanjen, H.D., and Schaefer, H.E., “Structure and Diffusional Properties of Nanocrystalline Pd,” Philos. Mag. B, 76:407–417 (1997). 16. Brossmann, U., Untersuchung der Sauerstoffdiffusion in Nanokristallinem Monoklinem Zirkondioxid, Dissertation, University of Stuttgart, Germany (1998). Ossmann, U., Würschum, R., Södervall, U., and Schaefer, H.E., “Oxygen Diffusion in Ultra-Fine Grained Monoclinic ZrO2,” J. Appl. Phys., 85:7646–7654 (1999). 17. Würschum, R., Farber, P., Dittmar, R., Scharwaechter, P., Frank, W., and Schaefer, H.E., “Thermal-Vacancy Formation and Self-Diffusion in Inter-

358

18.

19. 20. 21. 22. 23. 24. 25.

26. 27. 28.

29. 30. 31. 32.

Properties metallic Fe3Si-Crystallites of Nanocomposite Alloys,” Phys. Rev. Lett., 79:4918–4921 (1997). Würschum, R., Kübler, A., Gruß, S., Scharwaechter, P., Frank, W., Valiev, R.Z., Mulyukov, R.R., and Schaefer, H.E., “Tracer Diffusion and Crystallite Growth in Ultra-Fine Grained Pd Prepared by Severe Plastic Deformation,” Annales de Chimie–Science des Matériaux, 21:471–482 (1996). Surholt, T., Mishin, Y.M., and Herzig, C., “Grain Boundary Diffusion and Segregation of Gold in Copper: Investigation in the Type-B and Type-C Kinetic Regimes,” Phys. Rev. B, 50:3577–3587 (1994). Surholt, T., and Herzig, C., “Grain Boundary Self- and Solute-Diffusion and Segregation in General Large Angle Grain Boundaries in Copper,” Defect and Diffusion Forum, 143–147:1391–1396 (1997). Qing Ma, and Balluffi, R.W., “Diffusion Along [001] Tilt Boundaries in the Au/Ag System, I. Experimental Results,” Acta Metall. Mater., 41:133–141 (1993). Harrison, L.G., “Influence of Dislocations on Diffusion Kinetics in Solids With Particular Reference to the Alkali Halides,” Trans. Faraday Soc., 57:1191–1199 (1961). Levine, H.S., and MacCallum, C.J., “Grain Boundary and Lattice Diffusion in Polycrystalline Bodies,” J. Appl. Phys., 31:595–599 (1960). Kaur, I., Mishin, Y., and Gust, W., Fundamentals of Grain and Interphase Boundary Diffusion, Wiley, Chichester (1995). Glaeser, A.M., and Evans, J.W., “Effect of Grain Boundary Migration on Apparent Boundary Diffusion Coefficients,” Acta Metall., 34:1545–1552 (1986). Güthoff, F., Mishin, Y., and Herzig, C., “Self-diffusion Along Stationary and Moving Grain Boundaries in α-Hf,” Z. Metallkd., 84:584–591 (1993). Klinger, L., Levin, L., and Petelin, A., “A Model of Accelerated Diffusion Along Triple Junction of Grain Boundaries,” Mater. Sci. Forum, 207–209:501–504 (1996). Bokstein, B.S., Bröse, H.D., Trusov, L.I., and Khvostantseva, T.P., “Diffusion in Nanocrystalline Ni,” Nanostruct. Mater., 6:873–876 (1995). Divinski, S.V., Hisker, F., Kang, Y.-S., Lee, J.-S., and Herzig, C., “Fe-59 Grain Boundary Diffusion in Nanostructured gamma-Fe –Ni–Part I: Radiotracer Experiments and Monte-Carlo Simulation in the Type-A and B Kinetic Regimes,” Z. Metallkd., 93:256–264; Part II: Effect of bimodal microstructure on diffusion behavior in type-C kinetic regime Z. Metallkd., 93:265–272 (2002). Sutton, A.P., and Balluffi, R.W., Interfaces in Crystalline Materials, Monographs on the Physics and Chemistry of Materials, Clarendon Press, Oxford (1995). Langmuir, I., “Thoriated Tungsten Filaments,” J. Franklin Inst., 217:543–570 (1934). Martin, G., Blackburn, D.A., and Adda, Y., “Autodiffusion Au Joint de Grains de Bicristaux d’argent Soumis à une Pression Hydrostatique,” Phys. Status Solidi, 23:223–228 (1967). Robinson, J.T., and Peterson, N.L., “Correlation Effects in Grain Boundary Diffusion,” Surf. Sci., 31:586–616 (1972). Robinson, J.T., and Peterson, N.L., Acta Metall., 21:1181–1186 (1973).

8: Diffusion in Nanocrystalline Materials, Sprengel

359

33. Kwok, T., Ho, P.S., Yip, S., Balluffi, R.W., Bristowe, P.D., and Brokman, A., “Evidence for Vacancy Mechanism in Grain Boundary Diffusion in bcc Iron: A Molecular-Dynamics Study,” Phys. Rev. Lett., 47:1148–1151 (1981). Balluffi, R.W., Kwok, T., Bristowe, P.D., Brokman, A., Ho, P.S., and Yip, S., “Determination of Vacancy Mechanism for Grain Boundary Self-Diffusion by Computer Simulation,” Scripta Metall., 15:951–956 (1981). 34. Qing Ma, Liu, C.L., Adams, J.B., and Balluffi, R.W., “Diffusion Along [001] Tilt Boundaries in the Au/Ag System-II. Atomistic Modelling and Interpretation,” Acta Metall. Mater., 41:143–151 (1993). 35. Nomura, M., and Adams, J.B., “Self-Diffusion Along Twist Boundaries in Cu,” J. Mater. Res., 7:3202–3212 (1992). 36. Sommer, J., Herzig, C., Mayer, S., and Gust, W., “Grain Boundary SelfDiffusion in Silver Bicrystals,” Defect and Diffusion Forum, 66–69:843–848 (1989). 37. Budke, E., Orientierungsabhängigkeit der Radiotracer-Diffusion von 195Au, 64 Cu und 75Se in Symmetrischen [001]-Kupferkippkorngrenzen, Dissertation, University of Münster, Germany (1994). 38. Wolf, D., and Merkle, K.L., “Correlation Between the Structure and Energy of Grain Boundaries in Metals,” in: Material Interfaces: Atomic-Level Structure and Properties, (D. Wolf, and S. Yip, eds.), pp. 87–150, Chapman and Hall, London (1992). 39. Seeger, A., and Schottky, G., “Die Energie und der elektrische Widerstand von Großwinkelkorngrenzen in Metallen,” Acta Metall., 7:495–503 (1959). 40. Schumacher, S., Birringer, R., Strauß, R., and Gleiter, H., “Diffusion of Silver in Nanocrystalline Copper Between 303 K and 373 K,” Acta Metall., 37:2485–2488 (1989). 41. Würschum, R., Michel, T., Scharwaechter, P., Frank, W., Schaefer, H.-E., “Fe Diffusion in Nanocrystalline Alloys and the Influence of Amorphous Intergranular Layers,” Nanostruct. Mater., 12:555–558 (1999). 42. Eggersmann, M., Ye, F., Herth, S., Gutfleisch, O., Würschum, R., “Intergranular Melting of Ultrafine Grained Nd2Fe14B Studied by Means of Radiotracer Diffusion,” Interface Sci., 9:337–341 (2001). 43. Tanimoto, H., Farber, P., Würschum, R., Valiev, R.Z., and Schaefer, H.E., “Self-Diffusion in High-Density Nanocrystalline Fe,” Nanostruct. Mater., 12:681–684 (1999). 44. Würschum, R., Structure and Diffusion Properties of Nanocrystalline Metals and Alloys, University Stuttgart, Germany (1997). 45. Kolobov, Y.R., Grabovetskaya, G.P., Ivanov, M.B., Zhilyaev, A.P., Valiev, R.Z., “Grain Boundary Diffusion Characteristics of Nanostructured Nickel,” Scripta Mater., 44:873–878 (2001). 46. Erdélyi, Z., Girardeaux, Ch., Langer, G.A., Beke, D.L., Rolland, A., Bernardini, J., “Determination of Grain-Boundary Diffusion of Ag in Nanocrystalline Cu by the Hwang-Balluffi Method,” J. Appl. Phys., 89:3971–3975 (2001). 47. Ovidko, I.A., and Sheinerman, A.G., “Diffusion Percolation along Triple Junctions in Nanocrystalline Materials,” Rev. Adv. Mater. Sci., 6:41–47 (2004).

360

Properties

48. Herth, S., Eggersmann, M., Eversheim, P.-D. Eversheim, and Würschum, R., “Interface Diffusion and Amorphous Intergranular Layers in Nanocrystalline Fe90Zr7B3,” J. Appl. Phys., 95:5075–5080 (2004). 49. Herth, S., Michel, T., Tanimoto, H., Eggersmann, M., Dittmar, R., Schaefer, H.-E., Frank, W., and Würschum, R., “Self-diffusion in Nanocrystalline Fe and Fe-rich Alloys,” Defect and Diffusion Forum, 194–199:1199–1204 (2001). 50. Herth, S., Eggersmann, M., Herzer, G., and Würschum, R., “Diffusion and Induced Magnetic Anisotropy in Nanocrystalline Fe73.5Si13.5B9Nb3Cu1, Philos. Mag. Lett., 84:531–537 (2004). 51. Grünberger, W., Hinz, D., Kirchner, A., Müller, K.-H., and Schultz, L., “Hot Deformation of Nanocrystalline Nd-Fe-B alloys,” J. Alloys Comp., 257:293–301 (1997). 52. Herth, S., Ye, F., Eggersmann, M., Gutfleisch, O., and Würschum, R., “Self-Diffusion in Liquid Interfaces,” Phys. Rev. Lett., 92:095901/1–4 (2004). 53. Maier, K., “Self-Diffusion in Copper at ‘Low’ Temperatures,” Phys. Status Solidi A, 44:567–576 (1977). 54. Birringer, R., Krill, C.E., and Klingel, M., “Orientation-Phase-Space-Averaged Properties of Grain Boundaries,” Philos. Mag. Lett., 72:71–77 (1995). 55. Reimann, K., and Würschum, R., “Distribution of Internal Strains in Nanocrystalline Pd Studied By X-ray Diffraction,” J. Appl. Phys., 81:7186–7192 (1997). 56. Balandin, I.L., Bokstein, B.S., Egorov, V.K., and Kurkin, P.V., “Antimony Diffusion in Fine Scale Copper,” Defect and Diffusion Forum, 143–147:1475–1480 (1997). 57. Höfler, H.J., Diffusion und Festkörperreaktionen in Nanokristallinen Materialien, Dissertation, University of Saarlandes, Germany (1991) (See also Ref. 11.) 58. Höfler, H.J., Averback, R.S., Hahn, H., and Gleiter, H., “Diffusion of Bismuth and Gold in Nanocrystalline Copper,” J. Appl. Phys., 74:3832–3939 (1993). 59. Bonzel, H.P., and Latta, E.E., “Surface Self-Diffusion on Ni(110): Temperature Dependence and Dislocation Anisotropy,” Surf. Sci., 76:275–295 (1978). 60. Maier, K., Mehrer, H., Lessmann, E., and Schüle, W., “Self-Diffusion in Nickel at Low Temperatures,” Phys. Status Solidi B, 78:689–698 (1976). 61. Peterson, N.L., “Isotope Effect in Self-Diffusion in Palladium,” Phys. Rev., 136:568–574 (1964). 62. Gust, W., Mayer, S., Bögel, A., and Predel, B., “Generalized Representation of Grain Boundary Self-Diffusion Data,” J. de Physique, 46 Colloq., C4:537–544 (1985). 63. Höfler, H.J., Averback, R.S., and Gleiter, H., “Diffusion of Boron in Nanocrystalline Iron–A New Type of Diffusion Kinetics—Type-C,” Philos. Mag. Lett., 68:99–105 (1993). 64. Sommer, J., and Herzig, C., “Direct Determination of Grain-Boundary and Dislocation Self-Diffusion Coefficients in Silver from Experiments in TypeC Kinetics,” J. Appl. Phys., 72:2758–2766 (1992).

8: Diffusion in Nanocrystalline Materials, Sprengel

361

65. Ma, Q., and Balluffi, R.W., “Diffusion along [001] Tilt Boundaries in the Au/Ag System 1. Experimental Results,” Acta Metall. Mater., 41:133–141 (1993). 66. Minamino, Y., Saji, S., Hirao, K., Ogawa, K., Araki, H., Miyamoto, Y., and Yamane, T., “Diffusion of Copper in Nanocrystalline Al-7.8 at%Ti-0.3 at% Fe Alloy Prepared by Mechanical Alloying,” Mater. Trans. JIM, 37:130–137 (1996). 67. Fujita, T., Hasegawa, H., Horita, Z., and Langdon, T.G., “Diffusion in Ultrafine-Grained Al-Mg alloys,” Defect and Diffusion Forum, 194–199:1205–1210 (2001). 68. Babcock, S.E., and Balluffi, R.W., “Grain Boundary Kinetics-II: In-situ Observation of the Role of Grain Boundary Dislocations in High-Angle Grain Boundary Migration,” Acta Metall., 37:2367–2376 (1989). 69. Malow, T.R., and Koch, C.C., “Grain Growth of Nanocrystalline Materials–A Review,” in: Synthesis and Processing of Nanocrystalline Powder (D. L. Bourrell, ed.), p. 33, TMS, Warrendale, PA (1996). 70. Humphreys, F.J., and Hatherly, M., Recrystallization and Related Annealing Phenomena, Pergamon, Elsevier, Oxford (1995). 71. Liu, Y., and Patterson, R.R., “Grain Growth Inhibition by Porosity,” Acta Metall. Mater., 41:2651–2656 (1993). 72. Mayo, M.J., and Hague, D.C., “Porosity–Grain Growth Relationships in the Sintering of Nanocrystalline Ceramics,” Nanostruct. Mater., 3:43–52 (1993). 73. Würschum, R., Gruß, S., Gissibl, B., Natter, H., Hempelmann, R., and Schaefer, H.-E., “Free Volumes and Thermal Stability of Electro-Deposited Nanocrystalline Pd,” Nanostruct. Mater., 9:615–618 (1997). 74. Gertsman, V.Y., and Birringer, R., “On the Room-Temperature Grain Growth in Nanocrystalline Copper,” Scripta Metall. Mater., 30:577–581 (1994). 75. Lebedev, A.B., Pulnev, S.A., Kopylov, V.I., Burenkov, Y.A., Vetrov, V.V., and Vylegzhanin, O.V., “Thermal Stability of Submicrocrystalline Copper and Cu:ZrO2 Composite,” Scripta Mater., 35:1077–1080 (1996). 76. Morris, D.G., and Morris, M.A., “Microstructure and Strength of Nanocrystalline Copper Alloy Prepared by Mechanical Alloying,” Acta Metall. Mater., 39:1763–1770 (1991). 77. Krill, C.E., Klein, R., Janes, S., and Birringer, R., “Thermodynamic Stabilization of Grain Boundaries in Nanocrystalline Alloys,” Mater. Sci. Forum, 179–181:443–448 (1995). 78. Bernardini, J., Gas, P., Hondros, E.D., and Seah, M.P., “The Role of Solute Segregation in Grain Boundary Diffusion,” Proc. Roy. Soc. London A, 379:159–178 (1982). 79. Konrad, H., Haubold, T., Birringer, R., and Gleiter, H., “Nanostructured Cu–Bi Alloys Prepared by Co-evaporation in a Continuous Gas Flow,” Nanostruct. Mater., 7:605–610 (1996). 80. Akhamadeev, N.A., Kobelev, N.P., Mulyukov, R.R., Soifer, Y.M., and Valiev, R.Z., “The Effect of Heat Treatment on the Elastic and Dissipative Properties of Copper with the Submicrocrystalline Structure,” Acta Metall. Mater., 41:1041–1046 (1993). Jianshe Lian, Valiev, R.Z., and Baudelet, B.,

362

81. 82. 83.

84.

85. 86. 87. 88. 89. 90. 91.

92.

93. 94.

Properties “On the Enhanced Grain Growth in Ultrafine Grained Metals,” Acta Metall. Mater., 43:4165–4170 (1995). Sanders, P.G., Weertman, J.R., Barker, J.G., and Siegel, R.W., “Small Angle Neutron Scattering From Nanocrystalline Palladium as a Function of Annealing,” Scripta Metall. Mater., 29:91–96 (1993). Hofmann, B., Reininger, T., and Kronmüller, H., “Influence of the Microstructure on the Magnetization Processes in Nanocrystalline Fe73.5Si13.5B9Nb3Cu1,” Phys. Status Solidi A, 134:247–261 (1992). Yoshizawa, Y., Oguma, S., and Yamauchi, K., “New Fe-Based Soft Magnetic Alloys Composed of Ultrafine Grain Structure,” J. Appl. Phys., 64:6044–6046 (1988). Herzer, G., “Grain Structure and Magnetism of Nanocrystalline Ferromagnets,” IEEE Tran. Magnetics, 25:3327–3329 (1989). Hampel, G., Pundt, A., and Hesse, J., “Crystallization of Fe73.5Si13.5B9Nb3Cu1: Structure and Kinetics Examined by X-ray Diffraction and Mössbauer Spectroscopy,” J. Phys. C: Condens. Matter, 4:3195–3214 (1992). Lübbehusen, M., and Mehrer, H., “Self-Diffusion in α-Iron: The Influence of Dislocations and the Effect of the Magnetic Phase Transition,” Acta Metall. Mater., 38:283–292 (1990). Gude, A., and Mehrer, H., “Diffusion in the D03 Type Intermetallic Phase Fe3Si,” Philos. Mag. A, 76:1–30 (1997). Kümmerle, E.A., Badura, K., Sepiol, B., Mehrer, H., and Schaefer, H.E., “Thermal Formation of Vacancies in Fe3Si,” Phys. Rev. B, 52:R6947–R6950 (1995). Ulfert, W., Horváth, J., and Frank, W., “Self-Diffusion of 59Fe Tracer Atoms in Amorphous Fe78Si9B13 and Fe40Ni40B20,” Cryst. Lattice Def. Amorph. Mater., 18:519–531 (1989). Wang, J., Wolf, D., Phillpot, S.R., and Gleiter, H., “Computer Simulation of the Structure and Thermo-Elastic Properties of a Model Nanocrystalline Material,” Philos. Mag. A, 73:517–555 (1996). Nishi, Y., “Solid-Undercooled Liquid Interfacial Energy of Pd—Cu—Si Alloy Glass,” in: Rapidly Quenched Metals (S. Steeb, and H. Warlimont, eds.), pp. 231–234, Elsevier, Amsterdam (1985). Herth, S., Michel, T., Tanimoto, H., Eggersmann, M., Dittmar, R., Schaefer, H.-E., Frank, W., and Würschum, R., “Self-Diffusion in Nanocrystalline Fe and Fe-Rich Alloys,” Defect and Diffusion Forum, Proc. Int. Conf. on Diffusion in Materials, 194–199:1199–1204 (2001). Sprengel, W., Herth, S., Barbe, V., Schaefer, H.-E., Wejrzanowski, T., Gutfleisch, O., and Würschum, R., “Grain-Boundary Diffusion of 147 Nd in Nanocrystalline Nd2Fe14B,” J. Appl. Phys., 98:074314/1–5. (2005). Lübbehusen, M., and Mehrer, H., “Self-Diffusion in alpha-Iron–The Influence of Dislocations and the Effect of the Magnetic Phase-Transition,” Acta Metall. Mater., 38:283–292 (1990). Bernardini, J., Gas, P., Hondros, E.D., and Seah, M.P., “The Role of Solute Segregation in Grain-Boundary Diffusion,” Proc. Roy. Soc. Lond. A, 379:159–178 (1982).

8: Diffusion in Nanocrystalline Materials, Sprengel

363

95. Kirchheim, R., Mütschele, T., Kieninger, W., Gleiter, H., Birringer, R., and Koblé, T.D., “Hydrogen in Amorphous and Nanocrystalline Metals,” Mater. Sci. Eng., 99:457–462 (1988). 96. Palumbo, G., Doyle, D.M., El-Sherik, A.M., Erb, U., and Aust, K.T., “Intercrystalline Hydrogen Transport in Nanocrystalline Nickel,” Scripta Metall. Mater., 25:679–684 (1991). 97. Maier, C.U., and Kronmüller, H., “Diffusion and Relaxation Processes of Hydrogen in Heterogenous Microcrystalline Fe-Zr Alloys,” Z. Metallkd., 84:410–418 (1993). 98. Hirscher, M., Zimmer, S., and Kronmüller, H., “Diffusion of Hydrogen in Nanocrystalline Fe90Zr10,” Z. Phys. Chem., 183:51–58 (1994). 99. Mössinger, J., Hirscher, M., and Kronmüller, H., “Diffusion of Hydrogen in Nanocrystalline Pd3Fe,” Philos. Mag. B, 73:503–510 (1996). 100. Sinning, H.-R., “Mechanical Spectroscopy With Hydrogen in Intermetallic Phases,” J. Alloys Comp., 211/212:216–221 (1994). 101. Kappesser, B., Stuhr, U., Wipf, H., Weissmüller, J., Clos, C., and Gleiter, H., “Hydrogen-Induced Internal Friction in Nanocrystalline Palladium,” J. Alloys Compounds, 231:337–342 (1995). 102. Stuhr, U., Wipf, H., Udovic, T.J., Weissmüller, J., and Gleiter, H., “Inelastic Neutron Scattering of Hydrogen in Nanocrystalline Pd,” Nanostruct. Mater., 6:555–558 (1995). 103. Janßen, S., Natter, H., Hempelmann, R., Striffler, T., Stuhr, U., Wipf, H., Hahn, H., and Cook, J.C., “Hydrogen Diffusion in Nanocrystalline Pd by Means of Quasielastic Neutron Scattering,” Nanostruct. Mater., 9:579–582 (1997). 104. Hirscher, M., “Diffusion of Hydrogen in Amorphous and Nanocrystalline Alloys,” in: Interstitial Intermetallic Alloys (F. Grandjean, et al., eds.), pp. 333–347, Kluwer Academic, Dordrecht (1995). 105. Würschum, R., Greiner, W., Valiev, R.Z., Rapp, M., Sigle, W., Schneeweiss, O., and Schaefer, H.-E., “Interfacial Free Volumes in Ultra-Fine Grained Metals Prepared by Severe Plastic Deformation, by Spark-Erosion, or by Crystallization of Amorphous Alloys,” Scripta Metall. Mater., 25:2451–2456 (1991). 106. Würschum, R., and Schaefer, H.-E., “Interfacial Free Volumes and Atomic Diffusion in Nanostructured Solids,” in Ref. 2, pp. 277–301. 107. Höfler, H.J., Hahn, H., and Averback, R.S., “Diffusion in Nanocrystalline Materials,” Defect and Diffusion Forum, 75:195–210 (1991). 108. Chen, I.W., and Xue, L.A., “Development of Superplastic Structural Ceramics,” J. Am. Ceram. Soc., 73:2585–2609 (1990). 109. Brossmann, U., Würschum, R., Södervall, U., and Schaefer, H.-E., “Oxygen Diffusion in Ultrafine Grained Monoclinic ZrO2,” J. Appl. Phys., 85:7646–7654 (1999). 110. Knöner, G., Reimann, K., Röwer, R., Södervall, U., and Schaefer, H.-E., “Enhanced Oxygen Diffusivity in Interfaces of Nanocrystalline ZrO2·Y2O3,” Proc. Natl. Acad. Sci., 100:3870–3873 (2003). 111. Atkinson, A., “Diffusion in Ceramics,” in: Materials Science and Technology, vol. 11, Structure and Properties of Ceramics (M.V. Swain, ed.), pp. 299–338, VCH Weinheim (1994).

364

Properties

112. Subbarao, E.C., “Zirconia—An Overview,” in: Advances in Ceramics, Science and Technology of Zirkonia (A.H. Heuer, and L.W. Hobbs, eds.), 3:1–13, American Ceramic Society, Columbus, OH (1981). 113. Scott, H.G., “Phase Relationships in the Zirconia–Yttria System,” J. Mater. Sci., 10:1527–1535 (1975). 114. Würschum, R., Soyez, G., and Schaefer, H.-E., “Preparation and Sintering Studies of Nanometer-Sized Polycrystalline ZrO2,” in: Structure and Properties of Interfaces in Materials (W.A.T. Clark, C.L. Briant, and U. Dahmen, eds.), 238:733–738, Mater. Res. Soc. Symp. Proc., Pittsburgh, PA (1992). 115. Madeyski, A., and Smeltzer, W.W., “Oxygen Diffusion in Monoclinic Zirconia,” Mater. Res. Bull., 3:369–375 (1968). 116. Keneshea, F.J., and Douglas, D.L., “The Diffusion of Oxygen in Zirconia as a Function of Oxygen Pressure,” Oxidation of Metals, 3:1–14 (1971). 117. Simpson, L.A., and Carter, R.E., “Oxygen Exchange and Diffusion in Calcia-Stabilized Zirconia,” J. Am. Ceram. Soc., 49:139–144 (1966). 118. Tannhauser, D.S., Kilchner, J.A., and Steele, B.C.H., “The Determination of the Oxygen Self-Diffusion and Gas-Solid Exchange Coefficients for Stabilized Zirconia by SIMS,” Nucl. Instrum. Methods, 281:504–508 (1983). 119. Kim, B.K., Park, S.J., and Hamaguchi, H., “Determination of the Oxygen Self-Diffusion Coefficients in Y2O3-Containing Tetragonal Zirconia Polycrystals by Raman Spectrometric Monitoring of the 16O–18O Exchange Reaction,” J. Am. Ceram. Soc., 76:2119–2122 (1993). 120. Ikuma, Y., Komatsu, K., and Komatsu, W., in: Advances in Ceramics, vol. 24: Science and Technology of Zirkonia III (S. Somiya, N. Yamamoto, and H. Yanagida, eds.), pp. 749–758, American Ceramic Society, Columbus, OH (1988). 121. Chien, F., and Heuer, A., “Lattice Diffusion Kinetics in Y2O3-Stabilized ZrO2 Single Crystals: A Dislocation Loop Annealing Study,” Philos. Mag. A, 73:681–697 (1996). 122. Schaefer, H.-E., Reimann, K., Straub, W., Phillipp, F., Tanimoto, H., Brossmann, U., and Würschum, R., “Interface Structure Studies by Atomic Resolution Electron Microscopy, Order–Disorder Phenomena and Atomic Diffusion in Gas-Phase Synthesized Nanocrystalline Solids,” Mater. Sci. Eng. A, 286:24–33 (2000). 123. Würschum, R., Shapiro, E., Dittmar, R., and Schaefer, H.-E., HighTemperature Studies of Grain Boundaries in Ultrafine Grained Alloys by Means of Positron Lifetime, Phys. Rev. B, 62:12021–12027 (2000). 124. Madubuonu, A., Drings, H., Röwer, R., and Schaefer, H.-E., “Grain Size Reduction in Fully Dense Nanocrystalline Yttria-Stabilized Zirconia by Al Doping,” Phys. Stat. Sol. (a) 203:R64–R65 (2006). 125. Kingery, W., Bowen, H.K., and Uhlmann, D.R., Introduction to Ceramics, Wiley, New York (1976). 126. Hoshino, K., Peterson, N.L., and Wiley, C.L., “Diffusion and Point Defects in TiO2,” J. Phys. Chem. Solids, 46:1397–1411 (1985). 127. Rhodes, W.H., and Carter, R.E., “Cationic Self-Diffusion in CalciaStabilized Zirconia,” J. Am. Ceram. Soc., 49:244–249 (1966). 128. Heitjans, P., and Indris, S., “Diffusion and Ionic Conduction in Nanocrystalline Ceramics,” J. Phys.: Condens. Matter, 15:R1257–R1289 (2003).

9 Nanostructured Materials for Gas Reactive Applications Michel L. Trudeau Chimie et Matériaux, Institut de Recherche d’Hydro-Québec, Varennes, Quebec, Canada

9.1 Introduction The last 15 years have revealed the unique properties of nanostructured materials as reflected by the various chapters in this book, and many technological areas clearly stand to benefit from the development of materials on the nanoscale. However, there is one field already based on nanometer structures that could probably benefit most by such developments: materials used in gas reactive applications. Interestingly, this topic is not the first that comes to mind when discussing nanostructured materials, even if their potential for technological applications is tremendous. It is not clear whether this is because the majority of researchers who first explored the field of nanocrystalline material belonged to science and physics groups, or because chemists and chemical engineers had already been working in this field for so many decades that they did not notice the headline news about nanosize crystals. Examples of the technological importance of nanostructured materials in this field abound. For instance, metallic nanoparticulate highly dispersed heterogeneous catalysts are important to 20% of the GNP of the United States[1] while in Japan in 1986 the number of gas sensors used for home gas-leak alarms, based on SnO2, was about five million units,[2] most of which were based on nanostructured grains. The main purpose of this chapter is to present some of the recent developments in the field of nanostructured materials in some gas reactive applications from a materials science point of view. Three different areas are reviewed: catalysis and electrocatalysis, an area that has been working with nanostructured materials for more than fifty years; semiconductor gas sensors, materials that represent a perfect example of nanostructurerelated properties; and hydrogen storage, an area where, in latter years, nanostructured materials and designs have succeeded in considerably Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 365–438 © 2007 William Andrew, Inc.

365

366

Properties

improving the hydrogen absorption and desorption cycle and, in view of the recent increase in oil prices, could be the key for a more rapid use of hydrogen as a fuel than one would have anticipated just a couple of years ago. The main objective is to briefly describe some of the basic principles in order to show the advantages of nanostructured materials and to describe recent synthesis processes of new nanostructured materials. The purpose of this chapter is not to go into detail about the different chemical processes and reactions, neither is it to present all the recent studies that have used nanomaterials, an impossible task in view of the large amount of researches done in recent years, but rather to show nanostructured materials developments in regard to the large potential of these fields and to present the already diverse studies that have emerged. It is hoped that these examples will encourage researchers from different branches of materials science to combine their efforts to produce new materials and to “nanodesign” new structures that will show improved or unique properties.

9.2 Catalysis and Electrocatalysis Fundamentally, catalysis involves a cyclic process in which a site on a catalyst forms a complex with reactants from which products are desorbed, thereby restoring the original site and continuing the cycle.[3] In these materials, the active site of a reaction may be a group or cluster of neighboring atoms on the catalyst surface, or a species adsorbed onto the catalyst. These sites are associated with surface structures, lattice defects, and edges or corners of crystallites. Using a number of different synthesis processes, it is now possible to derive nanocrystals with a specific average crystallite size, and to do so with some control over aspects such as the lattice parameters and the amount or type of internal defects with a metastable structure quite different from what is expected with a normal equilibrium process. This will result, for instance, in an extension of the concentration range of a structural phase or of the solubility limit of an element. Furthermore, as the crystallite size decreases below 10 nm and even below 5 nm, surface atoms will start to dominate while, at the same time, other structures, such as triple junctions, will be more and more present. The surface design of materials, as well as the increased presence of these structures, could have a large impact on the properties of catalytic materials. One reason could be changes to the electronic structure for nanoscale particles, since there seems to be a relationship between the electronic structure and the reactivity of metal surfaces, in a way that

9: Gas Reactive Applications, Trudeau

367

could explain why, when the coordination of surface atoms decreases, the adsorption energy and the dissociation rate increase.[4] Nanostructured materials are not something new in catalysis or electrocatalysis. Metallic nanoparticles with an average grain size between 1 and 20 nm, such as Pt or Rh, dispersed on various substrates such as SiO2 or Al2O3 have been used for the past fifty years in heterogeneous catalysis.[1,5,6] These materials are used extensively in many industries such as petrochemical production, automobile emission control, and finechemicals synthesis. The exciting news generated by recent work on nanostructured materials is that it is now possible to consciously “nanodesign” the structure of a material in order to improve its catalytic response or to better understand a particular catalytic process. Because nanostructured materials have played a major role in catalysis for the past fifty years, this section is far from an exhaustive survey of all the relevant studies that show the impact of a crystalline structure on the nanometer scale. Its main purpose is to look at recent studies based on new materials synthesis techniques that have been developed and that show the importance of nanoscale materials for the development of catalytic research and the related industries.

9.2.1 Impact of Structure on Catalysis and Electrocatalysis Processes Before looking at the different methods for producing nanoscale particles, it is important to consider briefly how the structure can be related to a catalytic reaction. To arrive at new catalysts with nanostructure-level design, different phenomena need to be considered:[1,7] the cluster size and distribution, the sensitivity of the reaction to the structure of the catalyst (crystalline structure, interatomic spacing, number of defects, etc.), the thermodynamic properties (surface energy, heat of adsorption), the electronic state of the active site, the interaction with the support (if one is used), and the spillover of one reaction species to another component of the catalyst or to the catalytic support.[8] It is well understood that many catalytic reactions depend on the structure of the catalyst, such as its particle size, the amount and types of internal defects, the surface structure,[9–11] and so on. Since it is now well known that all these morphological parameters are influenced by the size of the crystals,[12] it is clear that catalytic activity will also be different for large-crystalline materials (>20 nm) and nanocrystalline compounds (<20 nm), especially at very low crystallite sizes (∼5 nm). Zhdanov and

368

Properties

Kasemo[13] modeled the effect of catalyst size on the reaction kinetics and have shown that the reaction on small nanometer particles could be very different from that on larger surfaces. Such behavior was, in fact, observed by Peuckert et al.,[14]who found an optimum size of about 3.5 nm for the oxygen reduction kinetics on carbon-supported Pt particles in acid electrolytes. For their part, Zhao et al.,[15] and Feng et al.[16] have also shown that the catalytic activity of a 3-nm iron oxide catalyst used for direct coal liquefaction was greatly dependent on its surface chemistry and, in particular, on the number of water molecules adsorbed at the surface of coordinately unsaturated sites. Another good example of the importance of the crystallite size was presented by Wang et al.[17] for the photocatalytic decomposition of CHCl3 by nanocrystalline TiO2 ranging in size from 6 to 21 nm. They found a blue shift of the absorption edge with decreasing particle size in optical absorption, confirming that quantum-sized particles were produced, and determined that particles with an average size of 11 nm had the highest photoactivity. Also, as catalytic materials are designed, the structure sensitivity of some reactions has to be considered. Davis and Klabunde[18] have shown that some reactions depend on the presence of a single active surface but, in other cases, on several adjacent active sites. Another example is the enhanced adsorption at low pressure of CO2 and SO2 of nanoscale MgO (4 nm) produced by an autoclave compared to larger (150 nm) crystalline materials, owing to the abundance of edge and corner sites and crystal planes.[19] One of the best recent examples of the impact of structures at the nanometer level on catalytic properties is gold, which was thought to be an inert material. Indeed, gold has been the subject of renewed interest in recent years since it was demonstrated to be catalytically active at dimensions below 5 nm and at lower temperature than for other catalysts for reaction such as CO and benzene oxidation, hydrogenation, partial oxidation of hydrocarbons and selective oxidation of higher alkenes.[20–24] This enhanced oxidation reactions activity for nanocrystalline gold could be related to a stronger bonding toward oxygen and the concomitant enhanced dissociative adsorption rate of O2.[11] In heterogeneous catalysis, the active catalytic component may be a metal or a metal oxide, which is usually present in small quantities. It relies on a relatively inert support component to disperse it and to stabilize it as a reactive phase. The importance of the catalytic support as well as the amount of catalytic material has been reported in many studies.[25–28] Lalande et al.[29] have shown that, for nanocrystalline Co particles supported on high-surface-area carbon used for the electrocatalytic oxygen reduction reaction in fuel cells, there is an optimum Co concentration.

9: Gas Reactive Applications, Trudeau

369

These nanoparticles were produced by the pyrolysis of cobalt phthalocyanine (CoPc). The maximum activity, achieved at 3.5 wt% of Co, corresponded to full monolayer coverage of the CoPc on the support, indicating that the activity of the nano-Co was directly affected by its interaction with the carbon support. In another example, Yao et al.[30] found that, during the thermal processing of a Pt catalyst supported on TiO2 (Pt/TiO2), there was a catalyst—support interaction that led to the formation of a Ti suboxide coating at the Pt particle surface, resulting in a decrease in the catalyst activity. No such interaction was found by Zafiris and Gorte[31] in the case of Pt/amorphous CeO2. Yao et al.[30] noted hat the structure of the Pt particles was dependent on the nature of the support used. For example, in the case of TiO2 or CeO2 support, the small Pt particles (diameter about 4 nm) were well-faceted cuboctahedral, whereas Pt particles on Al2O3 were found to be more spherical. Fukushima et al.[32] found that Au 1 nm clusters prepared by an ionized cluster beam had a catalytic activity for the oxidation of CO that was dependent on both the size of the clusters and the nature of the support. Catalytic activity at 150°C was higher than for platinum catalyst at 300°C for Au islands on TiO2 anatase. If the substrate used was TiO2 rutile, the activity was found to be reduced by half. A recent work has also demonstrated the impact on ethane hydrogenolysis of the internal strain produced in Ni nanocrystallites by the strong interaction with the graphene edges of the carbon nanofibers used for support.[33] In addition, the pore structure of the support can contribute to the selectivity of a catalytic reaction by allowing only molecules that are smaller than the pore opening to gain access to the catalytic sites or to be produced. Detailed calculations of different bimetallic clusters have shown the possibility of variation in the surface site composition as a function of the bimetal composition.[34,35] The interaction between a catalytic material and its support has also been modeled in some cases.[36] One interesting result is the size dependence of this interaction. For alloy particles, surface segregation was shown to be affected by the particle size. For instance, in systems where segregation of one component is expected in normal coarse-grained particles, a reduction in segregation will result from grain size decrease. This behavior was also demonstrated to be dependent on the bond strength between the different alloy components and the support. For example, an increase in segregation with increased crystallite size was observed if both elements bonded equally strongly to the support. A reverse of this trend would be observed if the segregating component had a stronger bond with the support than the nonsegregating one. Control of the particle size and its relation to the designed surface composition can,

Properties

370

thus, be seen as an important factor for the development of new catalysts, as was found for PtRu nanocatalysts prepared by different methods for methanol fuel cell applications, which revealed the importance of alloy homogeneity.[37]

9.2.2 Nanostructure Design The development of gas-phase condensation has led to fundamental advances in the synthesis of new catalytic materials.[38] Some of the first results, presented by Beck and Siegel, studied the activity for the dissociative adsorption of H2S in an H2 environment on nanocrystalline TiO2, produced by the gas-phase evaporation of Ti, followed by postoxidation.[39] The nano-TiO2 had a rutile structure with an average diameter of 12 nm and a specific surface area of 76 m2/g. Better initial activity was found for the nanocrystalline materials than for the commercially available materials. This was partially attributed to the high surface area of the gas-phase-condensed rutile and, also, to its high oxygen deficiency, suggesting a relatively high concentration of anion vacancies or defects in the structure. These anion vacancies were shown to be partly responsible for the dissociation of the H2S at the surface, which was verified by oxygen annealing at 800°C. After a 10-minute treatment, the surface area of the nano-TiO2 did not decrease but the stoichiometry showed a measurable increase, which corresponded to a reduction in the overall activity of this catalyst. Sarkas et al.[40] looked at the catalytic activity of nanocrystalline 5 wt% Li—MgO produced by co-evaporation of Li and Mg metals followed by post-oxidation at room temperature for 12 hours in 20% wt% O2/He. The resulting crystallites had an average diameter of about 5.2 nm with a surface area of 750 m2/g. The study of this nanocomposite catalyst for the oxidative conversion of methane to higher hydrocarbons revealed excellent results. The nanocomposite provided catalytic activity at least 200°C below the temperature requirements of conventional Li-impregnated MgO catalysts. At higher temperatures (above 300°C), even though some sintering occurred, the average activity of the nanostructured material was about 3.3 times greater than that of the conventional sample. In another example, nonstoichiometric and highly surface-reactive nanocrystalline cerium oxides have been tailored with a view to catalyzing redox reactions[25,41] such as selective SO2 reduction by CO: SO2 + 2CO → S + 2CO2

Eq. (9-1)

9: Gas Reactive Applications, Trudeau

371

For this reaction, the surface of an oxide catalyst first needed to be reduced by oxidation of the adsorbed CO species. The oxygen vacancy produced in the catalyst could then enable an adsorbed SO2 species to be reduced to SO. If there were another oxygen vacancy in the vicinity, then the SO species could be further reduced to elemental sulfur. Cerium oxide was the catalytic material of choice because of its high oxygen mobility. The rare earth Ce may exist in +3 and +4 oxidation states. Reduction of CeO2, however, typically requires high-temperature treatment, which will eliminate the surface area of the catalyst. Through magnetron sputtering of Ce in Ar followed by controlled post-oxidation, Tschöpe and Ying were able to derive unique CeO2−x nanocrystals with a high surface area and oxygen vacancy concentration.[41,42] X-ray photoelectron spectroscopy (XPS) indicated that the as-prepared sample consisted of a 22% Ce3+ component and a 78% Ce4+ component, and required an oxidizing heat treatment at ≥500°C to be fully oxidized to CeO2.[43,44] The nanocrystalline material was found to be an excellent catalyst, which selectively reduced SO2 to elemental sulfur with 100% conversion at 500°C[45] compared to the 600°C needed by ultrafine stoichiometric CeO2 powder derived by chemical precipitation.[46] If Cu dopants were used, the SO2 reaction temperature could be further reduced to 420°C.[47] This nanocomposite system was obtained by sputtering from a mixed Cu—Ce target. From controlled post-oxidation, Cu was forced to segregate to the surface of the CeO2−x nanocrystals, since the two components have low miscibility in their oxide phases. The resulting ultrahigh Cu dispersion on CeO2−x nanocrystals[48,49] created many semiconductor Schottky junctions that facilitated the formation of oxygen vacancy. It also improved the CO adsorption of the catalyst through spillover effects such that CO oxidation could occur at the remarkably low temperature of 80°C over the Cu/CeO2−x catalyst.[50] This model nanocomposite catalyst also has excellent poisoning resistance against CO2 and is promising as an effective catalyst for SO2 removal from the industrial flue gas streams of fossil fuel combustion. Similarly, nanocrystalline Cu with an average size ≤8 nm was prepared by quenching ultrafine particles or clusters from a high-temperature furnace into condensed inert substrates at cryogenic temperature.[51] Using substrates that can be dissolved allows the formation of heterogeneous catalysts that have activity for the oxidation of CO to CO2 several orders of magnitude higher than that of similar catalysts prepared by conventional methods. Another synthesis technique offering great potential for the development of new active materials is high-energy mechanical alloying or highenergy milling. The main advantages of mechanical milling over most

372

Properties

other techniques are its simplicity, the ease of scale-up, and the cost, especially when compared to processes such as gas-phase condensation. A major disadvantage, however, is the specific surface of the milled powders, which normally varies between 1 and 10 m2/g, too small to be used directly for most catalytic applications. Another drawback of this process is the volume of impurities unavoidably incorporated in the powders during milling. However, in catalysis, these impurities are not necessarily detrimental to the material’s properties and may even be beneficial in some cases, although their presence often complicates the phenomenon under study and makes the experimental data more difficult to interpret. Because the specific surface is not such an important issue in electrocatalysis compared to direct catalysis, some of the first studies on the properties of milled materials were performed on specific electrocatalytic reactions. Trudeau et al.[52] and Huot et al.[53] examined the electrocatalytic activity for the hydrogen evolution reaction (HER) in an alkaline solution of nanocomposite Fe-based powders produced by the mechanical crystallization of Fe—(Co,Ni)—Si—B amorphous alloy. High mechanical deformation made these amorphous alloys crystallize, producing a two-phase material: α-Fe-rich crystallites surrounded by an amorphous Fe—B-based region. Compared to amorphous ribbons, these nanocomposite powders showed an improved activity for the HER. Also, before use, amorphous ribbons of similar composition first had to be activated by means of an oxidation/reduction treatment, which resulted in the formation of Fe crystals at the surface. No such activation was necessary for the nanocomposite powders because α-Fe nanocrystals were already present. Moreover, the Ni-containing amorphous ribbons were found not to be active even after the oxidation/reduction treatment, since, in this case, a passive Ni-oxide layer was produced in the process. On the other hand, a milled powder of similar composition revealed an activity comparable to the non-Ni-containing nanocomposites, even if its structure remained Xray amorphous. This improved activity for the ball-milled sample is still not well understood, but could be due to the presence of a large number of structural defects and a rough surface morphology, or to the presence of very small iron nanoclusters in the powder not detectable using X-rays. Benameur et al.[54] for their part, have looked at the HER activity of metastable nickel produced by milling 65 at% of Al and 35 at% of Ni for 150 hours followed by leaching in alkaline solution (20 wt% KOH). In agreement with Ivanov et al.,[55,56] they found that mechanical alloying gave rise to the formation of a CsCl-type NiAl structure. After leaching, nearly all the aluminum could be removed, leading to the formation of a

9: Gas Reactive Applications, Trudeau

373

porous bcc Raney nickel. Interestingly, even though the remaining material is structurally different from the conventional fcc Raney nickel, the first HER study revealed no major difference in activity between these two materials. Another system offering considerable interest for HER is Ni—Mo, which was first investigated more than nineteen years ago for its high activity. At that time, some Ni—Mo-coated electrodes were produced by a complex oxidation—reduction process at elevated temperatures.[57,58] Based on these studies, substantial work was devoted to investigating the structural transformation produced by mechanically alloying Ni and Mo powders, and the impact of such transformation on the electrocatalytic activity for HER.[59,60] The results showed that Ni nanocrystals, with up to 27 at% of Mo in solution, could be produced by mechanical alloying. The pressed-powder electrodes produced an HER activity superior to that of any known materials in alkaline solution at 70°C. These studies suggested that the increased Mo dissolution in the Ni lattice resulting from a reduction in the crystallite size to the nanometer range during milling was the main reason for the improved properties of this material, as presented in Fig. 9.1(a). However, further investigation revealed (Fig. 9.1(b)) that oxygen impurities, which were unintentionally introduced to the level of 2 at% in these first experiments, were also key to the unique performance of these electrodes.[61] Figure 9.1(b) indicates that, for the material to be active, at least 1 wt% of oxygen had to be incorporated into the milled powder. Surface analysis indicated that a Mo suboxide is present at the surface of the active nanostructured grains. As mentioned previously, nonstoichiometric oxides are often good catalytic materials and these Ni—Mo—O powders are probably another example of such systems. At present, the detailed microstructure is still not well understood. In the presence of oxygen, reactive milling could have incorporated oxygen into the Ni structure or caused some Mo atoms to segregate to the surface of the nanocrystals to form an oxide or suboxide shell structure affecting the surface potential, leading to the observed high activity. One result that tends to support this second hypothesis is the decrease in the amount of Mo in solution with further milling in oxygen, which is followed by the formation of MoO2. Similarly, new nonstoichiometric Ti—Fe—Ru—O nanostructured high-performance cathodes for the electrochemical synthesis of sodium chlorate were produced by milling pure metals, oxide powders, and mixtures of the two.[62,63] Partially nanocrystalline Fe78Si12B10 materials, produced by milling crystalline foils, were also found to be much more active for the catalytic hydrogenation of CO than cast or thermally crystallized materials.[64,65]

Properties

374

a

b

Figure 9.1 Hydrogen overpotential at 250 mA cm−2: (a) as a function of milling time and, in the inset, in relation with the crystallite size; (b) as a function of milling time for a Ni75Mo25 powder mixture milled under air and argon.[61]

The structure of the low-temperature annealed sample (500°C) was found to be mainly amorphous with very small α-Fe crystallites of <10 nm. The activity of this material was found to be three times higher than that of the fully crystallized sample (crystallite size ∼22 nm) and six times higher than the crystalline cast sample (crystallite size >30 nm). Trovarelli et al.[66] and Miani et al.[67] studied the catalytic properties of iron carbides produced by mechanical alloying for the hydrogenation of CO2. Using WC vial and balls, iron powder (75 at%) and graphite (25 at%) were milled together for 7.5 hours. The result was the production of a composite powder composed of Fe, orthorhombic Fe3C, hexagonal ε-Fe2C and monoclinic χ-Fe5C2 phase, with an average crystallite size of about 13 nm. The catalytic activity of this powder for the hydrogenation of CO2 was found to be comparable to that of noble metal catalysts dispersed on high-surface-area supports. As mentioned previously in this section, the use of high-energy milled powders in catalysis is restricted because of the low surface area of the powders. One way to circumvent this difficulty is to synthesize a heterogeneous nanocomposite catalyst using high-energy mechanical milling. For example, this technique was used to mill α-Al2O3 nanopowders together with nanocrystalline Ni-Ru powders, resulting in a high dispersion of metallic particles on the alumina support with a high specific surface area.[68] TPD experiments indicated that the specific surface of the metallic catalyst increased considerably upon milling with the 100 m2/g

9: Gas Reactive Applications, Trudeau

375

nanoscale α-alumina. In this case, the α-Al2O3 was also obtained by highenergy milling from the transformation of γ-Al2O3 to the α-phase.[69] Highenergy mechanochemistry was also used to prepare metallic nanocrystal (10 nm) supported catalyst directly by high-energy milling of metallic oxide MxOy (such as Fe2O3) and aluminum powder following the reaction: x 3 M x O y + 2 Al = Al 2 O 3 + 3 M y y

Eq. (9-2)

Surface activation can also be used to improve the surface area of highenergy-produced powders. For example, Sun et al.[70] looked at the catalytic activity of mechanically milled Cu30Al70 activated in NaOH nanocrystalline for the hydrogenation of xylose. After activation, the powder had a specific surface of about 24 m2/g and an activity six times higher than that of vacuum-melted samples with a high surface area. Even if high-energy milling mostly yields low-surface-area powders, high-surface-area active materials have been synthesized using highsurface-area precursors. De Leitenburg et al.[71] produced Ce—Zr—O solid solution by milling CeO2 (surface area of 55 m2/g) with ZrO2 (surface area of 15 m2/g) and obtained compounds with an average size between 9 and 13 nm and a surface area around 25 m2/g for Ce-rich compounds, which show a high oxygen mobility and a large volume of surface defects introduced by the milling, which should enhance its catalytic activity. Another synthesis technique that shows interesting potential is the sonochemical decomposition of volatile organometallic precursors in a high-boiling solvent. Found to be a very efficient technique for producing highly porous nanostructured catalysts,[72–74] this process is based on the formation, growth, and collapse of bubbles in a liquid, which can generate hot spots of ∼5000 K and pressures of ∼1800 atm with a cooling rate that exceeds 1010 K/s. Fe, Co, Fe—Co, and Mo2C metallic clusters prepared using this technique, with sizes ranging from 2 to 20 nm, were used to form colloids and nanostructured supported metal catalysts that showed better catalytic properties than normal catalysts for the dehydrogenation and hydrogenolysis of cyclohexane. The dehydrogenation catalytic activity of molybdenum carbide was found to be comparable to that of commercial ultrafine platinum powder.[72,73] Chemical precipitation, followed in some cases by calcination, has been used extensively in the past to prepare catalytic materials. New refinements in this processing have emerged to enable a better control of the structure and the average particle size of the end product. For instance, a variety of pure nanoscale metallic or even intermetallic compounds, with

376

Properties

an average grain size between 2 and 15 nm and with little or no surface oxidation, were prepared by a reduction process in alkalides or electrides.[75] Wang et al.[76] used reductive synthesis of colloidal particles in solution at room temperature and was able to prepare not only Pt particles with an average size between 5 and 15 nm but also cubic, tetrahedral, and octahedral particles with a particular atomic structure such as steps, ledges, and kinks, which could be critical for catalytic activity and selectivity. Cao and Bergens used the reduction of RuCl3 in aqueous solution to deposit a controlled amount of Ru adatoms onto the surface of Pt nanoparticles producing an optimized electrode for the electrooxidation of methanol for fuel cell applications.[77] The catalyst for the CO oxidation, Au/Fe2O3, was produced by co-precipitation from hydrogen tetrachloroaurate trihydrate, ferric nitrate, and sodium carbonate precursors dissolved in deionized water, followed by heat treatment between 473 and 773 K. The gold crystallite size was found to vary from 2–6 nm for the lower annealing temperature to 8–11 nm after calcination at 773 K.[78] In addition to the difference in crystallite size, the structures of the nano-Au crystals and the iron oxide support were also found to vary with the temperature of the heat treatment. A study of the catalytic CO oxidation in an oxygen-rich environment revealed a strong sensitivity to these morphological differences, the only stable catalyst being the one annealed at the lowest temperature. The formation of solid solution can also be used to prevent grain growth during high-temperature annealing. For instance, complex Ce—Zr—O solid solution was found to be more thermally stable than pure CeO2, with an average crystallite size of about 10 nm after annealing at 1000°C (compared to 100 nm for pure ceria).[79] Moreover, the complex oxides were found to have a higher effective surface (∼10 m2/g) and a much better catalytic activity for the removal of CO, NOx, and hydrocarbons. The sol-gel method, based on SnCl2·2H2O and H2PtCl6·6H2O, was used to produce SnO2 electrode films doped with metallic Pt nanoparticles of 5–10 nm that showed very good activity toward the oxidation of formaldehyde.[80] Gold nanoparticles of 2 nm core size encapsulated with some alkanethiolate monolayer shell were also recently synthesized chemically.[81] These nanoparticles were deposited as thin films on various supports, such as glassy carbon, electrochemically or thermally activated, and studied for methanol oxidation and oxygen reduction in fuel cell applications. The authors found that depending on the activation process, and thus a difference in the removal of the shell structure, different sites were activated. In a similar scheme, 0.5–1.0 at% Pd-coated Ni nanoparticles (50 ± 15 nm) were prepared using a polyol reduction method that produced a nanocomposite material that was found

9: Gas Reactive Applications, Trudeau

377

to better than normal commercial Pd/C catalyst for hydrogenation reaction.[82] Using these materials it was possible to convert nitrobenzene to aniline with the same efficiency but with 2–3 times less Pd. Using an electrochemical process, Reetz and Helbig[83] have been able to produce transition-metal nanoclusters with control of the particle size. In their system, an anode of the desired metal is decomposed in a tetraalkylammonium salt electrolyte, which also serves as a stabilizer for the nanoparticles. The average particle diameter was found to be a function of the current density during the electrolysis, varying from 1.4 to 4.8 nm in the case of a Pd electrode. The concentration of metal was about 71 at% for this electrode, the remainder being elements such as C, H, N, and Br (in the case of a bromide salt). Using a slurry of carbon black, in addition to the salt, allowed them to produce dispersed nanoparticles on a carbon support (in this case 5.5% Pd/C). For their part, Kost et al.[84] used electrodeposition of metal microparticles of Pd, Ir, Ru, and Pt in poly(4-vinylpyridine) to prepare films for electrocatalysis. The crystallite size varied between 10 and 100 nm and showed good properties for hydrogen evolution in 0.5M sulfuric acid. Thermal decomposition of an organometallic complex has been used successfully to produce nanoscale metallic particles embedded in a support matrix. One such system is the nano-Co/C produced by the pyrolysis of carbon supported cobalt phthalocyanine at temperatures between 600 and 1000°C. This material is intended to replace platinum as an electrocatalyst in solid polymer electrolyte fuel cells. Thermal treatment of the organometallic—carbon complex at a temperature around 900°C produced a material with a structure illustrated in Fig. 9.2. The TEM micrograph reveals a bcc-Co nanocrystal embedded in an amorphous carbon matrix. A layered structure typical of graphite can be observed around the crystallite, which seems to be the basis for the long-term stability of the electrode.[29,85,86] Similarly, very active monodisperse 7 nm metallic Ni nanoparticles were produced by the thermal decomposition of Nioleylamine complexes.[87] Microwave radiation was also used to thermally decompose (η-C2H4)(Cl)Pt(μ-Cl)2Ru(Cl)(η3 :η3-2,7-dimethyloctadienediyl)/ Vulcan carbon in order to produce PtRu/carbon nanocomposite as an anode catalyst for methanol fuel cell. The average nanoclusters size, for a 50 wt% total metal loading, was found to be 5.4 ± 3,2 nm and the catalyst had a performance equivalent to that of a 60 wt% commercial catalyst and equivalent to an unsupported PtRu catalyst.[88] Miquel and Katz[89] used a flame burner reactor and subsequent thermal annealing to produce VPO powders, the most widely used catalysts for the selective oxidation of butene and n-butane to maleic anhydride. The

378

Properties

design of the catalyst structure in this case is crucial since this catalytic oxidation has been shown to be structure-sensitive (i.e., the selectivity of a given product is related to the crystallographic plane exposed at the surface of the catalyst). Liquid VOCl3 and PCl3 were used as precursors, and the morphology and the crystalline structure of the final powder could be changed through the variation of flame temperature. At high temperatures (maximum flame temperature of 2800 K), the powder collected is transformed to γVOPO4 upon subsequent reheating, Figure 9.2 High-resolution TEM while at low temperatures micrograph of a cobalt nanocrystal embedded in an amorphous carbon (maximum flame temperature of layer. Note the graphitic layers that 2300 K), the final crystalline phases surround the particle.[3] produced are β-VOPO4 and V(PO3)3. Nanostructured catalysts have also been prepared by laser vaporization. Samy El-Shall has synthesized a number of metal oxide nanoparticles or nanocomposites using laser vaporization of metal or mixed-metal targets followed by gas-phase chemical reactions and by controlled condensation from the vapor phase. Nanoparticles with an average size between 10 and 20 nm can be obtained by this process as well as nanostructured (CexZr1−x)O2−y catalysts for automobile exhaust treatment. In another study, Rousset et al.[90] investigated the formation of PdPt bimetallic nanocrystals with an average size of about 2.5 nm and found that Pd segregation was present at the surface of these crystals. Laser vaporization of equally mixed Pd, Ag, and Au colloid was also used to produce 4.4 nm homogeneous nanoparticles that showed better properties than traditional Pd complex for the Heck reaction for the synthesis of nabumetone (a nonsteroidal anti-inflammatory drug).[91] The thermal crystallization of an amorphous precursor was also used to produce nanostructured reactive materials. Yamashita et al.[92] studied the hydrogenation of buta-1,3-diene over Fe90Zr10 alloys after different annealing temperatures. They found that the maximum activity was attained when the alloy was in a precrystallized state, with numerous fine

9: Gas Reactive Applications, Trudeau

379

particles of α-Fe (5–40 Å) present at the surface, and a distorted crystal lattice from the dissolution of Zr. Peuckert and Baiker[93] have examined the hydrogenation of CO over Fe81B13.5Si3.5C2 ribbons with different average α-Fe particle sizes. They found that the surface chemistry of their materials was modified upon annealing, as the average iron crystallite size became larger. These results seemed to indicate that maximum activity was obtained when precrystallite nuclei were present at the surface of the materials, and that the activity would decrease rapidly with grain growth. Brunelli et al.[94] investigated the electrocatalytic activity of HF-treated amorphous Au30Zr70 amorphous alloys and found that the porous structure in nanocrystalline gold particles had better activity for HER than untreated ribbons or polycrystalline gold electrodes. In recent years a new variety of nanocomposite porous materials have demonstrated high promises for various catalytic applications, as direct catalysts or for the separation of gas mixtures.[95] Antonelli and Ying[96] have shown that it is possible using liquid crystal phase, which acts as a template around which inorganic materials can nucleate and crystallize, to produce transition-metal oxide-based porous materials, of Nb, Ta, and Ti, with pore size ranging from 2.0 to 5.0 nm, an amorphous wall structure, and specific surface areas up to 1000 m2/g. The possibility to produce host materials with finely tuned pore sizes, high surface areas, enhanced accessibility of the active surface sites, and walls that have d-state available is an important feature for some catalytic applications.[97,98] Moreover, it is possible, using different chemical species, to modify the oxidation state of the wall structure, tailoring it to some specific catalytic process. One example of such specific manipulation on the nanometer scale can be seen in Fig. 9.3.[99] In this example, mesoporous Nb oxide (Nb2O5) was obtained with a specific surface area of 947 m2/g and a pore size of 2.31 nm. This material was treated with excess bis(toluene)niobium in toluene over several days under argon, producing a new porous material with a reduced specific surface of 702 m2/g and pore size of 1.99 nm. Since the bis(toluene)-niobium is unstable, it is suggested that the Nb mesostructure will induce decomposition of the organometallic compound and the Nb will deposit itself on the inner and outer surfaces of the material, giving rise to a mesoporous composite with a very high electrical conductivity. Moreover, the amount of Nb atoms deposited inside the pore structure could be further increased by further treatment at 80°C, giving rise to a material with smaller pore size and higher conductivity. More interestingly, this new conductivity-tailored nanocomposite was found to be able to cleave dinitrogen at room temperature, a reaction that normally requires very forcing conditions or low-valent

Properties

380

a 550

Intensity [a.u.]

500

450

400

350 300 410

b

405 400 395 390 Binding Energy (cV)

385

Figure 9.3 (a) Schematic representation of deposition of Nb atoms from bis(toluene)-niobium to the inner surface of a mesoporous oxide to form a lowvalent metallic oxide coating on the internal surface. (b) N 1s region of an XPS spectrum of a reduced mesoporous niobium oxide sample after treatment for 24 h at RT with dinitrogen, revealing the presence of surface N3− species coming from the cleavage of N2 molecules.99

coordinatively stressed metal centers. In another example, Fe3+-doped mesoporous tantalum oxide was used as Schrauzer-type photocatalyst for the conversion of dinitrogen to ammonia.[100] Recently, Haoshen and Honna obtained a self-ordered mesoporous composite with a nanocrystalline-amorphous hybrid wall structure that

9: Gas Reactive Applications, Trudeau

381

could have improved properties compared to an amorphous wall structure, since some materials properties are related to crystallographic constitution or a periodicity over a particular length scale.[101] In their work they succeeded in producing hexagonally ordered mesoporous materials with pore size and specific surface around 4.0 nm and 300 m2/g and having a variety of wall structure and composition, such as TiO2 (anatase) nanocrystals surrounded by a TiO2—P2O5 amorphous phase. To better control the amorphous phase and the subsequent crystallization process, they also added small amount of a third metal oxide such as Li2O, CuO, or Fe2O3. The presence of this third phase could also have some impact on some of the material properties. For instance, the addition of 5– 10 mol% of MxOy with ionic conductivity or semiconductivity improved the electrochemical properties of binary TiO2—P2O5 system. Zeolites and mesoporous materials are also used to incorporate metal clusters, such as PtRe, PtIr, or PtCo, in such a way as to prevent facile migration and cluster sintering, producing in fact what is now called “nanometer-sized microreactor” or “ship-in-bottle” catalysts. They can also be used as templates to produced different nanostructures, as described in Fig. 9.4 for different Pt structures created using mesoporous FSM-16.102 A sonochemical process was recently developed for the insertion of tin nanoparticles into mesoporous carbon, produced by templating from mesoporous silica, which was used with high efficiency as anode in a rechargeable lithium battery.[103] In a new direction, nanocomposites based on nanotube materials are also being designed and investigated for their catalytic properties. For instance, 1–2 nm Pt particles supported on single-walled carbon nanotubes have been synthesized and studied for the partial hydrogenation of 3-methyl-2-butane to 3-methyl-2-buthenol,[104] and single-wall nanohorns[105] and multiwalled nanotubes[106] for proton-exchange membrane fuel cell applications. Carmo et al.[107] compared the properties of single-walled nanotubes, multiwalled nanotubes, and carbon powder (Vulcan XC-72) as support for noble metal nanocatalysts also for protonexchange fuel cells. They found that the activity was greatly influenced by the nature of the support and that nanoparticles of PtRu (4–6 nm) supported on multiwalled nanotubes had better performance than more conventional carbon powder. Zhao et al.[108] found improved properties for Pt—Sn of average diameter 2.7 nm deposited on multiwalled carbon nanotubes, as compared to XC-72 carbon, for the electro-oxidation of ethanol. They suggested that this improved activity could be due to the unique structure, higher electric properties, and lower organic impurities of the multiwalled nanotubes, or to a direct enhancement of the activity by this special catalytic support.

382

Properties

Figure 9.4 Representation of the template fabrication of platinum nanoparticles prepared by the controlled removal of CO from Chini cluster anions [Pt3(CO)6]n2− (n = 5, 6) in FSM-16 and of platinum nanowires by exposure of H2PtCl6/FSM-16 in propan-2-ol and water to X-rays or UV illumination.102

It should be mentioned that surface lithography now makes it possible to prepare nanometer-scale model catalysts. For example, Jacobs et al.[109] reported the catalytic activity of a platinum array formed by 50-nmdiameter disks about 200 nm apart that were prepared on an oxidized silicon wafer as shown in Fig. 9.5. Ethylene hydrogenation at high temperature revealed that the properties of this catalyst array match results on platinum foils and dispersed or supported porous platinum catalysts. Such high-precision nanodesign of catalysts offers a unique possibility to understand the relation between nanostructures and the reaction activities, especially when coupled with analytical tools such as scanning tunneling microscopy (STM), which can monitor surface changes on the atomic

9: Gas Reactive Applications, Trudeau

383

Figure 9.5 SEM micrographs of the platinum cluster array prepared by electron beam lithography.110

scale during catalytic reactions.[110] Furthermore, this understanding can only improve as smaller structures are developed. In a similar concept, Berko et al. developed a two-dimensional nanocrystalline model catalysts by vapor deposition of a few percent of a monolayer of metallic Rh onto polished TiO2(110) followed by annealing at 1100 K in vacuum to produce seeded regions. By subsequent exposure to further amount of Rh at 1100 K, it was possible to produce nanoparticles of controlled sizes with a constant distance between them.[111]

9.3 Gas Sensors Another field where nanostructured materials already play a prominent role is gas detection. New technological developments, terrorism

Properties

384

scares,[112] and environmental concerns are creating the need to rapidly detect various kinds of toxic gases, such as CO2, CH4, C2H4O, fluorocarbons, N2O, O3 (greenhouse effects), fluorocarbon, halocarbon (ozone layer destruction), SOx, and HCl (acid rain),[113–115] with great sensitivity and selectivity. The level of detection for such pollutants would be in the ppm range and even down to the ppb level for some particular cases. Moreover, to optimize combustion and to reduce emissions it is necessary to monitor the exhaust gases from boilers, combustion furnaces, and automobile engines using feedback controllers, which require a rapid response time.[116] A number of different physical principles are presently used for detection purposes:[117–119] • • • • • • •

Changes in mass Changes in transport properties Heat of reaction measurements Work function measurements Capacitance measurements Electrochemical detection Optical absorption and reflection

Typical examples of the sensitivity of gas sensors are approximately 1 ppm for CO and 0.1 ppm for NO2.[120] Various chapters of this volume show that a number of material properties are modified as the crystallite size is reduced to the nanosize, and it can easily be concluded that the sensing properties of gas-detector materials will also be affected by a reduction in the crystallite size down to a few nanometers. Although sensing properties could be based on other properties of materials that are modified at the nanoscale, such as optical transmittance[121] or capacitance variation,[122] because of the cost of the materials and the simplicity of the measurement principles, methods based on changes in the material transport properties are chosen for many industrial applications, even if they are not the most sensitive, the most accurate, or the most selective. For this reason, and since this field is one where nanostructured materials are already playing a significant role, the present section will be devoted to this type of sensing technology and related materials. Again it should be mentioned that, in this technological area, the development of nanostructured materials should not be seen as a recent phenomenon. As early as 1981, in fact, Ogawa et al.[123–125] were using gasphase evaporation of tin in an oxygen atmosphere to produce nanoscale SnO2 crystals with an average size of 6 nm. The novelty in the different researches that are presently underway relates more to our understanding

9: Gas Reactive Applications, Trudeau

385

of the importance of the structure on the nanoscale, better control of grain growth,[126] and the application of new processing methods that offer numerous possibilities for easier design of materials on this scale. It should also be stressed that nanostructure-based sensors are being developed for other applications such as single-molecule detection systems.[127]

9.3.1 Physical Principles of Semiconductor Sensors The possibility of using the interaction of semiconductors with reactive gases was applied more than 35 years ago.[128] Albeit not giving the ultimate in accuracy or stability, one of the most common gas sensors is tin oxide, due mainly to its cost and sensitivity.[129–131] SnO2 is an n-type semiconductor with a direct band gap of 3.6 eV as shown in Fig. 9.6, with the donor levels formed by oxygen vacancies in the lattice. The model most frequently mentioned in the literature is based on the fact that oxygen atoms absorbed or adsorbed at the surface form negative ions, resulting in a decrease in the surface conductivity. Three different types of ionic species have been observed on the surface in addition to adsorbed O2 molecules: O 2−, O−, and O2−, which are desorbed at characteristic temperatures.[132] The presence of a reactive gas (oxidant or reducer) in the atmosphere, its adsorption at the surface, and its reaction with the oxygen atoms will result in a decrease or an increase in surface electrons and, hence, in the conductivity. Typically, the value measured is the resistance of the semiconductor, Rs, Atmosphere Conduction Band EF ED

-

O O- ++++++ O- ++++ OValence Band

L

Distance

Figure 9.6 Schematic representation of the band structure in the presence of absorbed oxygen. EF, Fermi level; ED, donor level.

Properties

386 Rs = KC s−α

Eq. (9-3)

where Cs is the concentration of gas in air, K is the sensitivity coefficient, and α is the dimensional power exponent between 0 and 1.[129,133] In most cases, however, the value given is the sensitivity value, SG, or its logarithm form: SG =

Ra Rs

R R log Sg = log⎛ a ⎞ − log⎛ s ⎞ ⎝ R0 ⎠ ⎝ R0 ⎠

Eq. (9-4)

Eq. (9-5)

where Ra is the resistance of the sensor in clean air and R0 is the value of the resistance at a specific gas concentration, a specific operating temperature, or the relative-humidity value.[129] However, because this value cannot fall below 1, some researchers express the sensitivity as a variation of the conductance, G: SG′ =

Gs − Ga Ga

Eq. (9-6)

With hydrogen, three different reactions have been identified, depending on the surface species considered.[134,135] At low temperatures (below 150°C), the negative-charged adsorption of hydroxyl species takes place: H2 + O2 + 2e− → 2OH− (below 100°C)

Eq. (9-7)

H2 + O2− + e− → 2OH− (below 150°C)

Eq. (9-8)

both of which result in a decrease in conductivity. At temperatures above 200°C, the desorption of O− will release one electron in the conduction band: H2 + O− → H2O + e−

Eq. (9-9)

and, finally, positive-charged adsorption of hydrogen to the surface oxygen occurs at temperatures above 200°C: H2 + 2O− → 2OH− + 2e−

Eq. (9-10)

9: Gas Reactive Applications, Trudeau

387

A different point of view was presented recently by Kanamori et al.,[136] who, from ESR (electron spin resonance) on thin films composed of 10 nm SnO2, concluded that the source of carrier electrons was related more to the amount of unpaired electrons localized in oxygen vacancies. However, taking into account the amount of adsorbed oxygen, it can be calculated that the observed conductivity variations cannot be explained solely by the above reaction. For SnO2 the amount of adsorbed species was estimated to be between 1012 and 1014/cm2, which is negligible compared to the carrier concentration of 1018 to 1020/cm3.[132,137] A good demonstration of the variation in carrier concentration was made by Ogawa et al.,[123] who measured it using the Hall effect in ultrafine SnO2 particle films for the detection of C2H5OH. To explain the high sensitivity of some particular semiconductors for gas detection based on conductivity changes, it is necessary to consider the structure of the semiconductor itself. In fact, early researchers found that the grain size is critical for the response to gas contaminants.[137] Studies of monocrystalline thin films have shown only small electrical resistance variations in the presence of an oxidant or reducing gas.[138] Polycrystalline materials, for their part, show a strong response. This effect of polycrystallinity was related to the formation of space charge layers with their potential barriers between grains, the grain boundaries behaving as double Schottky junctions.[139] The height of these barriers was found to depend on the equilibrium concentration of adsorbed oxygen. Charge transport across grain boundaries is thus a major process governing the material response.[138] Figure 9.7 shows two different examples of polycrystalline sensing materials: one where the particle size (diameter for spherical grains) is larger than twice the value of the depletion region L, and the other where D is smaller than 2L, as well as a schematic description of the gas sensing process for H2 detection using a typical variation in the material conductivity with the concentration of reducing gas. The value of L corresponds to the Debye length of the materials and the strength of the oxygen chemisorption following the relation[123,140] 2 εE L =⎛ 2 s⎞ ⎝ e0 N ⎠

12

Eq. (9-11)

where ε is the static dielectric constant and N is the concentration of completely ionized donors or acceptors. It can be seen that if D ≤ 2L, then all the crystals will be depleted, causing the gas sensitivity of the element to

Properties

388 Adsorbed

SnO2

SnO2

Crystallite size > LD

Crystallite size < LD

Es Ec Eπ H2

H2 SnO2

H2 H2

H2O H2

SnO2 H2

Ec

H2

H2O

Es

Eπ H2

σ

100 ppm

100 ppm

200 ppm

200 ppm 300 ppm

Time (min)

Figure 9.7 Schematic diagram of the gas sensing process for SnO2 for two different materials with an average crystallite size larger or smaller than the Debye length (LD) and with or without the presence of a reducing gas such as H2. The lower curve represents a typical conductivity variation as a function of the concentration of reducing gas in the surrounding sensor atmosphere.

change with D in the presence of a reducing gas.[141] For SnO2 with the value of ε = 1.2 × 10−10 F/m, Es = kT, and with a carrier concentration of about 3.6 × 1018 cm−3, L at 250°C can be estimated to be around 3 nm.[81] Nanostructured materials with their very small size and their large number of grain boundaries are, thus, fundamental to this field. If the crystallite diameter is more than twice the depth of the space charge layer, then the material resistance is directly related to the potential barrier, Es. On the

9: Gas Reactive Applications, Trudeau

389 Nanostructured material

Surface Bulk

Large Crystal Grain Boundary Controlled Material

Figure 9.8 Equivalent circuits in relation to the size and structure of the sensing materials.131

other hand, if the crystallite diameter is less than twice the depth of the space charge, each crystallite contributes to what is effectively the channel of a field-effect transistor, wherein the negative surface charge (due to the adsorbed oxygen ions) acts like a gate voltage.[2,129] The existence of this critical size and also morphological considerations could be responsible for the different models proposed by different researchers. For instance, the impact of grain boundaries will be less if a small continuous channel exists between large grains (D ≥ 2L), the depletion region mainly affecting the neck between the grains.[142] The impact of size can also be schematized by an equivalent circuit as presented in Fig. 9.8 for different materials: large grains (monocrystals), grain boundary controlled materials, and nanostructured compounds. The choice of crystallite size is, thus, critical for the properties of a particular compound, especially its sensitivity to a particular gas.129 It was also shown that a relation exists between the average concentration of the conduction electron and the reciprocal of the semiconductor temperature:[143] ln nc ≈ B +

Ea kT

Eq. (9-12)

where Ea is the activation energy equal to (Es − Vs), Es being the surface energy level and Vs the maximum Schottky-barrier height due to chemisorption. Assuming that the electron mobility is independent of the temperature, then lnσ ≈ c +

Ea kT

Eq. (9-13)

Other factors found to have an impact on the sensing properties include the crystalline orientation[144] and the presence of surface defects or

Properties

390

impurities.[139] Indeed, semiconducting metal oxides were demonstrated to show surface sensitivity to redox reactions: for example, during hydrocarbon oxidation reactions on MoO3, hydrogen abstraction occurs at (001) and (100) crystal phases, while oxygen addition occurs at the (010) basal plane.[144] The possibility of adding dopants[145,146] to the semiconductor to improve its detection properties was also found as early as the 1960s for the case of Pd- and Pt-[147] doped tungsten oxides (WO3). Catalytic dopants have the effect of modifying the selectivity mainly by changing the rate of the redox reactions. Chang[148] found that by depositing a 3.5 nm Pd–Au film on a SnO2 substrate, the sensing properties for H2 and propylene (C3H6) were greatly enhanced compared to CO or methane. Cricenti et al.[149] found that Pt doping decreases the sensitivity threshold for CO detection by a factor of almost 2, while at the same time reducing the peak sensitivity temperature by about 100°C. Recently, Vlachos et al.[150] proposed that the enhanced sensitivity due to metallic additives was related to an electronic interaction between the metal and the semiconductor. In their model, the presence of metallic additives has a direct impact on the depletion depth in the semiconductor, the depth increasing with the metal work function according to W=

2ε s ⎛ δ ϕ m − X − ( EC − EF ) − q 2 Ds y0 ⎞ 2 ⎠ εi q n0 ⎝

Eq. (9-14)

where ϕm is the metal work function, X is the electron affinity of the semiconductor, δ is the thickness of the interfacial layer between the metal and the semiconductor, Ds is the density of surface states per unit area, and eV, y0 is the level below which the surface states are occupied for charge neutrality, and εi is the dielectric constant of the interfacial layer. The net effect of adding metallic nanoclusters would be to decrease the electron availability of semiconducting grains, which results in an active size of the semiconductor particles smaller than the geometrical size in relation to the metal work function. A similar result found by Zhang et al.[151] was that the addition of dispersed nanoclusters (<10 nm) of Pt on α- and βCdSnO3 improved the sensitivity for ammonia detection at temperatures below 240°C. Cao et al.[152] found, by XPS analysis, that the addition of Pd to SnO2, and the subsequent formation of PdO and PdO2 during oxidation treatments of thin films produced by sol-gel, decreased the Fermi level of SnO2 by 0.2 eV. Recently, Matsushima et al.[153] correlated the sensing enhancement for C2H5OH detection of La2O3-doped SnO2 with the basic nature of the additives, which favors the reaction selectivity in

9: Gas Reactive Applications, Trudeau

391

the oxidation of C2H5OH. Sysoev et al. found, for their part, that Cu-doped nanocrystalline SnO2, with columnar structure with an average diameter of 50 nm, shows very good selectivity for detection of ammonia compared to acetone, ethanol, or propanol.[154] Rastomjee et al.[155] used ion implantation to modify the surface structure on the nanoscale and the sensing properties of SnO2. They found that selectivity toward CO was enhanced by Bi implantation. Barbi et al.[156] used Pt-implanted nanostructured SnO2 for CO detection and found a reduced sensitivity, as well as a decrease in the response rate, which could be caused in part by strong alterations of the lattice structure due to bombardment with heavy platinum shells. Finally, Stankva et al.[157] studied seven types of dopants in WO3 and found good selectivity for different gases. For instance, they found that WO3 doped with Au, Ag, and Pt can selectively detect H2S, NO2, and NH3, respectively. As stressed previously, one disadvantage of most semiconductor-based sensors, and in particular SnO2, is that they are simultaneously sensitive to most reducing gases and often two or more gases will be present at the same time. For example, the detection of a major pollutant such as H2S around gas extraction and processing plants will be complicated by the presence of other accompanying gases, such as H2 and CnHm.[158] Various approaches can be chosen to improve selectivity. First, by modifying the operating temperature it is possible to promote one oxidation reaction more than another. For example, CO will oxidize more rapidly than CH4 at high temperature, resulting in a smaller length of interaction for the CO and thus a smaller resistive change and, hence, a lower sensitivity. At lower temperatures, the contrary phenomenon takes place: CO oxidizing at a lower rate will interact more with the material.[129] Modifications in the response of solid-state sensors have also been obtained through treatment involving special gases. Zubknas et al.[159] changed the sensing response for NOx (from an increase to a decrease in conductivity) of a field-effect transistor covered with a 10-nm Pt film by prior exposure to ammonia at 150°C. Yamazoe et al.[132] in their study of the reactivity of SnO2 with O2, H2, and H2O, found the formation of a number of oxygen species on SnO2 as a function of the temperature and revealed that the process of conductivity change as a function of gas present and the detector temperature is quite complex. For example, they found that H2 will react first with the adsorbed oxygen with an irreversible or reductive adsorption of the H2 to form surface hydroxyls, which would lead to the adsorption or desorption of the H2O present on the surface. At higher temperatures

392

Properties

(>500°C) the adsorption of H2O and H2 is no longer significant and the change is mainly from the consumption of adsorbed (and lattice) oxygen with H2.[132] Finally, mention should be made of one of the important problems of solid-state gas sensors and, in particular, tin oxide, namely the adsorption of water vapor, which results in a large increase in the conductivity[132] due to the presence of hydroxyl groups at the surface that contribute positively to the conductivity. This effect is one of the reasons for the instability of this material and variation in its relative humidity as well as its long-term conductivity.

9.3.2 Nanostructured Design of Sensing Materials A number of techniques exist for preparing gas-sensing materials of various nanostructures, and most of the different synthesis methods described in this volume can be used to prepare useful materials.[160] An important fact is that compositionally identical materials have a different response to various gases depending on their method of synthesis,[161] their thickness, and the nature of the substrate. For example, Ansari et al.,[162] produced a number of SnO2 films using chemical vapor deposition, spray pyrolysis, and physical vapor deposition, and found very different responses to H2 for these three materials. Apart from gas-phase evaporation,[123,125] two of the main techniques used to prepare nanostructured films are sputtering and, more often, reactive sputtering. Rickerby et al.[120] studied the structure of SnO2 nanocrystals produced by radiofrequency (RF) reactive sputtering. The morphology and nanostructure of their films are presented in Fig. 9.9. The SEM micrograph of Fig. 9.9 (left) clearly shows the granular nature of the films, while the bright- and dark-field TEM micrographs (a) and (b) present grains with values ranging from 2 to 20 nm. For their materials, they found that porous films have a higher sensitivity up to a thickness of about 200 nm (at constant grain size) which increase with diminishing grain size, while thick and compact columnar films generally decrease in sensitivity, with increasing film thickness in the range 0.1–1.0 μm. This improved sensitivity of porous films was attributed to the columnar structure, which allows gas molecules to permeate along vertical fissures, which in thin films will also increase the surface area exposed to the gas with increasing thickness. In a similar work, Vlachos et al.[163] produced different SnO2 films with average crystallite sizes of about 11 nm and a high surface area (50 m2/g) after deposition at around 460°C. They show

9: Gas Reactive Applications, Trudeau

393

Figure 9.9 (Left) SEM micrograph and (a) bright-field and (b) dark-field highresolution TEM micrographs of nanostructured SnO2 gas sensing materials prepared by reactive sputtering.120

that one advantage of reactive sputtering revolves around the potential control of the O : Sn ratio. By controlling the flow of oxygen during deposition, it is possible to vary the nature of the films from metallic to semiconductor (nearly stoichiometric SnO2). Such control allows the synthesis of materials with a large amount of oxygen vacancies that show a higher sensitivity than compounds with a high O : Sn ratio, because the chemisorption of oxygen atoms is directly related to the free electron concentration (or the oxygen vacancy concentration). Yoo et al. used also RF sputtering with a 0.5 wt% Pd-catalyzed ITO target (In2O3 : SnO2) to deposit amorphous films that were subsequently annealed at 650°C to form thin film having grains averaging 35 nm in size that showed very good sensitivity and selectivity to H2 at 300°C.[164] In an interesting study, Serrini et al.[165] studied the effect of the oxygen concentration during RF sputtering, grain sizes, and the amount of adsorbed oxygen on the sensing properties for CO and NO2 of SnO2 grains. Figure 9.10, which presents the interrelation between some of these properties, reveals that the sensitivity increases for both gases with decreasing grain sizes and that, as mentioned earlier, the critical size for SnO2 indeed seems to be around 6 nm if we consider a depletion region of about 3 nm. More interestingly, it can also be seen that the grain size is not the only factor influencing the sensitivity since the response is greatly affected by the amount of adsorbed oxygen. Two samples (4 and

394

Properties

Figure 9.10 Response to NO2 (150–200°C) and CO (210 ppm at about 300–350 °C) in air as a function of the adsorbed oxygen in the first two layers of SnO2 films and of average grain size.[165]

6), which have an average grain size below the threshold for graincontrolled conductivity (as compared to grain boundaries for larger grains), showed a marked difference for NO2 detection, with the sensitivity nearly doubling when the amount of adsorbed O in the first layers of the materials increased from 17% to 25%. Mixed-oxide nanocomposite materials, which showed good properties for the detection of NOx, ethanol, and CO, produced by RF sputtering based on Ti, W, and Mo, have also been studied for their sensing properties. Enhancement in performance, especially the inhibition of coalescence during the thermal stabilization stage of the sensing materials, was attributed to the presence of a suitable mixture of a dopant or a foreign phase to a particular oxide.[126] Hu et al.[166] used laser ablation on SnO2 and Sn targets to produce nanostructured materials. The lowest average grain sizes were obtained from amorphous films deposited at 100°C followed by crystallization and oxidation, in an oxygen atmosphere at 400°C for 4 hours. The resulting materials had grain sizes between 4.0 and 5.2 nm and showed an improvement in sensitivity by a factor of approximately 2–5 for C2H5OH detection when compared to films having larger grain sizes (30–46 nm).

9: Gas Reactive Applications, Trudeau

395

Dieguez et al.[141] found that the nature of the substrate and the nature of the oxidizing atmosphere had an influence on the structure of the materials. They revealed, for nanocrystalline SnO2 prepared by PVD, that the crystallite size was different if the annealing was done in synthetic air or in oxygen. They also observed a difference in the crystallite structure between materials deposited on sapphire as compared to SnO2 deposited on SiO2. In the latter case, diffusion of Sn into the silicon substrate occurred. Also, they observed that nanocrystalline powders produced by precipitation had a large number of defects after annealing below 450°C. Other work has also shown that the active area can be considerably increased by depositing the film on a rough surface, the surface texture being replicated by the film.[120] High-energy mechanical alloying was used also to prepare different nanostructured sensor materials. Jiang et al.[167] prepared (α-Fe2O3)x— (SnO2)1−x powders by milling hematite (α-Fe2O3) and cassiterite (τ-SnO2) in air using tungsten carbide balls and vials. They found that their materials, especially one with 85 mol% of Fe2O3, had a much higher sensitivity to alcohol compared to the unmixed powders at temperatures around 250°C. At the same time, the sensitivity to other gases such as CO and CH4 was minimal and did not change with milling. For this system, highenergy milling formed an α-Fe2O3—SnO2 solid solution with a hematitelike structure with an average crystallite size that decreased to a stable value of the order of 8 nm after about 25 hours. An amorphous phase and tungsten carbide contamination were also present after a long milling time. Even if the grain size of the hematite phase was stable after 25 hours of milling, the sensitivity to alcohol continued to increase with milling time. This improvement could be related to the fact that, even if the crystallite size remained stable, the lattice parameter of the Fe2O3-type crystals showed a continuous increase with milling due to the continuous dissolution of SnO2. This improvement could also be due to the presence of the amorphous phase or of WC—Co impurities, the possible presence of WO3, or catalytic effects related to Co. Hu et al. studied the oxygensensing properties of SrTiO3 prepared by high-energy milling followed by thermal annealing and found an optimum annealing temperature that allowed their sensor to be operated at 40°C with the same sensitivity as commercial sensors that have to be operated in the range 300–500°C.[168] However, in this case also, they did not discuss the effect on the sensing properties of potential contaminants since they milled their materials for 120 hours. TiO2—SnO2 nanofibers or structures with nanogrooves have been prepared by milling a mixture of SnO2 and TiO2 in isopropanol followed by compaction, sintering at 1200 or 1450°C for 6 hours, and

396

Properties

exposure to a mixture of H2 and N2 at 700°C for 8 hours.[169] The materials with nanocarving revealed a higher sensitivity for H2 detection that could be correlated to the increased surface area. Finally, reactive milling of SnCl2, Na2CO3, and NaCl was used to synthesize SnO nanoparticles, which were oxidized to SnO2 particles of size ranging from 5 to 30 nm after heat treatment in air.[160] Liu et al.[167] used plasma-enhanced chemical vapor deposition followed by thermal annealing to produce a mixture of α-, β-, and γ-Fe2O3 powders with an average size of about 6 nm (which is about 2L) after annealing below 600°C, which has good alcohol-sensing capability at around 400°C while at the same time being less sensitive to methane. Bhowmik et al.[170] studied the electrical characteristics of nanocrystalline TiO2 prepared by inert-gas-phase condensation and found that the conductivity is dependent on the porosity, grain size, and grain boundary structure. They also observed that materials annealed at low temperature (400°C) had an enhanced sensitivity to changes in the ambient atmosphere, especially the humidity level, which could make them useful for oxygen detection. Jun et al. found that the thermal oxidation of Ti in air at 900°C produced a specific microstructure of both short and continuous cracks exhibiting a very high sensitivity of 1.2 × 106 to 1.0% H2 at 300°C with a very short response time of about 10 seconds.[171] Sol-gel was also used to prepare nanostructured or nanocomposite gas detection materials. Rella et al.,[172] for instance, produced both pure and Pd-doped SnO2 films with a mean grain diameter that, after annealing at 600°C, was smaller for Pd films compared to the undoped sample (10.6 nm as opposed to 7 nm) and also a higher specific area for the doped material (58 m2/g compared to 35 m2/g). These results indicate that the presence of Pd and its subsequent oxidation to PdO can serve as a grain-growth inhibitor by pinning the grain boundaries in some way. As for gas detection, they found that their Pd-doped films show a higher sensitivity to CO with a lower operating temperature. Xu et al.[173,174] produced a number of composite materials by impregnating hydrous SnO2 with different aqueous solutions in which a salt and/or polyoxy acid of different metal or nonmetal elements was dissolved in order to reduce grain growth during high-temperature sintering. They found that large-surface-area films (around 40 m2/g) could be obtained when tungsten, niobium, or phosphorus were used as dopants even after sintering at 900°C, which is about eight times larger than for undoped SnO2. Figure 9.11 presents variations in sensitivity for H2 and CO as a function of the crystallite size for a number of SnO2 films doped with 5 at% additives after thermal treatments at 300 and 400°C, showing, at the same

9: Gas Reactive Applications, Trudeau

a

397

b

Figure 9.11 Variation with crystallite size of the gas sensitivity toward (a) H2 and (b) CO for SnO2 doped with 5 at.% additives.174

time, the impact of various additives on the average crystallite size and the large increase in sensitivity with decreasing grain size, especially below 10 nm. Sun et al.[133] studied the sensing properties of solgel-derived thin films produced by the dissolution of iron ethoxide Fe(OC2H5)3 in benzene. After calcination at about 400°C the film consisted of α-Fe2O3 grains that varied in size from 20 to100 nm and showed good sensing properties for NO2 and CO below 300°C that depended on the nature and structure of the electrode materials. Addition of Cu to 5–7-nm SnO2 grains, prepared by aerosol pyrolysis and the subsequent formation of CuO during annealing treatments, was found to greatly increase the material resistance.[158,175] The conductance changed from a value in the range (0.2–5.0) × 10−4 Ω−1 at 100°C to a value between 10−8 and 10−9 Ω−1 for samples containing 1.2–1.5 at% Cu. This increased resistivity was found to have a major impact on the sensitivity to hydrogen sulfide, which was much greater than that of the pure SnO2 samples. At the same time, as seen in Fig. 9.12, compared to the sensing properties of pure SnO2 (labeled “1”) and of nanocomposites SnO2(CuO) (labeled “2”) films at 150°C, the sensitivity to other gases such as CO or ethyl alcohol was significantly reduced. It is speculated that this improvement in sensitivity was due to the reaction between CuO and H2S, which transforms the highly resistive p-CuO segregated at the grain boundaries to highly conductive CuS, a hypothesis that seems to be corroborated by

Properties

398

Sensitivity

10000

2 1000 100

10

1

1

H2S 100 ppm

C2H5OH 80 ppm

CO 300 ppm

Figure 9.12 Comparative sensitivity of pure SnO2 (1) and nanocomposite SnO2(CuO) (2) films to different gases at 150°C.[158]

surface analysis that reveals the presence of sulfur only on the surface of the SnO2(CuO) nanocomposites after exposure to H2S. A similar sulfur reaction was also found for the case of Ag-doped SnO2 films.[176] Neubecker et al.[177] used ozone-enhanced evaporation (OEE) to prepare oxygen-deficient NiO sensor materials through molecular beam deposition with an average grain size between 5 and 10 nm and high oxygen deficiency. They found that the sample with the highest oxygen deficiency had better sensing properties toward a number of toxic gases but also observed a tendency to further oxidation at elevated temperatures, which could result in a strong drift of the sensor. Modern nanoelectronic integration and patterning combined with data extraction techniques[178] are now being used to develop sensor arrays that will have detection capability with superior sensitivity for a large number of pollutants. For instance, Ivanov et al. produced sensor microarrays using screen printing. Pure nanopowder SnO2 as well as SnO2 doped with 1% of Au, Pd, and Pt were deposited on the surface Si devices. Using the difference in temperature detection and sensitivity for the undoped and doped layers for different gases, they obtained good sensitivity and selectivity for CO and NO2 detection, with minimal humidity effect.[116] More recently, they produced microarrays of pure SnO2 and WO3. The arrays

9: Gas Reactive Applications, Trudeau

399

were composed of thick films (5 μm) deposited by screen printing with grain size between 30 to 50 nm, and more compact thin films deposited by RF sputtering with grain size between 15 and 30 nm. Both materials were deposited on substrate into which electrodes of Cr/Pt were deposited with a gap of 50 or 100 μm. For their gas detection experiments they used an array of 16 sensors (four screen-printed SnO2, four screen-printed WO3, four sputtered SnO2, and four sputtered WO3). They obtained excellent detection and quantification of ethanol, acetone and ammonia binary vapor mixtures and toxic gases such as CO and NO2 by combining the results of the different sensors and using principal components analysis (PCA) or a fuzzy ARTMAP neural network.[179] In a more ominous field, Tomchenko et al.[112] have looked at the response of arrays of thick sensors based on nanocrystalline oxides (SnO2, Y2O3, Ga2O3, WO3, In2O3 and CuO mainly) for the detection of methanol but also chemical warfare agents such as sarin, soman, and mustard gas. By combining the response of various materials, their sensors were able to detect 10 ppb and higher of these agents with reliability.

9.3.3 Novel Nanostructured Systems As for most other gas reactive applications, sensors synthesis has been influenced by the recent occurrence of new classes of materials based on new structures at the nanometer level. For instance Ivanov et al.[180] combined the sensing properties of undoped and Pt-doped SnO2 with the molecular—pore wall strong interaction of nanoporous materials such as zeolite-A and silicalite to improved sensing sensitivity and selectivity. The combined Pt-doped SnO2—zeolite-A was found to have a dramatic reduction in sensitivity to ammonia and ambient moisture, while complementarily, the silicalite layer dramatically reduced the sensitivity to ethanol and benzene. The interaction between gases and porous media was recently analyzed and it was found that it is possible to predicts analytically the delay that will result from the diffusion, adsorption, and desorption processes through a microporous barrier, results that can be used to design better sensing devices.[181] Arbiol et al. investigated the properties doping mesoporous silica with 3 nm crystallites of Pt and Pd, a specific surface as high as 800 m2/g as a selective filter for gas sensors.[182] The electrical resistivity of porous silicon (pore width around 4.1 nm), produced by electrochemical anodic dissolution under different HF concentrations, was found to increase strongly in an oxidizing environment[183]

400

Properties

and porous TiO2, prepared in a similar fashion, was found to be quite sensitive to H2 in an N2 atmosphere.[184] Li and Kawi[185] prepared very highsurface-area SnO2 by surfactant-templating. They obtained mesoporous materials with a surface area as high as 157 m2/g after calcination at 450°C and with a pore diameter centered at about 4 nm. They found a linear increase in sensitivity toward 500 ppm H2 in air at 300°C with increased surface area. Mesoporous and macroporous materials were also developed by Shimizu et al.[186,187] to improve the sensing properties of SnO2 and TiO2. They managed to produce thermally stable porous structure with high surface area (253 m2/g) by a templating method employing supramolecular surfactants or poly(methyl methacrylate) (PMMA) microspheres that had improved sensitivity due mainly to the increase in specific surface. For macroporous SnO2 materials they also found that the H2 and NOx sensing properties were correlated with the size of the PMMA microspheres. Nanocomposite NiO- and Co3O4-doped porous SiO2 films prepared by sol-gel were also shown recently to be very good sensors for CO and H2 detection, with good selectivity in the 10–10,000 ppm range.[121] Carbon nanotubes, as well as other nanotube, nanobelt, and nanorodtype materials or structures, have attracted great interests in recent years for the development of various sensing devices. Kong et al. demonstrated that the conductance of semiconducting single-walled carbon nanotubes can be substantially increased or decreased by exposure to NO2 or NH3, respectively. The response time at room temperature was around 2–10 s after exposure to 200 ppm of NO2 and was 1–2 minutes after exposure to 1% NH3,[188] while detection at the 10 ppb level for multiwalled film was achieved at 165°C.[189] Valentini et al.[190] studied the effect of surface defects on carbon nanotubes and observed that O2 could be chemisorbed on topological defects and that the conductivity of semiconductor nanotubes could be changed from p-type to n-type by adsorption of O2, making them very efficient sensor for minute amounts of oxygen. Nguyen et al demonstrated that single-wall nanotubes fabricated by screen printing on al Al2O3 substrate could detect as low as 5 ppm NH3 in 500 scan 50 N2 fluxing at 80°C.[191] Picozzi et al.[192] demonstrated that carbon multiwalled nanotubes could be used for the for the detection of O3. Carbon nanotubes can also be used as a dopant. Doping of semiconducting materials with multiwall carbon nanotubes was found to improve the detection of WO3 for NO2 and CO at room temperature.[193] Carbon-based materials are the system that have attracted the most attention, but other materials are also been studied for chemical detection. Recently, Li and Yu described sensors based on single-walled carbon nan-

9: Gas Reactive Applications, Trudeau

401

otubes (in one case a nanotube-based electronic design), and also on metal oxide nanowires or nanoblets, for which the sensing properties could be enhanced through metallic doping or polymer coating.[194] Also, TiO2 nanotube arrays made by anodization techniques, coated with a 10 nm thick Pd discontinuous layer, have showed 170,000% change in electrical resistance upon exposure to 1000 ppm at 24°C. More interestingly, using the photocatalytic properties of TiO2 nanotubes, this sensor was found to be self-cleaning through UV illumination after poisoning of the surface with different grades of motor oil.[195,196] Individual β-Ga2O3 nanowires, prepared by conventional sublimation of Ga under Ar in a tube furnace and deposited on Al electrodes, showed very good sensitivity and response time to ethanol at 100°C.[197] The response of this sensor was found to be rapid enough to detect even the initial turbulence and gas pressure change. Vanadium pentoxide nanobelts, prepared by a mild hydrothermal method, have recently been shown to have good detection capability for ethanol in different humidity conditions and in the presence of other gases such as H2S, NH3, H2, C3H8, CO, and NOx.[198] Silicon nanowires of about 20 nm diameter, prepared by an oxide-assisted growth technique, have also shown some possibilities as a chemical sensor for N2 and NH3.[199] Finally, room-temperature H2 sensitivity was also observed for Pd nanowires, produced by template-manufacturing using electrodeposition onto the surface of highly oriented pyrolitic graphite. The detection was due to the closing of nanogaps caused by the Pd lattice expansion and the deformation of the metal lattice driven by a α- to βphase transition.[200]

9.4 Hydrogen Storage With the price of oil rising steeply, increasing consideration is given to the use of hydrogen as a new fuel for automotive applications. It still needs to be stressed that hydrogen is not a natural fuel but an energy vector that can be used for various applications, such as transportation, after being produced through a primary energy form: natural gas reforming, coal gasification, or electrolysis. One of the key roadblocks for the use of hydrogen remains the development of compact storage methods for 5–10 kg of hydrogen that will allow vehicle autonomy over about 500 km with a refueling time of less than 5 minutes and that will store and release hydrogen at temperatures between 0 and 100°C and pressures of 1–10 bar.[201] The Department of Energy has set a number of technical targets for 2015 that should insure mass production of vehicles. Targets include a hydrogen

402

Properties

capacity of 9.0 wt%, an energy density of 2.7 kWh/liter, a minimum full flow of 0.033 (g/s)/kW, a delivery pressure of 2.5 bar, a refueling rate of 2.0 kg H2/min; and a cost of $2/kWh.[202,203] At the same time, it is also required that the overall propulsion system (motor and storage system) should not operate above 100°C.[204] A number of storage methods are presently available:[204–207] hydrogen can be transported as compressed gas or in liquid form, both of which call for very special care to take into account all the safety concerns. Moreover, the liquefaction process consumes almost 30% of the hydrogen energy to compress and cool the hydrogen to 20 K, making any use of liquid hydrogen very difficult for large-scale transportation applications. Adsorption on activated carbon with a very high surface area (1500– 2000 m2/g, especially at low temperature (77 K) and high pressure (∼60 bar), can also be used, but again the problem of refrigeration increases the difficulties and the cost of this approach. Hydrogen reversible capacities around 10.5 wt% with a volumetric density of 41 kg/m3 are presently achievable.[205] Another possibility is to use pure iron and its transformation to oxide (rust) by water vapor to produce hydrogen in situ. One of the drawbacks of this process, apart from the weight aspect, is the need to recycle the oxides. A solution would be to use a high-tech process such as municipal waste to reduce the iron oxide. Another possibility is metallic or complex hydrides, materials that can absorb hydrogen at moderate pressure and are operated at temperatures ranging from ambient to about 200–300°C. They can store hydrogen with a higher volume density than in liquid form. Three types of hydride exist: ionic (such as magnesium), covalent, and metallic (hydrides of transition metals). A major drawback of hydrides is the increase in weight needed for storage. On the other hand, they are much safer, since the hydrogen must be released from the hydride before it can oxidize or burn rapidly. For nearly two decades, car manufacturers have been testing metal hydride storage in hydrogen propulsion test vehicles.[208] A number of alloys have been investigated and it can now be shown that structural design on the nanometer scale may play a significant role in the synthesis of new and more efficient hydrogen-storage materials.[209] It should also be mentioned that metallic hydrides are finding applications in the area of rechargeable batteries, with Ni-metal hydride batteries already on the market. Their main advantages are their high energy density, high dischargeability, long charge–discharge cycle life, and environmental cleanliness.[210] Also in this field, the design of new nanostructured materials is expected to prove critical for future technological development.[211–214]

9: Gas Reactive Applications, Trudeau

403

Finally, new nanostructured materials such as nanotubes or mesoporous composites are now starting to show very promising storage capacity.

9.4.1 Properties of Hydrogen-Storage Compounds Hydrogen gas storage compounds require many unique properties, which is probably the main reason why they are so few. In order for hydride materials to be of commercial value, it is essential that the following parameters be optimized:[208] • • • • • • •

weight capacity kinetics of H2-exchange sensitivity to impurity gases multi-cycle stability activation procedure large-scale production at low cost

Except for the last parameter, which is obvious, it is worth discussing these different parameters in order to understand the possibilities offered by the nanostructural design of hydride materials. Weight: The question of weight, as mentioned above, is essential. To be viable as a hydrogen-storage compound, the material used must be as light as possible, which is why Mg-based alloys have attracted attention, even if their oxidation characteristics make them more difficult to handle than other intermetallic compounds such as FeTi. Capacity: This is, in part, related to the weight aspect in that the aim is to be able to store as many hydrogen atoms per metallic atoms as possible. Kinetics of the H2-exchange: Typical hydrogen absorption desorption curves are presented in Fig. 9.13 for some LaNi alloys.[215] First-order kinetics describe the reaction rate according to c(t) = c0(1 − e−Kt)

Eq. (9-15)

where c(t) is the hydrogen concentration in metal hydride, c0 is the saturation concentration, and K = K(p,T) the reaction constant.[216] The

Properties

404

Figure 9.13 Typical H2 absorption–desorption curves.[215]

important point is that an appropriate material should have an absorption— desorption cycle well below 300°C, with rapid rates. In order to identify the rate-limiting step, the hydride formation has been divided into five intermediate processes by Martin et al.:[217] 1. 2. 3. 4.

Physisorption of hydrogen molecules Dissociation of hydrogen molecules and chemisorption Surface penetration of hydrogen atoms Diffusion of hydrogen atoms through the hydride layer, either by an interstitial or by a vacancy mechanism 5. Hydride formation at the metal/hydride interface A similar five-step process was also developed for hydrogen desorption. Sensitivity to impurity gases: The presence of gaseous impurities was shown to have an impact on the cycling stability and kinetics of materials. The effect of these gases depends on their nature. For example, N2 and CH4 were found to reduce only the kinetics of the hydrogen absorption, while even small amounts of impurities such as O2, CO, CO2, and H2O reduce the capacity and cycle stability.[208] In most cases, a reactivation process in pure hydrogen is necessary to regain the initial capacity of the material.

9: Gas Reactive Applications, Trudeau

405

Multi-cycle stability: The material must be able to undergo a number of absorption—desorption cycles without a significant loss in capacity. Moreover, one of the major problems of hydride materials relates to their decripitation with the charging and discharging cycles. A strong variation in the crystalline volume is caused by the absorption—desorption of hydrogen. For example, the unit cell volume in the case of Ti45Zr38Ni17 C14 hcp causes a phase increase of about 24% upon hydrogenation.[218] This variation in volume has a direct effect on the generation of crystalline defects, resulting in fracture of the crystals after a number of cycles. One main advantage of nanocrystalline materials can already be seen: high-energy milling studies have revealed that defect-induced breakdown of crystals (such as the presence of dislocations) is very limited at small sizes.[219] Activation procedure: Most metallic hydrides do not readily absorb hydrogen. Before they can do so, some kind of activation process, which normally comprises a number of cycles from room temperature to high temperature (e.g., 400°C), under high hydrogen pressure, must be performed. Two possible explanations have been proposed for this behavior. The first relates the main effect of this process to removal of the surface oxide layer or other impurities present on the surface of the grains. The second is that the increase in surface area caused by the decreased particle size (due to decripitation) provides a larger surface area upon which the hydrogen can react.[220]

9.4.2 Metallic Hydrides Until recently most studies to improve the storage capacity of materials using structural design on the nanometer scale have been done on metallic hydrides and in most cases through high-energy mechanical milling or alloying.[222] Some of the earlier work on nanostructured hydrogen-storage materials was done on FeTi alloys produced by high-energy mechanical milling. As presented in Table 9.1, FeTi is a well-known intermetallic hydrogen-storage compound. Trudeau et al.[223] and Zaluski et al.[224] have shown that it is possible to obtain FeTi powder with an average crystallite size of 7–10 nm by milling together the elemental components or by mechanically grinding microcrystalline powders. It was also demonstrated that the presence of a low concentration of oxygen (3.6 t%) was sufficient to produce an amorphous material. Figure 9.14 presents the first comparison of H absorption for microcrystalline (curve c), nanocrystalline (curve b), and amorphous (curve a) Fe50Ti50.[225,226] These data show that the nanocrystalline sample had absorption properties between normal

Properties

406 Table 9.1 Hydrogen Capacity for Some Hydrogen Storage Alloys[208,218,221]

Material Mg MgNi2 Mg1.92NiAl0.08 TiV1.5Fe0.4Mn0.1 Ti45Zr38Ni17 Ti.98Zr.02V.43Fe.09Cr.05Mn1.5 FeTi LaNi5

Hydrogen wt.% 7.7 3.6 3.5 3.3 2.5 1.8 1.6 1.5

Figure 9.14 Pressure–composition isotherms for (a) amorphous, (b) nanocrystalline (5 nm), and (c) microcrystalline Fe50Ti50 alloy.[224]

coarse-grained material and amorphous alloys. Nanocrystalline Fe50Ti50 was found to have a lower pressure plateau than the microcrystalline sample, albeit with a reduced capacity of one hydrogen atom per metallic element compared to about 1.2 for its microcrystalline counterpart (at low pressure).[226] Further studies on a number of alloys in the FeTi system

9: Gas Reactive Applications, Trudeau

407

revealed the presence of an amorphous layer (from 10% to 30% of the materials) at the surface or at the boundaries of FeTi crystals. It is assumed that, below the absorption plateau the hydrogen absorption in this amorphous layer exerts a negative pressure on the FeTi crystal while above the plateau the expansion of the crystal is restricted by the amorphous layer and that the stress increases with further concentration of the hydrogen in the crystal.[225] Modeling based on this layer has successfully explained the decrease in the absorption plateau as well as its narrowing. Hydrogen absorption also revealed that activation was much easier for the nanocrystalline materials than for the normal microcrystalline alloy. For the coarse-grained sample, activation consists of several cycles of heating up to the range of 400–450°C in vacuum and annealing in H2 under 7 bars, followed by cooling to room temperature and admission of hydrogen at a pressure between 35 and 65 bars. Apart from the decripitation hypothesis, two models have been proposed to explain the need for this activation. The first proposes that activation transforms the surface iron oxide to metallic iron with the formation of TiO2 and that the presence of metallic Fe domains would be responsible for dissociating molecular hydrogen. The second model proposes that the formation of an iron–titanium mixed oxide would produce the hydrogen sorption behavior.[227] On the other hand, activation of Fe50Ti50 nanocrystals requires only a single annealing treatment at 400°C for 0.5 hour under vacuum. In order to investigate this difference, the oxidation and the surface chemistry of nanocrystalline Fe50Ti50, were studied using XPS and compared to those of the microcrystalline alloys. This study shows that Fe50Ti50 oxidizes more easily than its microcrystalline counterpart and that, at temperatures below 500°C, Fe2TiO5 and TiO2 are mainly formed, resulting in dissociation of the intermetallic compound and the formation of metallic iron. Table 9.2 presents the surface elemental atomic concentration and the Fe : Ti metallic ratio of various Fe50Ti50 samples of different grain size. It has been proposed that the simpler activation needed for the nanocrystalline samples is related in part to Fe enrichment of the surface (the surface of nanocrystalline samples having twice the amount of Fe on the arc-melted coarse-grained sample) and to its ability to accomplish the reduction–dissociation transformation. This is further confirmed by the results of two samples analyzed after a hydrogenation cycle, which show a decrease in surface Fe and an increase in Ti and O simultaneously with a drastic hike in the Fe : Ti metallic component ratio, even if the sample was exposed to air prior to the XPS analysis. This tends to indicate that in this nanostructure configuration the Fe metallic surface is stable, since it does not completely oxidize in air. These results support the model that

Properties

408

Table 9.2 Variation of Surface Species Obtained by XPS for Different FeTi Samples[227]

Sample Type Arc melted Powder, 39 nm Powder, 24 nm Powder, 21 nm Powder, 13 nm Powder, 10 nm Powder Ha, 24 nm a

Powder H , 10 nm a

Fe (at%)

Ti (at%)

O (at%)

C (at%)

Surface Fe/Ti

Fe:Ti metallic

11.7 20.1 36.2 23.6 26.9 20.1

7.7 7.6 6.9 6.6 7.0 5.4

45.6 46.1 48.4 45.7 45.5 43.1

35.0 26.3 18.6 24.1 20.6 31.3

1.5 2.7 3.8 3.6 3.8 3.7

1 0.48 0.43 0.53 0.49 0.47

14.7 11.7

12.4 13.7

52.6 52.4

20.3 22.2

1.2 0.9

>4.8 >3.6

Powder H, after a hydrogenation cycle.

metallic iron acts as the catalytic surface for dissociating the hydrogen molecule in the absorption process, in agreement with other results.[228] In subsequent work, Zaluski et al.[229] studied the storage properties of nanocomposite Pd-doped FeTi prepared by milling FeTi with Pd powder. The result was the formation of a nanocomposite with Pd grains (30 nm) cold-welded to a number of FeTi nanocrystals. The nanocomposite had hydrogenation properties superior to the pure nanocrystalline material, requiring no activation annealing, for instance, even if stored in air. Moreover, absorption was found to be much easier with a hydrogen uptake of about 0.6 H atom/FeTi at pressures lower than 1 bar compared to about 0.2 for the undoped nanocrystalline powders. Wasz and Schwarz[230] studied the hydrogen storage properties of different LaNi5-based intermetallic alloys prepared by high-energy mechanical alloying, and especially LaNi5–ySny. The presence of Sn was found to reduce the plateau pressures for hydrogen absorption and desorption, to decrease the hysteresis between the pressure for hydride formation and decomposition, and, unfortunately, to decrease the hydrogen storage capacity. The latter behavior was explained by changes in the LaNi5 electronic 3d band structure with Sn alloying, resulting in a reduction in the number of holes that can be occupied by electrons from the hydrogen. As in other studies, they found that an activation cycle was not necessary for the powders to absorb hydrogen, which they attributed to the presence of

9: Gas Reactive Applications, Trudeau

409

Ni inclusions or, more probably, to the absence (or reduced presence) of La oxides in their powders. In their work, they also observed that nanocrystalline powders were less affected by the volume change variation due to hydrogen charging. In LaNi5S0.25 the surface-area increase between the first and second hydride cycles was only 1 × 10−3 m2/g compared to 8 × 10−2 m2/g in arc-cast powder. A similar improvement of the activation performance was found for quenched Ti28V15Mn10Cr that was surface modified by ball-milling with 10 wt% of LaNi3.55Co0.75Mn0.4Al0.3, producing a nanocomposite Ti28V15Mn10Cr material covered by LaNi3.55Co0.75Mn0.4Al0.3 that acts as a catalyst.[231] Probably because of their light weight and their high hydrogen capacity, nanostructured Mg and Mg-based alloys have drawn more attention in recent years in the field of nanostructured hydrogen-storage materials, even if the poor dissociation ability of metallic Mg due to the low probability of adsorption of H2 molecules on metallic Mg (around 10−6) seems problematic for the technical applications of Mg systems.[232] Zaluski et al.[233] found that nanocrystalline Mg2Ni formed by mechanical alloying, with average grain sizes between 20 and 30 nm, absorb hydrogen readily without the need for activation at temperatures lower than 250°C. They also observed that these nanocrystalline materials absorb hydrogen at lower temperatures (∼200°C), i.e., below the temperature of structural transformation of the Mg2NiH4 hydride. Moreover, as found for FeTi and also for LaNi5, the addition of Pd was found to enhance the hydrogen absorption kinetics at lower temperature, and absorption at room temperature was observed even without the need for an activation cycle.[234,235] Addition of Cu, on the other hand, was found to increase the plateau pressure. Li et al.[236] showed that mechanical alloying could be used to prepare Mg2Ni hydrogen-storage compounds with a very fine structure and a surface Mg/Ni atomic ratio of 1 : 3 after sintering at 400°C for 5 hours. This Ni concentration at the surface, and its aid in the diffusion and decomposition of the hydrogen, could be responsible for the good properties of these materials, especially the greatly reduced need for an activation process. The active role of Ni in Mg—Ni alloys was also demonstrated by Berlouis et al.,[237] who studied alloys with Ni content ranging from 0.1 at% to 10 at%. Imamura and Sakasai[238] studied the hydrogen absorption characteristics of mechanically milled Mg with and without 5 wt% Pd-supporting graphite in the presence of tetrahydrofuran (THF) with an average crystallite size between 17 and 34 nm. The complex nanocomposite materials (Mg/Pd/G)THF were found to have a high surface area and very good absorption properties that were dependent not so much on the crystallite

410

Properties

size as on the amount of THF used. Upon milling, THF was found to decompose the graphite and to form an intimate mixture of Mg (or Mg/Pd) and aromatic carbon molecules, resulting in a charge-transfer interaction that could be the reason for the observed enhanced activity.[239] Holtz and Imam[240] studied the hydrogen storage capacity of nanostructured magnesium alloys prepared using three different techniques: inert-gas condensation of sputtered nanocrystalline powder, co-sputtering of amorphous thin films, and high-energy milling. From these three synthesis methods, the best materials were alloys with 5 at% and 10 at% Ni that were high-energy-milled for 50 hours in mineral oil followed by consolidation, which achieved about 6 wt% hydrogen loading without requiring an activation cycle. The other synthesis methods were found to produce materials with low kinetics and very prone to grain growth and oxidation. Based on the previous work of Imamura and Sakasai,[238] it would be interesting to investigate the possible catalytic effect of mineral oil and its decomposition to C groups on the hydrogen absorption properties of the mechanically alloyed powder. Orimo and Fujii[241] studied the properties of the Mg2Ni—H system synthesized by milling the metallic powders in a hydrogen atmosphere. They found that the hydrogen content in the powder reached a stable value of about 1.6 wt% after 1 hour of milling, which is 1.5 times higher than the value for H absorption in Mg2Ni at ambient temperature. Also, the dehydriding reaction was found to occur at 440 K, which is much lower than for the normal low-temperature phase of Mg2NiH4 (520 K). To improve the hydrogen absorption–desorption kinetics of Mg-based alloys, Gross et al.,[242,243] investigated the hydriding properties of composites obtained by mechanically alloying La2Mg17 with various weight percentages of LaNi5. The composite materials, La2Mg17 + 40 wt% LaNi5, were found to have a reduced absorption capability with a hydrogen capacity of 3.7 wt% but an absorption–desorption rate 50 times that of pure La2Mg17. The decrease in capacity seems to have been circumvented by Li et al.,[244] who produced a new nano-ternary Mg–8mol% LaNi0.5 through melting and subsequent mechanical alloying, which had a hydrogen absorption capacity of 4.6–7.0 wt% under 30 bar and a desorption of 4.4–6.9 in 600 seconds for temperatures between 150 and 300°C without the need for a prior activation process. Similarly, Wang et al. investigated the properties of various Mg—ZrFe1.4Cr0.6 synthesized by mechanical grinding and observed also good hydrogen absorption capacity and very good kinetics that they attributed to the combined effects of the full reaction of the catalytic function of the Mg—ZrFe1.4Cr0.6 and the nanostructure of the Mg.[245]

9: Gas Reactive Applications, Trudeau

411

A variety of Mg nanocomposites have also been investigated in the last couple of years. Zhenxing et al.[246] studied the properties of Mg—2Ni— 2Cu (wt%) with the presence of 1 wt% CrCl3 produced by high-energy milling for 110 hours and found that the materials could absorb 5.5 wt% at 200°C in 60 seconds under 20 bar and that the presence of CrCl3 was critical to improve the kinetic of the hydrogenation. Oelerich et al.[232,247] studied the properties of a variety of nanocrystalline MgH2/MexOy composite powders prepared by high-energy ball milling (MexOy = Sc2O3, TiO2, V2O5, Cr2O3, Mn2O3, Fe3O4, CuO, Al2O3, and SiO2) and found that the transition-metal oxides act as catalysts for the magnesium–hydrogen reaction. The best catalyst was found to be Nb2O5, which allows absorption of 6 wt% at 250°C under 8.4 bar in 60 seconds and desorption in 500 seconds.[248] They also found that some carbines or nitrides could show significant enhancement of the hydrogen reaction kinetics.[249] Bobet et al. confirmed these improved kinetic properties in the case of the nanodispersion of Cr2O3 at 200 and 300°C and 16 bar.[250] Similarly, Wang et al. found that Mg95Ni3(MnO2)2 nanocomposite, produced by ball milling, can absorb more than 6 wt% in 60 seconds at 200°C and 20 bar, with desorption in 400 seconds at 310°C under 1 bar.[251] Recently, Yavari et al.[252] produced a new nanocomposite materials through the reactive high-energy milling of MgH2 with a few percent of FeF3 producing nanograins of MgH2 with a protective layer of MgF2 and the formation and dispersion of Fe nanoparticles. This nanocomposite was found to have very fast absorption and desorption kinetics at 300°C of about 200 seconds. The positive role of Fe was also found in nanostructured MgH2—Fe by Bessetti et al., who found a decrease of about 100°C in the desorption peak in samples with similar microstructure and particle size.[253] Other techniques have been used to improve the surface area and the absorption–desorption kinetics. For example, Huot et al.,[254] used a lixiviation technique to produce a highly porous hydride material. They produced Mg72Li28 with an average size of about 46 nm by high-energy mechanical alloying. Lixiviation was done by immersing the powder in distilled water, with the lithium dissolving with gaseous evolution, resulting in a 10-fold increase in the surface area (from 1 to about 10 m2/g). Although the lixiviated sample was found to have better absorption and desorption kinetics, its hydrogen capacity was lower than that of the nonlixiviated sample due to the presence of magnesium hydroxides produced by lixiviation. On the other hand, by better control of the Pd—Mg structure at the nanometer scale through the use of RF-associated magnetron sputtering, producing Mg layers with a controlled columnar-like textures, Fujii

412

Properties

et al.[255] managed to obtain excellent absorption of ∼5 wt% at 100°C and 1 bar for a three-layered Pd(50 nm)/Mg(800 nm)/Pd(50 nm) film. The excellent storage properties were explained by these researchers through the cooperative phenomena in nanocomposite regions through some elastic interactions. It is thus clear that the structural transformation at the nanometer level can play an important role in the hydrogenation properties of metallic hydride materials. The structural effects were further investigated in some Mg—Ni—RE (RE = Y- or Ce- or La-rich mischmetal) produced through rapid quenching and subsequent crystallization by Spassov et al.[256] Again, they found that the amorphous or partially or fully nanocrystallized materials had better hydrogenation kinetics than the normal polycrystalline materials, probably due to the faster hydrogen diffusion in the amorphous (disordered) phase. However, they also found, as most other researchers, that the hydrogen capacity was reduced compared to similar coarse-grained alloys. Dehouche et al.,[257] in studying the cycling stability of MgH2—Cr2O3, observed an increase in the hydrogen capacity by 8% after 1000 cycles at 300°C, but with a reduction of the desorption kinetics as the crystallite size increases from 21 to 84 nm. The decrease in hydrogen capacity could be due to the interaction between the hydrogen in solution and the presence of strain or stress fields near grain boundaries and phase boundaries.[258] This was demonstrated by Hanada et al.[259] in their study of the hydrogen-storage properties as a function of the crystallite size and lattice strain for mechanically milled MgH2, and was discussed in detail by Aizawa et al.[260] in their studies of the hydrogenation of nanostructured Mg2Ni produced by bulk mechanical alloying. They argued that the reduction in hydrogen capacity comes from the site energy distribution induced by the intrinsic strains. During size refinement by high-energy milling, each nano-grain is elastically strained and, because of the presence of intrinsic strain distribution over the nano-grains, the hydrogen site distribution is expanded in proportion to the amount of strain. In analogy to amorphous materials, for which the site energy distribution is typically continuous, because lower-energy sites might be present at or near grain boundaries, absorption and desorption temperatures can be expected to be lower in nanostructured materials than in normal coarse-grained ones.[260]

9.4.3 Complex Hydrides The difficulties in obtaining the desired storage capacity in metallic hydrides forced researchers to look at other systems. In recent years,

9: Gas Reactive Applications, Trudeau

413

Table 9.3 Hydrogen Capacity of Some Complex Hydride Compounds[261,262]

Materials LiBH4 Al(BH4)3 LiAlH2(BH4) Mg(BH4)2 Ti(BH4)3 NaBH4 LiAlH4 NaAlH4

Hydrogen wt% 18.4 16.9 15.3 14.9 13.1 10.6 9.5–10.6 7.5

complex hydrides, such as AlH4 or BH4, have started to attract attention because of their high hydrogen contents.[261] Table 9.3 presents some of these complex hydrides with their total hydrogen capacity. Until recently, little attention was given to these ionic or covalent complexes because it was thought that these systems would release hydrogen irreversibly at conditions outside the desired temperature and pressure ranges.[262] However, Bogdanovic and Schwickardi found that using suitable catalysts such as TiCl3, Ti- or Fe-alcoholates, or FeCl2, nanocrystalline NaAlH4 could reversibly absorb 5.6 wt%[263,26] hydrogen at temperature of 180 and 210°C following the reactions 1 2 3 NaAlH 4 ↔ Na 3 AlH 6 + Al + H 2 ↔ NaH + Al + H 2 3 3 2

Eq. (9-16)

As for metallic hydrides, various structural transformations at the nanometer level are being investigated in order to improve the storage properties of these compounds. Jensen et al.[265] further improved the properties of this compound by doping it with Ti(OBun)4 using mechanical grinding and also using wet chemical techniques as in Ref. 263. They found that the method of synthesis had a large impact on the hydrogen desorption, with the material produced by grinding having a significantly larger amount of catalytically active titanium sites. Gomes et al.[266] published results on the effects of milling and doping on storage properties of NaAlH4. They found that milling with and without dopants tends to increase the axial cell parameter ratios, cell volumes, atomic displacement amplitudes, and strain, and to decrease grain size. However, the presence

414

Properties

of dopants, such as Ti, could be seen to enhance the kinetics during the re-crystallization by diffusion and substitution into the alanate lattice. Finally, as another example, Easton et al. studied the effect of the addition of TiCl3 to LiAlH4 by high-energy milling and found that the decomposition temperature of LiAlH4 as well as that of Li3AlH6 was greatly reduced.[267] Researchers are also greatly interested by LiBH4 since it is the compound with the highest hydrogen content. Zuttel et al. have recently found that 13.5 wt% could be released at around 200°C if SiO2 was added to its structure.[268,269]

9.4.4 Porous or Surface-Related Nanostructures The cost, the low specific uptake per weight, the unfavorable kinetics that require heating cycles, and the susceptibility to contamination by impurities makes the use of metal hydrides at the present time somewhat problematic for economic hydrogen-storage applications.[270] Because of this, researchers have looked at a number of various nanostructured materials that have started to show some interesting hydrogen storage potential.[271] One of the best examples is the large amount of work that is ongoing in carbon nanotubes and other carbon nanostructures. Indeed, hydrogen storage on various forms of carbon has been in progress for a long time. However, because the binding energy of hydrogen on activated carbon is only about 60 meV, this leads to conditions of very high pressure (several hundreds of bars) and/or very low temperature (∼77 K) for the storage of meaningful amounts of hydrogen despite the high surface area of up to 2600 m2/g.[272] The main interest in nanotubes comes from the curvature of the graphene sheets and the cavity inside the tube, with a width not exceeding a few molecular diameters, which makes the potential fields from opposite walls overlap in such a way that the attractive force acting on adsorbate molecules will be increased compared with that on a flat carbon surface.[273] A first report on the hydrogen storage capacity of nanotubes was presented by Dillon et al.,[274] who gave values between 5 and 10 wt%. Moreover, in 1998, Chamber et al. reported a storage capacity of nearly 67 wt% in some carbon nanofibers,[275] results that have not been reproduced up to now. Since then, the scientific community has investigated all these new forms of carbon with, unfortunately, large variability in results. This inconsistency seems to be partly related to the amount of impurities still present in the nanotubes, as well as the pretreatment that

9: Gas Reactive Applications, Trudeau

415

they underwent prior to the H2 storage experiments.[276] Results for singlewall nanotubes with a mean diameter of 1.85 nm give values of the order of 4.2 wt% at room temperature and under 100 bar, with 78.3% that could be released under ambient room temperature and pressure.[277] For their graphitic nanofibers, Gupta and Srivastava found a storage capacity near 15 wt% after a “suitable” activation process.[278] On the other hand, Ritschel et al. found values below 1 wt% for H2 capacity in a variety of carbon nanostructures for different temperature and pressure conditions.[279] Tarasov et al.[280] found reversible hydrogen capacity of 2.4 wt% only at cryogenic temperature and at pressure of 25 bar. They also observed that single-walled nanotubes with a larger diameter of 1.4 nm produced using YNi2 catalyst had a higher storage capacity than other nanotubes of 1.2 nm diameter that were produced using a 3Co/Ni catalyst. So far, the indication in carbon nanostructures is that the low-temperature reversible hydrogen storage capacity is proportional to the specific surface area of the material, which indicates a physisorption process,[281] with a value of the order of 1.5 wt% per 1000 m2/g and thus a maximum of 2 wt% (for a specific surface of 1315 m2/g).[282] Recent data seem to indicate that the hydrogen storage capacity could be improved through surface modification. For instance, Zacharia et al. have increased the storage capacity of their nanotubes by 30% by doping them with Pd and V nanocrystals (of average size 2–3 nm), under 20 bar in what they supposed to be a spillover phenomenon.[283] However, at the present time, nanotubes are far from being the ideal hydrogen-storage system. The same variability is also found for other forms of carbon nanostructures. For instance Orimo et al.[282,284] found that the hydrogen storage capacity of nanostructured graphite prepared by mechanical milling in an H2 atmosphere reached 7.4 wt% after 80 hours of milling, while Shindo et al.[285] found, under similar milling conditions, no major effect of milling time, after 10 hours of milling, in a number of carbon materials with a final identical storage capacities of about 2.6–3.0 wt%. However, even if the hydrogen capacity is as high as Orimo et al. observed, the desorption temperature is high (above 700 K) and re-loading in a hydrogen atmosphere is impossible.[281] Carbon–metallic nanocomposites are also being investigated. Imamura et al.,[286] for instance, studied the storage capacity of mechanically milled magnesium and graphite carbon with organic additives (benzene, cyclohexane or tetrahydrofuran) and found in some cases a decrease in the MgH2 decomposition onset temperature and also the formation of additional hydrogen uptake sites. More recently, Ichikawa et al.[287] found a rechargeable hydrogen capacity of more than 4 wt% below 350°C in a

416

Properties

2 : 1 mixture of nanostructured carbon and lithium hydride prepared by ball milling. They mention that the properties of their mixture are due to the close contact on the nanometer scale of the strong C—H and Li—H ionic bonds. Other nanotube materials are also being investigated for their storage properties and could potentially be good hydrogen-storage media.[272] Chen et al.[288] have studied the storage capacity of multiwall MoS2 nanotubes treated with KOH and have found that they can store about 1.2 wt% at 25°C under 12 bar. Boron nitride nanotubes were found to be able to absorb between 1.8 and 2.6 wt% (1.8 in multiwall tube and 2.6 in bamboo-like tubes) of hydrogen at 100 bar,[289] while TiS2 nanotubes have been found to have a reversible hydrogen capacity of 2.5 wt% at 25°C and 40 bar.[290] Another variety of nanostructured materials that are studied for their potential storage capacities are new porous materials, with pore sizes in the nanometer range, that can now be synthesized with a high level of control over the pore size, structure, and shape. An example is microporous metal–organic frameworks (MOF), one of which, Zn4O(1,4benzenedicarboxylate)3, with a cubic three-dimensional extended porous structure, was found to adsorb 4.5 wt% of H2 at 77 K and 1 wt% at room temperature and 20 bar.[270] Pan et al.[291] found for a crystal of [Cu(hfipbb)(H2hfipbb)0.5][H2hfipbb=4, 4′-(hexafluoroisopropylidene)-bis(benzoic acid)], which has an accessible volume of 11.6% as compared to 76.8% for previous materials, a storage capacity of 1 wt% at 48 bar. These two results indicate that tailoring of the pore size at the nanometer level to fit the molecular diameter of the stored molecule seems to be required in order to achieve optimum storage capacity. The same can be said for new mesoporous materials whose composition, pore size, distribution, and connectivity could be tailored for this particular application. Some mesoporous titanium oxides were found to have a reversible absorption level between 3 and 5 wt% when suitably reduced by bis(toluene)-titanium, bis(toluene)-vanadium, or a variety of lithium fullerides.[271] One needs also to stress the vast number of possible combinations between the different groups of hydrogen-storage nanomaterials, such as the work of Chen et al.,[292] who studied the properties of nanocomposites based on Mg and multiwalled nanotubes prepared by catalytic reactive ball milling. For a sample with 5 wt% of nanotubes they found an absorption capacity of 5.34 wt% at 373 K under 20 bar in 15 minutes with a desorption of 3.62 wt% at 473 K in 30 minutes.

9: Gas Reactive Applications, Trudeau

417

9.5 Conclusion In this chapter, three surface reactive applications of nanostructured materials were presented. They bear witness to the great potential of nanostructured materials for large-scale technological applications. In fact, the examples used indicate that large-scale applications of nanostructured materials already exist and, in some circumstances, have been in place for more than fifty years, without the label “nanostructured” ever being voiced. The work described here also indicates that much more is needed and more efficient nanostructures need to be designed, in all these fields, to produce ideal catalysts, chemical sensors, and hydrogen-storage products. This review is far from being an exhaustive description of the possibilities of nanostructured materials as gas reactive applications. Other applications are readily envisaged. For example, because of their surface reactivity, there is an immense potential for the application of nanostructured materials as getters.[293] Moreover, because of their increased diffusivity, nanostructured materials could also be ideal materials for the development of new separating membranes.[294] It should be remembered, lastly, that the definition of nanostructured materials is somewhat ambiguous. However, it is the feeling of the present author that the large impact of the surface atoms, or of the high proportion of surface atoms, will make itself felt for materials with an average crystallite size well below 10 nm. At 10 nm the number of atoms at grain boundaries is about 20%, while this value increases to nearly 50% for an average value of 5 nm. In most studies described here, the average value of the crystals was 10 nm or varied between 10 and 20 nm. The fundamental question to be answered in the near future is what improvement we can expect when this average crystallite size can be efficiently tailored around 5 nm. For gas-sensing materials, for instance, it was mentioned that the value of the electron-surface-depleted layer is of the order of 3 nm in SnO2, which means that only grains below 6 nm will be completely depleted. One question that is still not resolved for sensors is the nature of the defects and their impact on the sensing properties. It is understood that oxygen vacancies play a major role, but what exactly role does the surface defect structure of nanostructured materials play? The stronger presence of these structures will have a direct impact on the concentration of vacancies while at the same time improving the surface diffusivity; both effects should enhance the sensitivity as well as the rapidity of detection.

418

Properties

It can now be foreseen that two main difficulties to be circumvented, once materials or composites of such small average size are easily synthesized, will be the control of the structure size during operation, which could probably be achieved by means of growth inhibitors or by lowering of the operating temperature, and the reproducibility of the materials with specific compositions and structures. For both these difficulties, new characterization tools will have to be developed to insure commercial viability. Much more work in material synthesis and characterization, as well as in the development of theoretical models, awaits researchers in all these domains if they are to bloom. However, the technological potential in all of them is dauntingly enormous. In order that rapid evolution in materials, properties, and technological concepts be achieved, it is essential that the findings from all fields be pooled by common assent and that researchers from all fields work together.

Acknowledgments The author would like to thank Professor Jackie Y. Ying, Dr. Virgil Provenzano, and Professor David Antonelli for their friendship and collaboration.

References 1. Klabunde, K.J., Li, Y.X., and Khaleel, A., “Catalysis and Surface Chemistry: Metal Clusters/Nanoscale Particles,” in: Nanophase Materials Synthesis–Properties–Applications (G.C. Hadjipanayis, and R.W. Siegel, eds.), 260:757–769, NATO ASI Series, Kluwer Academic, Dordrecht (1994). 2. Ippommatsu, M., Ohnishi, H., Sasaki, H., and Matsumoto, T., “Study of the Sensing Mechanism of Tin Oxide Flammable Gas Sensors using the Hall Effect,” J. Appl. Phys., 69:8368–8374 (1991). 3. Trudeau, M., and Ying, J.J., “Nanocrystalline Materials in Catalysis and Electrocatalysis: Structure Tailoring and Surface Reactivity,” Nanostruct. Mater., 7:245–258 (1996). 4. Henry, C.R., “Adsorption and Reaction at Supported Model Catalysts,” in: Catalysis and Electrocatalysis at Nanoparticles Surfaces (A. Wieckowski, E.R. Savinova, and C.G.Vayenas, eds.), pp. 239–280, Dekker, New York (2003). 5. Davis, S.C., and Klabunde, K.J., “Unsupported Small Metal Particles: Preparation, Reactivity, and Characterization,” Chem. Rev., 82:153–208 (1982). 6. Markovic, N.M., Radmilovic, V., and Ross, P.N., Jr., “Physical and Electrochemical Characterization of Bimetallic Nanoparticle Electrocatalysts,” in: Catalysis and Electrocatalysis at Nanoparticles Surfaces (A.

9: Gas Reactive Applications, Trudeau

7. 8.

9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

20. 21.

22.

419

Wieckowski, E.R. Savinova, and C.G.Vayenas, eds.), pp. 311–342, Dekker, New York (2003). Poirier, J.A., and Stoner, G.E., “Microstructural Effects on Electrocatalytic Oxygen Reduction Activity of Nano-Grained Thin-Film Platinum in Acid Media,” J. Electrochem. Soc., 141:425–430 (1994). Vayenas, C.G., Pliangos, C., Brosda, S., and Tsiplakides, D., “Promotion, Electrochemical Promotion, and Metal-Support Interactions: The Unifying Role of Spillover,” in: Catalysis and Electrocatalysis at Nanoparticles Surfaces (A. Wieckowski, E.R. Savinova, and C.G.Vayenas, eds.), pp. 667–744, Dekker, New York (2003). Rupprechter, G., Hayek, K., and Hofmeister, H., “Characterization of Oxide Supported Rhodium Nanocrystals: Habit, Surface Structure and Catalytic Activity,” Nanostruct. Mater., 9:311–314 (1997). Claus, P., Hofmeister H., and Mohr, C., “Identification of Active Sites and Influence of Real Structure of Gold Catalysts in the Selective Hydrogenation of Acrolein to Allyl Alcohol,” Gold Bull., 37:181–186 (2004). Heiz, U., and Schneider, W.D., “Nanoassembled Model Catalysts,” J. Phys. D, 33:R85–R102 (2000). Gamarnik, M.Y., “The Physical Nature of Changes of Lattice Parameters in Small Particles,” Phys. Status Solidi B, 178:59–69 (1993). Zhdanov, V.P., and Kasemo, B., “Kinetics of Rapid Reaction on Nanometer Catalyst Particles,” Phys. Rev. B, 55:4105–4108 (1997). Peuckert, M., Yoneda, T., Dalla Betta, R.A., and Boudart, M., “Oxygen Reduction on Small Supported Platinum Particles,” J. Electrochem. Soc., 133:944–947 (1986). Zhao, J., Huggins, F.E., Feng, Z., Lu, F., Shah, N., and Huffman, G.P., “Structure of a Nanophase Iron Oxide Catalyst,” J. Catal., 143:499–509 (1993). Feng, Z., Zhao, J., Huggins, F.E., and Huffman, G.P., “Agglomeration and Phase Transition of a Nanophase Iron Oxide Catalyst,” J. Catal., 143:510–519 (1993). Wang, C.C., Zhang, Z., and Ying, J.Y., “Photocatalytic Decomposition of Halogenated Organics Over Nanocrystalline Titania,” Nanostruct. Mater., 9:583–586 (1997). Davis, S.C., and Klabunde, K.J., “Unsupported Small Metal Particles: Preparation, Reactivity, and Characterization,” Chem. Rev., 82:153–208 (1982). Stark, J.V., Park, D.G., Lagadic, I., and Klabunde, K.J., “Nanoscale Metal Oxide Particles/Clusters as Chemical Reagents. Unique Surface Chemistry on Magnesium Oxide as Shown by Enhanced Adsorption of Acid Gases (Sulfur Dioxide and Carbon Dioxide) and Pressure Dependence,” Chem. Mater., 8:1904–1912 (1996). Chen, M.S., and Goodman, D.W., “The Structure of Catalytically Active Gold on Titania,” Science, 306:252–255 (2004). Santra, A.K., and Goodman, D.W., “Size-Dependent Electronic, Structural and Catalytic Properties of Metal Clusters Supported on Ultrathin Oxide Films,” in: Catalysis and Electrocatalysis at Nanoparticles Surfaces (A. Wieckowski, E.R. Savinova, and C.G.Vayenas, eds.), pp. 281–309, Dekker, New York (2003). Andreeva, D., Tabakova, T., Ilieva, L., Naydenov, A., Mehanjiev, D., and Abrashev, M.V., “Nanosize Gold Catalysts Promoted by Vanadium Oxide

420

23. 24. 25. 26. 27. 28. 29.

30. 31. 32.

33. 34. 35. 36. 37. 38.

Properties Supported on Titania and Zirconia for Complete Benze Oxidation,” Appl. Catal. A, 209:291–300 (2001). Corti, C.W., Holliday, R.J., and Thompson, D.T., “Developing New Industrial Applications for Gold: Gold Nanotechnology,” Gold Bull., 35:111–117 (2002). Haruta, M., “Gold as a Novel Catalyst in the 21st Century: Preparation, Working Mechanism and Applications,” Gold Bull. 37:27–36 (2004). Ying, J.Y., “Surface Interaction and Interfacial Chemistry in Multicomponent Nanocrystalline Materials,” Ceram. Trans., 44:67–76 (1995). Chernavskii, P.A., and Lunin, V.V., “Oxide-Oxide Interaction in Ni, Co and Fe Supported Catalyst,” Kinetics Catal., 34:470–477 (1993). Prairie, M.R., Renken, A., Highfield, J.G., Thampi, K.R., and Grätzel, M., “A Fourier Transform Infrared Spectroscopic Study of CO2 Methanation on Supported Ruthenium,” J. Catal., 129:130–144 (1991). Trovarelli, A., Mustazza, C., Dolcetti, G., Kaspar, J., and Graziani, M., “Carbon Dioxide Hydrogenation on Rhodium Supported on Transition Metal Oxides,” Appl. Catal., 65:129–142 (1990). Lalande, G., Tamizhmani, G., Côté, R., Dignard-Bailey, L., Trudeau, M.L., Schulz, R., Guay, D., and Dodelet, J.P., “Influence of Loading on the Activity and Stability of Heat-Treated Carbon-Supported Cobalt Phthalocyanine Electrocatalyst in Solid Polymer Electrolyte Fuel Cells,” J. Electrochem. Soc., 142:1162–1168 (1995). Yao, M.H., Smith, D.J., Kalakkad, D., and Datye, A.K., “HREM Study of Structure of Supported Pt Catalyst,” Z. Phys. D, 26:S79–S81 (1993). Zafiris, G.S., and Gorte, R.J., “A Study of CO, No, and H2 Adsorption on Model Pt/CeO2 Catalysts,” Surf. Sci., 276:86–94 (1992). Fukushima, K., Takaoka, G.H., Matsuo, J., and Yamada, I., “Effects on CO Oxidation Activity of Nano-Scale Au Islands and TiO2 Support Prepared by the Ionized Cluster Beam Method,” Jpn. J. Appl. Phys., 36:813–818 (1997). Ochoa-Fernandez, E., Chen, D., Yu, Z., Totdal, B., Ronning, M., Holmen, A., “Effect of Carbon Nanofiber-Induced Microstrain on the Catalytic Activity of Ni Cyrstals,” Surf. Sci., 554:L107–L112 (2004). Yang, L., Raeker, T.J., and DePristo, A.E., “Surface segregation in bimetallic clusters: predictions using a molecular dynamics/Monte Carlo corrected effective medium theory,” Surf. Sci., 290:195–205 (1993). Schoeb, A.M., Raeker, T.J., Yang, L., Wu, X., King, T.S., and DePristo, A.E., “Driving Force for Surface Segregation in Bimetallic Catalysts,” Surf. Sci., 278:L125–129 (1992). Gijzeman, O.L.J., “Surface Segregation in Small Supported Particles,” Appl. Surf. Sci., 64:9–20 (1993). Deivaraj, T.C., and Lee, J.Y., “Preparation of Carbon-Supported PtRu Nanoparticles for Direct Methanol Fuel Cell Applications—a Comparative Study,” J. Power Sources, 142:43–49 (2005). Ying, J.Y., “Synthesis of Nanostructured Catalysts and Catalytic Supports,” in: Nanophase Materials Synthesis–Properties–Applications (G.C. Hadjipanayis, and R.W. Siegel, eds.), 260:37–44, NATO ASI Series, Kluwer Academic, Dordrecht (1994).

9: Gas Reactive Applications, Trudeau

421

39. Beck, D.D., and Siegel, R.W., “The Dissociative Adsorption of Hydrogen Sulfide Over Nanophase Titanium Dioxide,” J. Mater. Res., 7:2840–2845 (1992). 40. Sarkas, H.W., Arnold, S.T., Hendricks, J.H., Kidder, L.H., Jones, C.A., and Bowen, K.H., “An Investigation of Catalytic Activity in Mixed Oxide Nanophase Materials,” Z. Phys. D, 26:46–50 (1993). 41. Tschöpe, A.S., and Ying, J.Y., “Synthesis of Nanostructured Catalytic Materials Using a Modified Magnetron Sputtering Technique,” Nanostruct. Mater., 4:617–623 (1994). 42. Ying, J.Y., “Surface Structure of Nanocrystalline Oxides,” in: Nanophase Materials Synthesis–Properties–Applications (G.C. Hadjipanayis, and R.W. Siegel, eds.), 260:197–204, NATO ASI Series, Kluwer Academic, Dordrecht (1994). 43. Tschöpe, A.S., Ying, J.Y., Amonlirdviman, K., and Trudeau, M.L., “Surface Chemistry of Nanocrystalline Cerium Oxide,” Mater. Res. Soc. Symp. Proc., 351:251–256 (1994). 44. Trudeau, M.L., Tschöpe, A.S., and Ying, J.Y., “XPS Investigation of Surface Oxidation and Reduction in Nanocrystalline CexLa1–xO2–y,” Surf. Interface Anal., 23:219–226 (1995). 45. Tschöpe, A.S., Ying, J.Y., Liu, W., and Flytzani-Stephanopoulos, M., “Nanocrystalline Doped Cerium Oxide as a Catalyst for SO2 Reduction by CO,” Mater. Res. Soc. Symp. Proc., 344:133–138 (1994). 46. Liu, W., Sarofim, A.F., and Flytzani-Stephanopoulos, M., “Reduction of Sulfur Doixide by Carbon Monoxide to Elemental Sulfur over Composite Oxide Catalysts,” Appl. Catal. B, 4:167 (1994). 47. Tschöpe, A.S., Liu, W., Flytzani-Stephanopoulos, M., and Ying, J.Y., “Redox Activity of Nonstoichiometric Cerium Oxide-Based Nanocrystalline Catalysts,” J. Catal., 157:42–50 (1995). 48. Miani, F., Matteazzi, P., Dolcetti, G., Lutman A., and Trovarelli, A., “Catalytic Properties of Mechanosynthesized Nanophase Iron Carbides,” Mater. Sci. Eng. A, 168:153 (1993). 49. Ying, J.Y., Tschöpe, A.S., and Levin, D., “Synergistic Effects and Catalytic Properties Tailored by Nanostructure Processing,” Nanostruct. Mater., 6:237–246 (1995). 50. Tschope, A.S., Schaadt, D., Birringer, R., and Ying, J.Y., “Catalytic Properties of Nanostructured Metal Oxides Synthesized by Inert Gas Condensation,” Nanostruct. Mater., 9:423–432 (1997). 51. Mile, B., Howard, J.A., Tomietto, M., Joly, H.A., and Sayari, A., “Preparation of Small Copper Particles of High Catalytic Activity Using a Rotating Cryostat,” J. Mater. Sci., 31:3073–3080 (1996). 52. Trudeau, M.L., Huot, J.Y., Schulz, R., Dussault, D., Van Neste, A., and L’Espérance, G., “Nanocrystalline Fe—(Co, Ni)—Si—B: The Mechanical Crystallization of Amorphous Alloys and the Effects on Electrocatalytic Reactions,” Phys. Rev. B, 45:4626–4636 (1992). 53. Huot, J.Y., Trudeau, M.L., and Schulz, R., unpublished. 54. Benameur, T., Yavari, R., and Durand, R., “Electrocatalytic Behavior of Non-Equilibrium bcc Nickel,” Mater. Sci. Eng., A181/A182:1145–1149 (1994).

422

Properties

55. Ivanov, E.Y., Golybkova, G.V., and Grigorieva, T.F., “The CsCl Structure of the Raney Nickel Catalyst Obtained from Mechanically Alloyed Nickel Aluminides,” Reactivity Sol., 8:73–76 (1990). 56. Ivanov, E.Y., Makhlouf, S.A., Sumiyama, K., Yamauchi, H., Suzuki, K., and Golubkuva, G., “Structural and Magnetic Properties of Non-equilibrium b.c.c. Nickel Prepared by Leaching of Mechanically Alloyed Ni35Al65,” J. Alloys Compounds, 185:25 (1992). 57. Brown, D.E., Mahmood, M.N., Turner, A.K., Hall, S.M., and Fogarty, P.O., “Low Overvoltage Electrocatalysts for Hydrogen Evolving Electrodes,” Int. J. Hydrogen Energy, 7:405 (1982). 58. Brown, D.E., Mahmood, M.N., Man, M.C. M., and Turner, A.K., “Preparation and Characterization of Low-Overvoltage Transition Metal Alloy Electrocatalysts for Hydrogen Evolution in Alkaline Solutions,” Electrochim. Acta, 29:1551 (1984). 59. Trudeau, M.L., Huot, J.Y., and Schulz, R., “Mechanically Alloyed Nanocrystalline Ni—Mo Powders: a New Technique for Producing Active Electrodes for Catalyst,” Appl. Phys. Lett., 58:2764–2766 (1991). 60. Huot, J.Y., Trudeau, M.L., and Schulz, R., “Low Hydrogen Overpotential Nanocrystalline Ni—Mo Cathodes for Alkaline Water Electrolysis,” J. Electrochem. Soc., 138:1316–1321 (1991). 61. Schulz, R., Huot, J.Y., Trudeau, M.L., Dignard-Bailey, L., Yan, Z.H., Jin, S., Lamarre, A., Ghali, E., and Van Neste, A., “Nanocrystalline Ni—Mo Alloys and their Applications in Electrocatalysis,” J. Mater. Res., 9:2998–3008 (1994). 62. Van Neste, A., Yip, S.H., Jin, S., Boily, S., Ghali, E., Guay, D., and Schulz, R., “Low Overpotential Nanocrystalline Ti—Fe—Ru—O Cathodes for the Production of Sodium Chlorate,” Mater. Sci. Forum, 225–227:795–800 (1996). 63. Blouin, M., Guay, D., Boily, S., Van Neste, A., and Schulz, R., “Electrocatalytic Properties of Nanocrystalline Alloys: Effect of the Oxygen Concentration in Ti2RuFeOx Alloy on the Structural and Electrochemical Properties,” Mater. Sci. Forum, 225–227:801–806 (1996). 64. Chui, Z.J., Wang, L., Wang, K.Y., Sun, L., Qiao, G.W., and Wang, J.T., “The Surface Structure and Catalytic Properties of Partially Crystallized Fe—Si—B Alloys,” J. Non-Cryst. Solids, 150:440–443 (1992). 65. Chui, Zh.J., Wang, L., Wang, K.Y., Song, Q.H., Wang, J.T., and Lu, K., “The Catalytic Properties of Fe78Si12B10 Alloys Prepared by Ball Milling,” Mater. Sci. Eng. A, 134:1037–1040 (1991). 66. Trovarelli, A., Matteazzi, P., Dolcetti, G., Lutman, A., and Miani, F., “Nanophase Iron Carbides as Catalyst for Carbon Dioxide Hydrogenation,” Appl. Catal., 95:L9–L13 (1993). 67. Miani, F., Matteazzi, P., Dolcetti, G., Lutman, A., and Trovarelli, A., “Catalytic Properties of Mechanosynthesized Nanophase Iron Carbides,” Mater. Sci. Eng. A, 168:153–155 (1993). 68. Van Neste, A., Zielinski, P.A., Kaliaguine, S., Trudeau, M.L., and Schulz, R., “Metal Dispersion Enhancement by Ball-Milling with Alumina,” in: The Proceeding of the Powder Metallurgy Conference, Paris, France (1994).

9: Gas Reactive Applications, Trudeau

423

69. Zielinski, P.A., Schulz, R., Kaliaguine, S., and Van Neste, A., “Structural Transformation of Alumina by High Energy Ball Milling,” J. Mater. Res., 8:2985–2992 (1993). 70. Sun, L., Liu, H., and Dou, S., “The Effect of Ball Milling and Rapid Quenching on the Catalytic Activity of the Cu30Al70 Alloy,” Mater. Lett., 33:79–83 (1997). 71. De Leitenburg, C., Trovarelli, A., Zamar, F., Maschio, S., Dolcetti, G., and Liorca, J., “A Novel Simple Route to Catalyst with a High Oxygen Storage Capacity: The Direct Room-Temperature Synthesis of CeO2—ZrO2 Solid Solutions,” J. Cem. Soc., Chem. Commun., 1995:2181 (1995). 72. Suslick, K.S., Hyeon, T., Fang, M., and Cichowlas, A.A., “Sonochemical Synthesis of Nanostructured Catalyst,” Mater. Sci. Eng. A, 204:186–192 (1995). 73. Suslick, K.S., Hyeon, T., and Fang, M., “Nanostructured Materials Generated by High-Intensity Ultrasound: Sonochemical Synthesis and Catalytic Studies,” Chem. Mater., 8:2172–2179 (1996). 74. Takatani, H., Kago, H., Nakanishi, M., Kobayashi, Y., Hori, F., and Oshima, R., “Characterization of Noble Metal Alloy Nanoparticles Prepared by Ultrasound Irradiation,” Rev. Adv. Mater. Sci., 5:232–238 (2003). 75. Tsai, K.L., and Dye, J.L., “Nanoscale Metal Particles by Homogeneous Reduction with Alkalides or Electrides,” J. Am. Chem. Soc., 113:1650– 1652 (1991). 76. Wang, Z.L., Ahmad, T.S., and El-Sayed, M.A., “Steps, Ledges, and Kinks on the Surfaces of Platinum Nanoparticles of Different Shapes,” Surf. Sci., 380:302–310 (1997). 77. Cao, D., and Bergens, S.H., “A Nonelectrochemical Reductive Deposition of Ruthenium Adatoms onto Nanoparticle Platinum: Anode Catalysts for a Series of Direct Methanol Fuel Cells,” Electrochem. Acta, 48:4021–4031 (2003). 78. Sze, C., Gulari, E., and Demczyk, B.G., “Surface Structure—Catalytic Function in Nanophase Gold Catalyst,” Mater. Res. Soc. Symp. Proc., 286:143–148 (1993). 79. Ozawa, M., Kimura, M., and Isogai, A., “The Application of Ce—Zr Oxide Solid Solution to Oxygen Storage Promoters in Automotive Catalyst,” J. Alloys Compounds, 193:73–75 (1993). 80. Villullas, H.M., Mattos-Costa, F.I., Nascente, P.A.P., and Bulhoes, L.O.S., “Anodic Oxidation of Formaldehyde on Pt-Modified SnO2 Thin Film Electrodes Prepared by a Sol-Gel Method,” Electrochem. Acta, 49:3909–3916 (2004). 81. Maye, M.M., Luo, J., Han, L., Kariuki, N.N., and Zhong, C.J., “Synthesis, Processing, Assembly and Activation of Core-Shell Structured Gold Nanoparticle Catalyst,” Gold Bull., 36:75–82 (2003). 82. Nagaveni, K., Gayen, A., Subbanna, G.N., and Hegde, M.S., “Pd-Coated Ni Nanoparticles by the Polyol Method: an Efficient Hydrogenation Catalyst,” J. Mater. Chem., 12:3147–3151 (2002). 83. Reetz, M.T., and Helbig, W., “Size-Selective Synthesis of Nanostructured Transition Metal Clusters,” J. Am. Chem. Soc., 116:7401–7402 (1994).

424

Properties

84. Kost, K.M., Bartak, D.E., Kazee, B., and Kuwana, T., “Electrodeposition of Palladium, Iridium, Ruthenium, and Platinum in Poly(4-vinylpyridine) Films for Electrocatalysis,” Anal. Chem., 62:151–157 (1990). 85. Ladouceur, M., Lalande, G., Guay, D., Dodelet, J.P., Dignard-Bailey, L., Trudeau, M.L., and Schulz, R., “Pyrolyzed Cobalt Phthalocyanine as Electrocatalyst for Oxygen Reduction,” J. Electrochem. Soc., 140:1974–1981 (1993). 86. Dignard-Bailey, L., Trudeau, M.L., Joly, A., Schulz, R., Lalande, G., Guay, D., and Dodelet, J.P., “Graphitization and Particle Size Analysis of Pyrolyzed Cobalt Phthalocyanine/Carbon Catalyst for Oxygen Reduction in Fuel Cells,” J. Mater. Res., 9:3203–3209 (1994). 87. Park, J., Kang, E., Son, S.U., Park, H.M., Lee, M.K., Kim, J., Kim, K.W., Noh, H.J., Park, J.H., Bae, C.J., Park, J.G., and Hyeon, T., “Monodisperse Nanoparticles of Ni and NiO: Synthesis, Characterization, Self-Assembled Superlattices, and Catalytic Applications in the Suzuki Coupling Reaction,” Adv. Mater. 17:429–434 (2005). 88. Boxall, D.L., Deluga, G.A., Kenik, E.A., King, W.D., and Lukehart, C.M., “Rapid Synthesis of a PtRu/Carbon Nanocomposite Using Microwave Irradiation: A DMFC Anode Catalyst of High Relative Performance,” Chem. Mater., 13:891–900 (2001). 89. Miquel, P.F., and Katz, J.L., “Formation and Characterization of Nanostructured V-P-O Particles in Flames: A New Route for the Formation of Catalysts,” J. Mater. Res., 9:746–754 (1994). 90. Rousset, J.L., Khanra, B.C., Cadrot, A.M., Cadete Santos Aires, F.J., Renouprez, A.J., and Pellarin, M., “Investigations on Supported Bimetallic PdPt Nanostructures,” Surf. Sci., 352–354:583–587 (1996). 91. Tsai, S.H., Liu, Y.H., Wu, P.L., and Yeh, C.S., “Preparation of Au—Ag— Pd Trimetallic Nanoparticles and their Application as Catalysts,” J. Mater. Chem., 13:978–980 (2003). 92. Yamashita, H., Sakai, K., Funabiki, T., Yoshida, S., and Isozumi, Y., “Catalysis by Amorphous Metal Alloys,” J. Chem. Soc. Faraday Trans., I83:2895–2900 (1987). 93. Peuckert, M., and Baiker, A., “Hydrogenation of Carbon Monoxide on Amorphous Fe81B13.5Si3.5C, J. Chem. Soc. Faraday Trans., I81:2797–2803 (1985). 94. Brunelli, K., Dabala, M., Frattini, R., and Magrini, M., “Structural Characterization and Electrocatalytic Properties of Au30Zr70 Amorphous Alloy Obtained by Rapid Quenching,” J. Appl. Electrochem., 33:995–1000 (2003). 95. van de Water, L.G.A., and Maschmeyer, T., “Mesoporous Membranes–a Brief Overview of Recent Developments,” Topics in Catalysis, 29:67–77 (2004). 96. Antonelli, D.M., and Ying, J.Y., “Synthesis of Stable Hexagonally-Packed Mesoporous Niobium Oxide Molecular Sieves Through a Novel LigandAssisted Templating Mechanism,” Angew. Chem. Int. Ed. Engl., 25:426 (1996). 97. Skadtchenko, B.O., Trudeau, M., Kwon, C.-W., Dunn, B., and Antonelli, D., “Synthesis and Electrochemistry of Li- and Na-Fulleride Doped Mesoporous Ta Oxides,” Chem. Mater. B, 16:2886–2894 (2004).

9: Gas Reactive Applications, Trudeau

425

98. Carreon, M.A., and Guliants, V.V., “Ordered Meso- and Macroporous Binary and Mixed Metal Oxides,” Eur. J. Inorg. Chem. 2005:27–43 (2005). 99. Vettraino, M., He, X., Trudeau, M., Drake, J.E., and Antonelli, D.M., “Synthesis of a Stable Niobium Oxide Molecular Sieve and Subsequent Room Temperature Activation of Dinitrogen,” Adv. Funct. Mater., 3:174–178 (2002). 100. Yue, C., Trudeau, M.L., and Antonelli, D., “Mesoporous Tantalum Oxide Photocatalysts for Schrauzer-Type Conversion of Dinitrogen to Ammonia,” Can. J. Chem, 83:308–314 (2005). 101. Li, D., Zhou, H., and Honma, I., “Design and Synthesis of Self-ordered Mesoporous Nanocomposite through Controlled in-situ Crystallization,” Nature Mater., 3:65–72 (2004). 102. Ichikawa, M, “ ‘Ship-in-Bottle’ Catalyst Technology Novel: Novel Templating Fabrication of Platinum Group Metals Nanoparticles and Wires in Micro/Mesopores,” Platinum Metals Rev., 44:3–14 (2000). 103. Grigoriants, I., Sominski, L., Li, H., Ifargan, I., Aurbach, D., and Gedanken, A., “The Use of Tin-Decorated Mesoporous Carbon as an Anode Material for Rechargeable Lithium Batteries,” Chem. Commun., 921–923 (2005). 104. Lordi, V., Yao, N., and Wei, J., “Method for Supporting Platinum on SingleWalled Carbon Nanotubes for a Selective Hydrogenation Catalyst,” Chem. Mater., 13:733–737 (2001). 105. Yoshitake, T., Shimakawa, Y., Kuroshima, S., Kimura, H., Ichihashi, T., Kubo, Y., Kasuya, D., Takahashi, K., Kokai, F., Yudasaka, M., and Iijima, S., “Preparation of Fine Platinum Catalyst Supported on Single-Wall Carbon Nanohorns for Fuel Cell Application,” Physica B, 323:124–126 (2002). 106. Mu, P., Haolin, T., Shichun, M., and Runzhang, Y., “Synthesis of Platinum/Multi-Wall Carbon Nanotube Catalysts,” J. Mater. Res., 19:2279– 2284 (2004). 107. Carmo, M., Paganin, V.A., Rosolen, J.M., and Gonzalez, E.R., “Alternative Supports for the Preparation of Catalysts for Low-Temperature Fuel Cells: the Use of Carbon Nanotubes,” J. Power Sources, 142:169–176 (2005). 108. Zhao, X., Li, W., Jiang, L., Zhou, W., Xin, Q., Yi, B., and Sun, G., “MultiWall Carbon Nanotube Supported Pt–Sn Nanoparticles as an Anode Catalyst for the Direct Ethanol Fuel Cell,” Carbon, 42:3263–3265 (2004). 109. Jacobs, P.W., Wind, S.J., Ribeiro, F.H., and Somorjai, G.A., “Nanometer Size Platinum Particle Arrays: Catalytic and Surface Chemical Properties,” Surf. Sci., 372:L249–L253 (1997). 110. Somorjai, G.A., “New Model Catalysts (Platinum Nanoparticles) and New Techniques (SFG and STM) for Studies of Reaction Intermediates and Surface Restructuring at High Pressure During Catalytic Reactions,” Appl. Surf. Sci., 121/122:1–19 (1997). 111. Berko, A., Szoko, J., and Solymosi, F., “Preparation and Reactivity of Rh Nanoparticles on TiO2 (110)-(1x2) Surface,” Solid Stae Ionics, 141/142:197–202 (2001). 112. Tomchenko, A.A., Harmer, G.P., and Marquis, B.T., “Detection of Chemical Warfare Agents Using Nanostructured Metal Oxide Sensors,” Sensors & Actuators B, 108:41–55 (2005).

426

Properties

113. Yamazoe, N., and Miura, N., “Environmental Gas Sensing,” Sensors and Actuators B, 20:95–102 (1994). 114. Yamazoe, N., “Toward Innovations of Gas Sensor Technology,” Sensors & Actuators B, 108:2–14 (2005). 115. Kugishima, M., Sakai, G., Shimanoe, K., and Yamazoe, N., “Development of SnO2-Based Gas Sensor Sensitive to Dilute Ethylene Oxide in Air,” Sensors & Actuators B, 108:130–133 (2005). 116. Ivanov, P., Llobet, E., Vilanova, X., Hubalek, J., Gracia, I., Cané, C., and Correig, X., “Gas Sensors Micro-array for Air Quality Monitoring Based on Pure and Doped SnO2 Thick Sensing Films,” 2005 Spanish Conference on Electron Devices, 545–548 (2005). 117. Tolles, W.M., “Nanoscience and Nanotechnology in Europe,” Nanotechnology, 7:59 (1996). 118. Korolkoff, N.O., “Survey of Toxic Gas Sensors and Monitoring Systems,” Solid State Techn., 49–64 (Dec. 1989). 119. Sbverveglieri, G. (ed.), Gas Sensors Principles, Operation and Developments, Kluwer Academic, Groningen (1992). 120. Rickerby, D.G., Horrillo, M.C., Santos, J.P., and Serrini, P., “Microstructural Characterization of Nanograin Tin Oxide Gas Sensors,” Nanostruct. Mater., 9:43–52 (1997). 121. Cantalini, C., Post, M., Buso, D., Guglielmi, M., and Martucci, A., “Gas Sensing Properties of Nanocrystalline NiO and Co3O4 in Porous Silica SolGel Films,” Sensors & Actuators B, 108:184–192 (2005). 122. Snow, E.S., Perkins, F.K., Houser, E.J., Badescu, S.C., and Reinecke, T.L., “Chemical Detection with Single-walled Carbon Nanotube Capacitor,” Science, 307:1942–1945 (2005). 123. Ogawa, H., Abe, A., Nishikawa, M., and Hayakawa, S., “Preparation of Tin Oxide Films from Ultrafine Particles,” J. Electrochem. Soc., 128:685–689 (1981). 124. Ogawa, H., Abe, A., Nishikawa, M., and Hayakawa, S., “Electrical Properties of Tin Oxide Ultrafine Particles Films,” J. Electrochem. Soc., 128:2020–2025 (1981). 125. Ogawa, H., Nishikawa, M., and Abe, A., “Hall Measurement Studies and an Electrical Conduction Model of Tin Oxide Ultrafine Particle Films,” J. Appl. Phys., 53:4448–4455 (1982). 126. Guidi, V., Comini, E., Faglia, G., and Ferroni, M., “Mixed Nanostructured Oxides of Ti and VI-B-Group Elements for Gas Sensing,” Sensors, 2002. Proceedings of IEEE, 1:398–403 (2002). 127. Cui, Y., Wei, Q., Park, H., and Lieber, C., “Nanowire Nanosensors for Highly Sensitive and Selective Detection of Biological and Chemical Species,” Science, 293:1289–1292 (2001). 128. Seiyama, T., Kato, A., Fujiishi, K., and Nagatani, M., “A New Detector for Gaseous Components Using Semiconductive Thin Films,” Anal. Chem., 34:1502–1053 (1962). 129. Kohl, D., “Surface Processes in the Detection of Reducing Gases with SnO2-Based Device,” Sensors and Actuators, 18:71–113 (1989). 130. Watson, J., Ihokura, K., and Coles, G.S.V., “The Tin Dioxide Gas Sensor,” Meas. Sci. Technol., 4:711–719 (1993).

9: Gas Reactive Applications, Trudeau

427

131. Gopel, W., and Schierbaum, K.D., “SnO2 Sensors: Current Status and Future Prospects,” Sensors and Actuators B, 26–27:1–12 (1995). 132. Yamazoe, N., Fuchigami, J., Kishikawa, M., and Seiyama, T., “Interactions of Tin Oxide Surface with O2, H2O, and H2,” Surf. Sci., 86:335–344 (1979). 133. Sun, H.-T., Cantalini, C., Faccio, M., and Pelino, M., “NO2 Gas Sensitivity of Sol-Gel-Derived α–Fe2O3 Thin Films,” Thin Solid Films, 269:97–101 (1995). 134. Yea, B., Konishi, R., Osaki, T., Abe, S., Tanioka, H., and Sugahara, K., “Analysis of the Sensing Mechanism of Tin Dioxide Thin Film Gas Sensors Using the Change of Work Function in Flammable Gas Atmosphere,” Appl. Surf. Sci., 100/101:365 (1996). 135. Chang, S.C., “Oxygen Chemisorption on Tin Oxide: Correlation Between Electrical Conductivity and EPR Measurements,” J. Vac. Sci. Technol., 17:366–369 (1980). 136. Kanamori, M., Suzuki, K., Ohya, Y., and Takahashi Y., “Analysis of the Changes in the Carrier Concentration of SnO2 Thin Film Gas Sensor,” Jpn. J. Appl. Phys., 33:6680–6683 (1994). 137. Pink, H., Treitinger, L., and Vite, L., “Preparation of Fast Detecting SnO2 Gas Sensors,” Jpn. J. Appl. Phys., 19:513–517 (1980). 138. Sberveglieri, G., “Classical and Novel Techniques for the Preparation of SnO2 Thin-Film Gas Sensor,” Sensors and Actuators B, 6:239 (1992). 139. Butta, N., Mell, M., and Pizzini, S., “Influence of Surface Parameters and Doping on the Sensitivity and on the Response Times of Tin Oxide Resistive Sensors,” Sensors and Actuators B, 2:151–161 (1990). 140. Windischmann, H., and Mark, P., “A Model for the Operation of a ThinFilm SnOx Conductance-Modulation Carbon Monoxide Sensor,” J. Electrochem. Soc., 126:627–633 (1979). 141. Dieguez, A., Romano-Rodriguez, A., Morante, J.R., Weimar, U., SchweizerBerberich, M., and Gopel, W., “Morphological Analysis of Nanocrystalline SnO2 for Gas Sensor Applications,” Sensors and Actuators B, 31:1–8 (1996). 142. Xu, C., Tamaki, J., Miura, N., and Yamazoe, N., “Grain Size Effects on Gas Sensitivity of Porous SnO2-Based Elements,” Sensors and Actuators B, 3:147–155 (1991). 143. Chang, S.C., “Thin-Film Semiconductor NOx Sensor,” IEEE Trans. Electron. Devices, ED-26:1875–1880 (1979). 144. Gouma, P.I., “Nanostructured Polymorphic Oxides for Advanced Chemosensors,” Rev. Adv. Mater. Sci., 5:147–154 (2003). 145. Kohl, D., “The Role of Noble Metals in the Chemistry of Solid-State Gas Sensor,” Sensors and Actuators B, 1:158–165 (1990). 146. Morrison, S.R., “Selectivity in Semiconductor Gas Sensors,” Sensors and Actuators, 12:425–440 (1987). 147. Shaver, P.J., “Activated Tungsten Oxide Gas Detectors,” Appl. Phys. Lett., 11:255–257 (1967). 148. Chang, S.C., “The Effects of Palladium-Gold on the Chemisorption Activities of Thin Film Tin Oxide,” J. Vac. Sci. Technol. A, 1:296–300 (1983). 149. Cricenti, A., Generosi, R., Scarselli, M.A., Perfetti, P., Siciliano, P., Serra, A., Tepore, A., Coluzza, C., Almeida, J., and Margaritondo, G., “Morpho-

428

150. 151. 152. 153. 154.

155.

156.

157.

158. 159.

160. 161.

162.

Properties logical, Chemical and Electrical Characterization of Pt–SnO2 Thin Film Grown on Rough and Mechanically Polished Al2O3 Substrates,” J. Phys. D, 29:2235–2239 (1996). Vlachos, D.S., Papadopoulos, C.A., and Avaritsiotis, J.N., “On the Electronic Interaction Between Additives and Semiconducting Oxide Gas Sensors,” Appl. Phys. Lett., 69:650–652 (1996). Zhang, T., Shen, Y., Zhang, R., and Liu, X., “Ammonia-Sensing Characteristics of Pt-Doped CdSnO3 Semiconducting Ceramic Sensor,” Mater. Lett., 27:161–164 (1996). Cao, X., Cao, L., Yao, W., and Ye, X., “Structural Characterization of PdDoped SnO2 Thin Films Using XPS,” Surf. Interface Anal., 24:662–666 (1996). Matsushima, S., Meakawa, T., Tamaki, J., Miura, N., and Yamazoe N., “Role of Additives on Alcohol Sensing by Semiconductor Gas Sensor,” Chem. Lett., 18:845–848 (1989). Sysoev, V., Kisin, V., Frietsch, Kiselev, I., Habicht, W., Bruns, M., and Goschnick, J., “The Characterization of Nanostructured Copper-Doped Tin Oxide Films for Gas Sensor Microarrays,” Sensors 2004. Proceedings of IEEE, 1:130–133 (2004). Rastomjee, C.S., Dale, R.S., Schaffer, R.J., Jones, F.H., Egdell, R.G., Georgiadis, G.C., Lee, M.J., Tate, T.J., and Cao, L.L., “An Investigation of Doping SnO2 by Ion Implantation and Application of Ion-Implanted Films as Gas Sensors,” Thin Solid Films, 279:98–105 (1996). Barbi, G.B., Santos, J.P., Serrini, P., Gibson, P.N., Horrillo, M.C., and Manes, L., “Ultrafine Grain-Size Tin-Oxide Films for Carbon Monoxide Monitoring in Urban Environments,” Sensors and Actuators B, 24–25:559–563 (1995). Stankova, M., Calderer, J., Vilanova, X., Llobet, E., Gracia, I., Cané, C., and Correig, X., “Improvement of the Gas Sensing Properties of RF Sputtered WO3 Thin Films Using Different Dopant,” Proc. 2005 Spanish Conference on Electron Devices, 553–556 (2005). Rumyantseva, M.N., Labeau, M., Senateur, J.P., Delabouglise, G., Boulova, M.N., and Gaskov, A.M., “Influence of Copper on Sensor Properties of Tin Dioxide Films in H2S,” Mater. Sci. Eng. B, 41:228–234 (1996). Zubkans, J., Lloyd Spetz, A., Sundgren, H., Winquist, F., Kleperis, J., Lusis A., and Lundstrom, I., “In-Situ Modification of the NOx Sensitivity of Thin Discontinuous Platinum Films as Gates of Chemical Sensors,” Thin Solid Films, 268:140–143 (1995). Cukrov, L.M., Tsuzuki, T., and McCormik, P.G., “SnO2 Nanoparticles Prepared by Mechanochemical Processing,” Scripta. Mater., 44:1787–1790 (2001). Jiang, J.Z., Lin, R., Lin, W., Nielsen, K., Morup, S., Dam-Johansen K., and Clasen, R., “Gas-Sensitive Properties and Structure of Nanostructured (αFe2O3)x–(SnO2)1–x Materials Prepared by Mechanical Alloying,” J. Phys. D, 30:1459–1467 (1997). Ansari, S.G., Gosavi, S.W., Gangal, S.A., Karekar, R.N., and Aiyer, R.C., “Characterization of SnO2-Based H2 Gas Sensors Fabricated by Different Deposition Techniques,” J. Mater. Sci., 8:23–27 (1997).

9: Gas Reactive Applications, Trudeau

429

163. Vlachos, D.S., Papadopoulos, C.A., and Avaritsiotis, J.N., “Dependence of Sensitivity of SnOx Thin-Film Gas Sensors on Vacancy Defects,” J. Appl. Phys., 80:6050–6054 (1996). 164. Yoo, K.S., Park, S.H., and Kang, J.H., “Nano-Grained Thin-Film Indium Tin Oxide Gas Sensors for H2 Detection, Sensors & Actuators B, 108:159–164 (2005). 165. Serrini, P., Briois, V., Horrillo, M.C., Traverse, A., and Manes, L., “Chemical Composition and Crystalline Structure of SnO2 Thin Films Used as Gas Sensors,” Thin Solid Films, 304:113–122 (1997). 166. Hu, W.S., Liu, Z.G., Wu, Z.C., and Feng, D., “Comparative Study of Laser Ablation Techniques for Fabricating Nanocrystalline SnO2 Thin Films for Sensors,” Mater. Lett., 28:369–372 (1996). 167. Liu, Y., Zhu, W., Tan, O.K., and Shen, Y., “Structural and Gas Sensing Properties of Ultrafine Fe2O3 Prepared by Plasma Enhanced Chemical Vapor Deposition,” Mater. Sci. Eng. B, 47:171–176 (1997). 168. Hu, Y., Tan, O.K., Pan, J.S., Huang, H., and Cao, W., “The Effects of Annealing Temperature on the Sensing Properties of Low Temperature Nano-Sized SrTiO3 Oxygen Gas Sensor,” Sensors & Actuators B, 108:244–249 (2005). 169. Carney, C.M., Yoo, S., and Akbar, S.A., “TiO2-SnO2 Nanostructures and their H2 Sensing Behavior,” Sensors & Actuators, 108:29–33. (2005). 170. Bhowmik, S., Constant, K.P., Parker, J.C., and Ali, M., “Electrical Characterization of Nanocrystalline Titania, I: Impedance Spectroscopy Studies between 300 K and 473 K,” Mater. Sci. Eng. A, 204:258–266 (1995). 171. Jun, Y.K., Kim, H.S., Lee, J.H., and Hong, S.H., “High H2 Sensing Behavior of TiO2 Films Formed by Thermal Oxidation,” Sensors and Actuators B, 107:264–270 (2005). 172. Rella, R., Serra, A., Siciliano, P., Vasanelli, L., De, G., and Licciulli, A., “CO Sensing Properties of SnO2 Thin Films Prepared by the Sol-Gel Process,” Thin Solid Films, 304:339–343 (1997). 173. Xu, C., Tamaki, J., Miura, N., and Yamazoe, N., “Promoting Effects of Additives on Thermal Stability of Tin (IV) Fine Particles,” J. Mater. Sci. Lett., 8:1092–1094 (1989). 174. Xu, C., Tamaki, J., Miura, N., and Yamazoe, N., “Correlation between Gas Sensitivity and Crystallite Size in Porous SnO2-Based Sensors,” Chem. Lett., 441–444 (1990). 175. Akimov, B.A., Albul, A.V., Gas’kov, A.M., Il’in, V. Yu, Rumyantseva, M.N., and Labeau, M., “The Electrical Conductivity of Polycrystalline SnO2(Cu) Films and Their Sensitivity to Hydrogen Sulfide,” Semiconductors, 31:335 (1997). 176. Liu, C.H., Zhang, L., and He, Y.J., “Properties and Mechanism Study of Ag Doped SnO2 Thin Films as H2S Sensors,” Thin Solid Films, 304:13–15 (1997). 177. Neubecker, A., Pompl, T., Doll, T., Hansch, W., and Eisele, I., “OzoneEnhanced Molecular Beam Deposition of Nickel Oxide (NiO) for Sensor Applications,” Thin Solid Films, 310:19–23 (1997).

430

Properties

178. Gualdron, O., Llobet, E., Brezmes, J., Vilanova, X., and Correig, X., “Fast Variable Selection for Gas Sensing Applications,” Sensors, 2004. Proceedings of IEEE, 2:892–895 (2004). 179. Ivanov, P., Stankova, M., Llobet, E., Vilanova, X., Brezmes, J., Gracia, I., Cané, C., Calderer, J., and Cooreig, X., “Nanoparticle Metal-Oxide Films for Micro-Hotplate-Based Gas Sensor Systems,” IEEE Sensors, 5:798–809 (2005). 180. Ivanov, P., Vilaseca, M., Llobet, E., Vilanova, X., Hubalek, J., Coronas, J., Santamaria, J., and Correig, X., “Selective Detection of Ammonia and Benzene via Zeolite Films Deposited on SnO2/Pt–SnO2 Thick Films Gas Sensors,” 2005 Spanish Conference on Electron Devices, 597–601 (2005). 181. Hossein-Babaei, F., and Orvatinia, M., “Transient Regime of a Gas Diffusion-Physisorption Through a Microporous Barrier,” IEEE Sensors J., 5:1004–1010 (2005). 182. Arbiol, J., Cabot, A., Morante, J.R., Chen, F., and Liu, M., “Distribution of Noble Metal Pd and Pt in Mesoporous Silica,” Appl. Phys. Lett., 81:3449–3451 (2002). 183. Di Francia, G., Della Noce, M., La Ferrara V., Laqncellotti, L., Morvillo, P., and Quercia, L., “Nanostructured Porous Silicon for Gas Sensor Applications,” Mater. Sci. Tech., 18:767–771 (2002). 184. Miyazaki, H., Hyodo, T., Shimizu, Y., and Egashira, M., “HydrogenSensing Properties of Anodically Oxidized TiO2 Film Sensors: Effects of Preparation and Pretreatment Conditions,” Sensors & Actuators B, 108:467–472 (2005). 185. Li, G.J., and Kawi, S., “High-Surface-Area SnO2: A Novel SemiconductorOxide Gas Sensor,” Mater. Lett., 34:99–102 (1998). 186. Shimizu, Y., Hyodo, T., and Egashira, M., “Meso- to Macro-Porous Oxides as Semiconductor Gas Sensors,” Catal. Surv. From Asia, 8:127–135 (2004). 187. Shimizu, Y., Jono, A., Hyodo, T., and Egashira, M., “Preparation of Large Mesoporous SnO2 Powder for Gas Sensor Application,” Sensors & Actuators B, 108:56–61 (2005). 188. Kong, J., Franklin, N.R., Zhou, C., Chapline, M.G., Peng, S., Cho, K., and Dai, H., “Nanotube Molecular Wires as Chemical Sensors,” Science, 287:622–625 (2000). 189. Cantalini, C., Valentini, L., Armentano, I., Kenny, J.M., Lozzi, L., and Santucci, S., “Carbon Nanotubes as New Materials for Gas Sensing Applications,” J. Eur. Ceram. Soc., 24:1405–1405 (2004). 190. Valentini, L., Armentano, I., Lozzi, L., Santucci, S., and Kenny, J.M., “Interaction of Methane with Carbon Nanotube Thin Films: Role of Defect and Oxygen Adsorption,” Mater. Sci. Eng. C, 24:527–533 (2004). 191. Nguyen, H.Q., Mai, V.T., and Huh, J.S., “Behavior of Single-Walled Carbon Nanotube-Based Gas Sensor at Various Conditions of Operation, SolidState Sensors, Actuators and Microsystems,” 2:1939–1942 (2005). 192. Picozzi, S., Lozzi, L., Santucci, S., Cantalini, C., Baratto, C., Sberveglieri, G., Armentano, I., Kenny, J.M., Valentini, L., and Delley, B., “Ozone Reactivity with Carbon Nanotubes: Experimental and Theoretical Studies,” Sensors, 2003. Proceedings of IEEE, 1:436–439 (2003).

9: Gas Reactive Applications, Trudeau

431

193. Ionescu, R., Espinosa, E., Bittencourt, C., Felten, A., Pireaux, J.J., and Llobet, E., “Gas Sensing Using CNT-doped WO3,” 2005 IEEE Spanish Conference on Electron Devices, 615–618 (2005). 194. Li, J., and Lu, Y., “Nanostructure Engineered Chemical Sensors for Hazardous Gas and Vapor Detection,” in Nanosensing: Materials and Devices(M.S. Islam, and A.K. Dutta, eds), Proc. SPIE, 5593:222–231 (2004). 195. Mor, G.K., Carvalho, M.A., Varghese, O.K., Pishko, M.V., and Grimes, G.A., “A Room Temperature TiO2-Nanotube Hydrogen Sensor Able to SelfClean Photoactively from Environmental Contamination,” J. Mater. Res., 19:628–634 (2004). 196. Varghese, O., Mor, G.K., Grimes, C.A., Paulose, M., and Mukherjee, N., “A Titania Nanotube-Array Room Temperature Sensor for Selective Detection of Hydrogen at Low Concentrations,” J. Nanosci. Nanotech., 4:733–737 (2004). 197. Yu, M.F., Atashbar, M.Z., and Chen, X., “Mechanical and Electrical Characterization of b-Ga2O3 Nanostructures for Sensing Applications,” IEEE Sensors J., 5:20–25 (2005). 198. Liu, J., Wang, X., Peng, Q., and Li, Y., “Vanadium Pentoxide Nanobelts: Highly Selective and Stable Ethanol Sensor Materials,” Adv. Mater., 17:764–767 (2005). 199. Zhou, X.T., Hu, J.Q., Li, C.P., Ma, D.D.D., Lee, C.S., and Lee, S.T., “Silicon Nanowires as Chemical Sensors,” Chem. Phys. Lett., 369:220–224. (2003). 200. Atashbar, M.Z., Benerji, D., and Singamaneni, S., “Room-Temperature Hydrogen Sensor Based on Paladium Nanowires,” IEEE Sensors J., 5:792–797 (2005). 201. Sandi, G., “Hydrogen Storage and Its Limitations,” The Electrochem. Soc. Interface, Fall 2004:40–44 (2004). 202. http://www.eere.energy.gov/hydrogenandfuelcells/pdfs/1_milliken_ final.pdf. 203. http://www.eere.energy.gov/hydrogenandfuelcells/pdfs/h2_storage_ think_tank.pdf. 204. Yartys, V.A., and Lototsky, “An Overview of Hydrogen Storage Methods,” in: Hydrogen Materials Science and Chemistry of Carbon Nanomaterials (T.N. Veziroglu, S.Y. Zaginaichenko, D.Y. Schur, B. Baranowski, A.P. Shpak, and V.V. Skorokhod, eds.), Mathematics, Physics and Chemistry 172:75–104, Kluwer AcademicDordrecht (2004). 205. Zhou, L., “Progress and Problems in Hydrogen Storage Mathods,” Renew. Sustain. Energy Rev., 9:395–408 (2005). 206. David, E., “An Overview of Advanced Materials for Hydrogen Storage,” J. Mater. Process. Technol., 162/163:169–177 (2005). 207. Sarkar, A., and Benerjee, R., “Net Energy of Hydrogen Storage Option,” Int. J. Hydrogen Energy, 30:867–877 (2005). 208. Topler, J., and Feucht, K., “Results of a Test Fleet with Metal Hydride Motor Cars,” Z. Phy. Chem. Neue Folge, 164:1451–1461 (1989). 209. Zaluski, L., Zaluska, A., Tessier, P., Strom-Olsen, J.O., and Schulz, R., “Nanocrystalline Hydrogen Absorbing Alloys,” Mater. Sci. Forum, 225–227:853–858 (1996).

432

Properties

210. Kuriyama, N., Sakai, T., Miyamura, H., Uehara, I., and Ishikawa, H., “Characterization of Metal Hydride Electrodes by Means of Electrochemical Impedance Spectroscopy,” J. Alloys Compounds, 192:161–163 (1993). 211. Jurczyk, M., Nowak, M., Jankowska, E., and Jakubowicz, “Structure and Electrochemical Properties of the Mechanically Alloyed La(Ni, M) Materials,” J. Alloys Compounds 339:339–343 (2003). 212. Han, S.S., Lee, H.Y., Goo, N.H., Jeong, W.T., and Lee, K.S., “Improvement of Electrode Performances of Mg2Ni by Mechanical Alloying,” J. Alloys Compounds, 330/332:841–845 (2002). 213. Ruggeri, S., Lenain, C., Roué, L., Liang, G., Huot, J., and Schulz, R., “Mechanically Driven Crystallization of Amorphous MgNi Alloy During Prolonged Milling: Applications in Ni–MH Batteries,” J. Alloys Compounds, 339:195–201 (2002). 214. Zhu, M., Wang, Z.M., Peng, C.H., Zeng, M.Q., and Gao, Y., “The Effect of Grain Refining on the Discharge Capacity of Mg2Ni/MmNi5–x(CoAlMn)x Composite Prepared by Mechanical Alloying,” J. Alloys Compounds, 349:284–289 (2003). 215. Meli, F., Zuettel, A., and Schlapbach, L., “Surface and Bulk Properties of LaNi5–xSix Alloys from the Viewpoint of Battery Applications,” J. Alloys Compounds, 190:17–24 (1992). 216. Bernauer, O., Topler, J., Noreus, D., Hempelmann, R., and Richter, D., “Fundamentals and Properties of Some Ti/Mn Based Laves Phase Hydrides,” Int. J. Hydrogen Energy, 14:187–200 (1989). 217. Martin, M., Gommel, C., Borkhart, C., and Fromm, E., “Absorption and Desorption Kinetics of Hydrogen Storage Alloys,” J. Alloys Compounds, 238:193–201 (1996). 218. Kim, J.Y., Gibbons, P.C., and Kelton, K.F., “Hydrogenation of Pd-Coated Samples of the Ti—Zr-Based Icosahedral Phase and Related Crystalline Phases,” J. Alloys Compounds, 266:311–317 (1998). 219. Koch, C.C., “Synthesis of Nanostructured Materials by Mechanical Milling: Problems and Opportunities,” Nanostruct. Mater., 9:13–22. (1997). 220. Murray, J., Miller, H., Bird, P., and Goudy, A.J., “The Effect of Particle Size and Surface Composition on the Reaction Rates of Some Hydrogen Storage Alloys,” J. Alloys Compounds, 231:841–845 (1995). 221. Wronski, Z.S., “Materials for Rechargeable Batteries and Clean Hydrogen Energy Sources,” Int. Mater. Rev., 46:1–49 (2001). 222. Varin, R.A., and Czujko, T., “Overview of Processing of Nanocrystalline Hydrogen Storage Intermetallics by Mechanical Alloying/Milling,” Mater. Manufact. Process., 17:129–156 (2002). 223. Trudeau, M.L., Schulz, R., Zaluski, L., Hosatte, S., Ryan, D.H., Doner, C.B., Tessier, P., Strom-Olsen, J.O., and Van Neste, A., “Nanocrystalline Iron-Titanium Alloys Prepared by High-Energy Mechanical Deformation,” Mater. Sci. Forum, 88–90:537–544 (1992). 224. Zaluski, L., Tessier, P., Ryan, D.H., Doner, C.B., Zaluska, A., Strom-Olsen, J.O., Trudeau, M.L., and Schulz, R., “Amorphous and Nanocrystalline Fe—Ti Prepared by Ball Milling,” J. Mater. Res., 8:3059–3068 (1993). 225. Tessier, P., Zaluski, L., Zaluska, A., Strom-Olsen, J.O., and Schulz, R., “Effect of Compositional Variations on Hydrogen Storage in Ball-Milled Fe—Ti,” Mater. Sci. Forum, 225–227:869–874 (1996).

9: Gas Reactive Applications, Trudeau

433

226. Zaluski, L., Zaluska, A., Tessier, P., Strom-Olsen, J.O., and Schulz, R., “Investigation of Structural Relaxation by Hydrogen Absorption in BallMilled Alloys,” Mater. Sci. Forum, 225–227:875–880 (1996). 227. Trudeau, M.L., Dignard-Bailey, L., Schulz, R., Tessier, P., Zaluski, L., Ryan, D.H., and Strom-Olsen, J.O., “The Oxidation of Nanocrystalline FeTi Hydrogen Storage Compounds,” Nanostruct. Mater., 1:457–464 (1992). 228. Selvam, P., Viswanathan B., and Srinivasan, V., “XPS and XAES Studies on Hydrogen Storage Magnesium-Based Alloys,” Int. J. Hydrogen Energy, 14:899–902 (1989). 229. Zaluski, L., Zaluska, A., Tessier, P., Strom-Olsen, J.O., and Schulz, R., “Hydrogen Absorption by Nanocrystalline and Amorphous Fe—Ti with Palladium Catalyst, Produced by Ball Milling,” J. Mater. Sci., 31:695–698 (1996). 230. Wasz, M.L., and Schwarz, R.B., “Structure and Properties of Metal Hydrides Prepared by Mechanical Alloying,” Mater. Sci. Forum, 225–227:859–868 (1996). 231. Yu, X.B., Wu, Z., Xia, B.J., and Xu, N.X., “Improvement of Activation Performance of the Quenched Ti—V-based BCC Phase Alloys,” J. Alloys Compounds, 386:258–260 (2005). 232. Oelerich, W., Klassen, T., and Bormann, R., “Mg—Based Hydrogen Storage Materials with Improved Hydrogen Sorption,” Mater. Trans., 42:1588–1592 (2001). 233. Zaluski, L., Zaluska, A., and Strom-Olsen, J.O., “Hydrogen Absorption in Nanocrystalline Mg2Ni Formed by Mechanical Alloying,” J. Alloys Compounds, 217:245–249 (1995). 234. Zaluski, L., Zaluska, A., Tessier, P., Strom-Olsen, J.O., and Schulz, R., “Catalytic Effect of Pd on Hydrogen Absorption in Mechanically Alloyed Mg2Ni, LaNi5, and FeTi,” J. Alloys Compounds, 217:295–300 (1997). 235. Janot, R., Rougier, A., Aymard, L., Lenain, C., Herrera-Urbina, R., Nazri, G.A., and Tarascon, J.M., “Enhancement of Hydrogen Storage in MgNi by Pd-coating,” J. Alloys Compounds, 356/357:438–441 (2003). 236. Li, J., Ji, X., Wu, F., and Wang. G., “A New Method for the Production of Mg—Ni Hydrogen Storage Materials,” Adv. Mater., 5:554–555. (1993). 237. Berlouis, L.E.A., Cabrera E., Hall-Barientos, E., Hall, P.J., Dodd, S., Morris S., and Imam, M.A., “A Thermal Analysis Investigation of the Hydriding Properties of Nanocrystalline Mg—Ni Alloys Prepared by High Energy Ball Milling,” J. Alloys Compounds, 305:82–89 (2000). 238. Imamura, H., and Sakasai, N., “Hydriding Characteristics of Mg–Based Composites Prepared Using a Ball Mill,” J. Alloys Compounds 231:810– 814 (1995). 239. Imamura, H., Sakasai, N., and Kajii, Y., “Hydrogen Absorption of Mg–Based Composites Prepared by Mechanical Milling: Factors Affecting its Characteristics,” J. Alloys Compounds, 232:218–223 (1996). 240. Holtz, R.L., and Imam, M.A., “Hydrogen Storage Capacity of Submicron Magnesium-Nickel Alloys,” J. Mater. Sci., 32:2267–2274 (1997).

434

Properties

241. Orimo, S., and Fujii, H., “Hydriding Properties of the Mg2Ni—H System Synthesized by Reactive Mechanical Grinding,” J. Alloys Compounds, 232:L16–L19 (1996). 242. Gross, K.L., Spatz, P., Zuttel, A., and Schlapbach, L., “Mechanically Milled Mg Composites for Hydrogen Storage. The Transition to a Steady State Composition,” J. Alloys Compounds, 240:206–213 (1996). 243. Gross, K.L., Spatz, P., Zuttel, A., and Schlapbach, L., “Mg Composites for Hydrogen Storage. The Dependence of Hydriding Properties on Composition,” J. Alloys Compounds, 261:276–280 (1997). 244. Li, Q., Lin, Q., Chou, K.C., Jiang, L.J., and Zhan F., “Hydrogen Storage Properties of Mechanically Alloyed Mg-8 mol% LaNi0.5 Composite,” J. Mater. Res., 19:2871–2876 (2004). 245. Wang, P., Zhang, H.F., Ding, B.Z., and Hu, Z.Q., “Structural and Hydriding Properties of Composite Mg—Mg—ZrFe1.4Cr0.6,” Acta Mater., 49:921–926 (2001). 246. Zhenxing, Y., Zuyan, L., and Erde, W., “Hydrogen Storage Properties of Nanocomposite Mg—Ni—Cu—CrCl3 Prepared by Mechanical Alloying,” Mater. Sci. Eng. A, 335:43–48 (2002). 247. Klassen, T., Oelerich, W., and Bormann, R., “Nanocrystalline Mg-based Hydrides: Hydrogen Storage for the Zero-Emission Vehicle,” Mater. Sci. Forum, 360/362:603–608 (2001). 248. Barkhordarian, G., Klassen, T., and Bormann, R., “Effect of Nb2O5 Content on Hydrogen Reaction Kinetics of Mg,” J. Alloys Compounds, 364:242–246 (2004). 249. Oelerich, W., Klassen, T., and Bormann, R., “Comparison of the Catalytic Effects of V, V2O5, VN and VC on the Hydrogen Sorption of Nanocrystalline Mg,” J. Alloys Compounds, 322:L5-L9 (2002). 250. Bobet, J.-L., Desmoulins-Krawiec, S., Grigorova, E., Cansell, F., and Chevalier, B., “Addition of Nanosized Cr2O3 to Magnesium for Improvement of the Hydrogen Sorption Properties,” J. Alloys Compounds, 351:217–221 (2003). 251. Wang, E.-d., Yu, Z.-x., and Liu, Z.-y., “Hydrogen Storage Properties of Nanocomposite Mg—Ni—MnO2 Made by Mechanical Milling,” Trans. Nonferrous Met. Soc. China, 12:227–232 (2002). 252. Yavari, A.R., LaMoulec, A., deCastro, F.R., Deledda, S., Friedrichs, O., Botta, W.J., Vaughan, G., Klassen, T., Fernandez, A., and Kvick, A., “Improvement in H-Sorption Kinetics of MgH2 Powders by Using Fe Nanoparticles Generated by Reactive FeF3 Addition,” Scripta Mater., 52:719–724 (2005). 253. Bassetti, A., Bonetti, E., Fiorini, A.L., Grbovic, J., Montone, A., Pasquini, L., and Vittori Antisari, M., “Microstructure and Hydrogen Desorption in Nanostructured MgH2—Fe,” Mater. Sci. Forum, 453/454:205–212. (2004). 254. Huot, J., Bouaricha, S., Boily, S., Dodelet, J.-P., Guay, D., and Schulz., R., “Increase of Specific Surface Area of Metal Hydrides by Lixiviation,” J. Alloys Compounds, 266:307–310 (1998). 255. Fujii, H., Higuchi, K., Yamamoto, K., Kajioka, H., Orimo, S., and Toiyama, K., “Remarkable Hydrogen Storage, Structural and Optical Properties in Multi-Layered Pd/Mg Thin Films,” Mater. Trans., 43:2721–2727. (2002).

9: Gas Reactive Applications, Trudeau

435

256. Spassov, T., Lyubenova, L., Köster, U., and Baro, M.D., “Mg—Ni—RE Nanocrystalline Alloys for Hydrogen Storage,” Mater. Sci. Eng. A, 375/377:794–799 (2004). 257. Dehouche, Z., Klassen, T., Oelerich, W., Goyette, J., Bose, T.K., and Schulz, R., “Cycling and Thermal Stability of Nanostructured MgH2-Cr2O3 Composite for Hydrogen Storage,” J. Alloys Compounds, 347:319–323 (2002). 258. Liang, G., and Schulz, R., “Mechanically Alloyed Nanocrystalline Hydrogen Storage Materials,” Mater. Trans., 42:1593–1598 (2001). 259. Hanada, N., Ichikawa, T., Orimo, S.I., and Fujii, H., “Correlation between Hydrogen Storage Properties and Structural Characteristics in Mechanically Milled Magnesium Hydride MgH2,” J. Alloys Compounds, 266:269–273 (2004). 260. Aizawa, T., Kuji, T., and Nakano, H., “Nanostructured Hydrogen Storage Materials via Bulk Mechanical Alloying,” Proceedings of Processing and Fabrication of Advanced Materials X, 5–8 November 2001, ASM International, 164–182 (2002). 261. Zuttel, A., Rentsch, S., Wenger, P., Sudan, P., Mauron, Ph., and Emmenegger, Ch., “Hydrogen Storage Materials: Metal, Carbon and Complexes,” Proceedings of Processing and Fabrication of Advanced Materials XI, 7–10 October 2002, ASM International, 107–122 (2003). 262. Ritter, J.A., Ebner, A.D., Wang, J., and Zidan, R., “Implementing a Hydrogen Economy,” Materials Today September:18–23 (2003). 263. Bogdanovic, B., and Schwickardi, M., “Ti-doped Alkali Metal Aluminum Hydrides as Potential Novel Reversible Hydrogen Storage Materials,” J. Alloys Compounds, 253/254:1–9 (1997). 264. Bogdanovic, B., Brand, R.A., Marjanovic, A., Schwickardi, M., and Tolle, J., “Metal-Doped Sodium Aluminum Hydrides as Potential New Hydrogen Storage Materials,” J. Alloys Compounds, 302:36–58 (2000). 265. Jensen, C.M., Zidan, R., Mariels, N., Hee, A., and Hagen, C., “Advanced Titanium Doping of Sodium Aluminium Hydride: Segue to a Practical Hydrogen Storage Material?” Int. J. Hydrogen Energy., 24:461–465 (1999). 266. Gomes, S., Renaudin, G., Hagemann, H., Yvon, K., Sulic, M.P., and Jensen, C.M., “Effects of Milling, Doping and Cycling of NaAlH4 Studied by Vibrational Spectroscopy and X-ray Diffraction,” J. Alloys Compounds, 390:305–313 (2005). 267. Easton, D.S., Schneibel, J.H., and Speakman, S.A., “Factors Affecting Hydrogen Release from Lithium Alanate (LiAlH4),” J. Alloys Compounds, 398:245–248 (2005). 268. Zuttel, A., Rentsch, S., Fisher, P., Wenger, P., Sudan, P., Mauron, P., and Emmenegger, C., “Hydrogen Storage Properties of LiBH4,” J. Alloys Compounds, 356/357:515–520 (2003). 269. Zuttel, A., Wenger, P., Rentsch, S., Sudan, P., Mauron, P., and Emmenegger, “LiBH4 a New Hydrogen Storage Material,” J. Power Sources, 118:1–7 (2003). 270. Rosi, N.L., Eckert, J., Eddaoudi, M., Vodak, D.T., Kim, J., O’Keeffe, M., and Yaghi, O.M., “Hydrogen Storage in Microporous Metal–Organic Framework,” Science, 300:1127–1129 (2003).

436

Properties

271. Seayad, A.M., and Antonelli, D.M., “Recent Advances in Hydrogen Storage in Metal-Containing Inorganic Nanostructured and Related Materials,” Adv. Mater., 16:765–777 (2004). 272. Jhi, S.-H., and Kwon, Y.-K., “Hydrogen Adsorption on Boron Nitride Nanotubes: A Path to Room-Temperature Hydrogen Storage,” Phys. Rev. B, 69:245407–245410 (2004). 273. Züttel, A., and Orimo, S-I., “Hydrogen in Nanostructured, Carbon-Related, and Metallic Materials,” MRS Bulletin, Sept,:705–711 (2002). 274. Dillon, A.C., Jones, K.M., Bekkedahl, T.A., Kiang, C.H., Bethume, D.S., and Heben, M.J., “Storage of Hydrogen in Single-walled Carbon Nanotubes,” Nature, 386:377–379 (1997). 275. Chamber, A., Park, C., Baker, R.T.K., and Rodriguez, N.M., “Hydrogen Storage in Graphitic Nanofibers,” J. Phys. Chem. B, 102:4253–4256 (1998). 276. Lueking, A., and Yang, R.T., “Hydrogen Spillover to Enhance Hydrogen Storage—Study of the Effect of Carbon Physicochemical Properties,” Appl. Catal. A, 265:259–268 (2004). 277. Liu, C., Fan, Y.Y., Liu., M., Cong, H.T., Cheng, H.M., and Dresselhaus, M.S., “Hydrogen Storage in Single-Walled Carbon Nanotubes at Room Temperature,” Science, 286:1127–1128 (1999). 278. Gupta, B.K., and Srivastava, O.N., “Investigation on the Carbon Special Form Graphitic Nanofibres as a Hydrogen Storage Materials,” in: Hydrogen Materials Science and Chemistry of Carbon Nanomaterials (T.N. Veziroglu, S.Y. Zaginaichenko, D.Y. Schur, B. Baranowski, A.P. Shpak, and V.V. Skorokhod, eds.), Mathematics, Physics and Chemistry 172:177–184, Kluwer Academic, Dordrecht (2004). 279. Ritschel, M., Uhlemann, M., Gutfleisch, O., Leonhardt, A., Graff, A., Taschner, Ch., and Fink, F., “Hydrogen Storage in Different Carbon Nanostructures,” Appl. Phys. Lett., 80:2985–2987 (2002). 280. Tarasov, B.P., Maehlen, J.P., Lototsky, M.V., Muradyan, V.E., and Yartys, V.A., “Hydrogen Sorption Properties of Arc Generated Single-Wall Carbon Nanotubes,” J. Alloys Compounds, 356–357:510–514 (2003). 281. Hirscher, M., Becher, M., Haluska, M., Zeppelin, F.V., Chen, X., DettlaffWeglikowska, U., and Roth, S., “Are Carbon Nanostructures an Efficient Hydrogen Storage Medium,” J. Alloys Compounds, 356–357:433–437 (2003). 282. Orimo, S., Züttel, A., Schlapbac, L., Najer, G., Fukunaga, T., and Fujii, H., “Hydrogen Interaction with Carbon Nanostructures: Current Situation and Future Prospects,” J. Alloys Compounds, 356–357:716–719 (2003). 283. Zacharia, R., Kim, K.Y., Kibria, A.K.M.F., and Nahm, K.S., “Enhancement of Hydrogen Storage Capacity of Carbon Nanotubes via Spill-Over from Vanadium and Palladium Nanoparticles,” Chem. Phys. Lett., 412:369–375 (2005). 284. Orimo, S., Maher, G., Fukunaga, T., Zuttel, A., Schlapbach, L., and Fujii, H., “Hydrogen in the Mechanically Prepared Nanostructured Graphite,” Appl. Phys. Lett., 75:3093–3095 (1999). 285. Shindo, K., Kondo, T., and Sakurai, Y., “Dependence of Hydrogen Storage Characteristics of Mechanically Milled Carbon Materials on their Host Structures,” J. Alloys Compounds, 372:201–207 (2004).

9: Gas Reactive Applications, Trudeau

437

286. Imamura, H., Tabata, S., Shigetomi, N., Takesue, Y., and Sakata, Y., “Composites for Hydrogen Storage by Mechanical Grinding of Graphite Carbon and Magnesium,” J. Alloys Compounds, 330–332:579–583 (2002). 287. Ichikawa, T., Fujii H., Isobe, S., and Nabeta, K., “Rechargeable Hydrogen Storage in Nanostructured Mixtures of Hydrogenated Carbon and Lithium Hydride,” Appl. Phys. Lett., 86:241914–241916 (2005). 288. Chen J., Li, S.L., and Tao, Z.L., “Novel Hydrogen Storage Properties of MoS2 Nanotubes,” J. Alloys Compounds, 356/357:413–417 (2003). 289. Ma, R., Bando, Y., Zhu, H., Sato, T., Xu, C., and Wu, D., “Hydrogen Uptake in Boron Nitride Nanotubes at Room Temperature,” J. Am. Chem. Soc., 124:7672–7673 (2002). 290. Chen, J., Li, S.L., Tao, Z.L., Shen, Y.T., and Cui, C.X., “Titanium Disulfide Nanotubes as Hydrogen-Storage Materials,” J. Am. Chem. Soc., 125:5284–5285 (2003). 291. Pan, L., Sander, M.B., Huang, X., Li, J., Smith, M., Bittner, E., Bockrath, B., and Johnson, J.K., “Microporous Metal Organic Materials: Promising Candidates as Sorbents for Hydrogen Storage,” J. Am. Chem. Soc., 126:1308–1309 (2004). 292. Chen, D., Chen, L., Liu, S., Ma, C.X., Chen, D.M., and Wang, L.B., “Microstructure and Hydrogen Storage Property of Mg/MWNTs Composites,” J. Alloys Compounds, 372:231–237 (2004). 293. Holtz, R.L., Provenzano, V., and Imam, M.A., “Overview of Nanophase Metals and Alloys for Gas Sensors, Getters, and Hydrogen Storage,” Nanostruct. Mater., 7:259–264 (1996). 294. Bryden, K.J., and Ying, J.Y., “Electrodeposition Synthesis and Hydrogen Absorption Properties of Nanostructured Palladium-Iron Alloys,” Nanostruct. Mater., 9:485–488 (1997).

10 Magnetic Nanoparticles and Their Applications Sara A. Majetich Physics Department, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA

10.1 Introduction The meaning of the term nanoscale varies depending on the property of interest. For the optical properties of semiconductors, it is the dimension of an exciton. For the mechanical properties of structural materials, it is the separation between dislocations. A natural length scale in ferromagnetic materials is the maximum size of a magnetic domain, which is on the order of tens of nanometers. There has been extensive research on nanoscale magnetic materials in recent years, and this review can cover only selected parts. The reader is referred elsewhere for more in-depth surveys of mesoscale magnetism,[1–4] magnetic nanocomposites,[5–6] and synthetic methods used to make magnetic nanoparticles.[7] There has been extensive research on self-assembly of ordered structures, but since this is not magnetically driven, this topic will also be omitted. There has been a recent spate of interest in the use of metallic Fe nanoparticles for environmental remediation of subsurface chlorinated hydrocarbons,[8–11] but here the Fe(0) chemistry is central, rather than the magnetism. Micrometer-scale particles are used in magnetorheological fluids,[12] whose viscosities change by orders of magnitude when a magnetic field is applied. Although nanoparticles may be mixed in as well,[13] the magnetism of these materials is dominated by the multidomain structure. For similar reasons, nanocrystalline soft materials for high-frequency applications,[14–16] and exchange spring nanocomposites for permanent magnet applications,[17–18] and magnetophotonic crystals,[19–23] will not be discussed here. Magnetocaloric materials based on coupled magnetic clusters showed promise in the 1990s,[24] but more recently bulk compounds with a structural phase transition have superseded them,[25–26] so this topic will be omitted. This review will focus on Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 439–486 © 2007 William Andrew, Inc.

439

440

Properties

Figure 10.1 Applications of magnetic nanoparticles take advantage of their ability to (a) generate flux when close to a magnetoresistive sensor; (b) create inhomogeneous magnetic fields that change the relaxation times of H nuclei in nearby water molecules; (c) move in response to an inhomogeneous magnetic field, dragging along entities such as cells linked through monoclonal antibodies; and (d) have metastability, so that particles or grains with different moment directions can be used for information storage.

monodomain nanoparticles and selected magnetic materials made from them in which the monodomain physics is critical to their behavior. Many of the important features of magnetic nanoparticles used in various applications can be related to basic magnetic properties (Fig. 10.1). The magnetic moment per particle, μ, depends on the particle volume and the material it is made of–or the number of atoms it has and the size of the atomic moment. The force that causes the particle to move  in an external magnetic field gradient is proportional to μ, so this is an important parameter for magnetic separation. The magnitude of the dipolar magnetic field generated by a particle is also proportional to μ. In magnetic resonance imaging (MRI), the inhomogeneous magnetic field of contrast agents modifies the relaxivity of water protons, or their ability to change their rate of magnetic relaxation in water relative to that in solid tissues. By removing the contribution of water to the MRI images, the structure of surrounding tissues can be observed more clearly. The rela tively small μ of a nanomagnet makes it possible to change its direction using a spin-polarized current. Micrometer-scale spin valve devices using monodomain thin-film layers are now in use in the read heads of magnetic disc drives. Related tunnel junction structures are currently being investigated as possible magnetic random access memory (MRAM). The switching field or coercivity, Hc, is a maximum for particles at the largest monodomain size. Magnetic recording media currently use thin films with

10: Magnetic Nanoparticles and Applications, Majetich

441

nanoscale grains, and nanoparticle-based media are an area of intense research interest. Magnetic nanoparticles can absorb energy proportional to the area within the hysteresis loop from an AC (electro)magnetic field during each cycle. This generates local heating that may be of use for hyperthermic cancer treatment. Following a brief introduction to ferromagnetic materials will be an indepth description of the physics of monodomain particles. Applications based on magnetic nanoparticles, both current and future, will be described in terms of these basic properties, and the materials challenges will be identified.

10.2 Fundamental Physics of Magnetic Nanoparticles 10.2.1 Bulk Ferromagnetism The total energy of a bulk ferromagnet is the sum of contributions from exchange, magnetostatic, anisotropy, and Zeeman energy terms. In a ferromagnetic material, quantum-mechanical exchange interactions make the magnetic moments of neighboring atoms parallel, even in the absence of an external magnetic field. This coupling of the atomic moments makes the sample have a large magnetization, or moment per unit volume. If the temperature is increased until the thermal energy is comparable to the exchange energy, long-range ferromagnetic ordering is lost. The temperature at which the spontaneous magnetization drops to zero is known as the Curie temperature, TC. Measurement of the Curie temperature of a material is used to determine the exchange stiffness, A: TC ≈ Aa/k

Eq. (10-1)[27]

where a is the nearest neighbor separation between spins, and k is the Boltzmann constant. Short-range (<1 nm) exchange interactions dominate the nearestneighbor coupling, but long-range magnetostatic interactions between moments in different locations within the ferromagnet are also important. In a bulk ferromagnet, the overall energy is minimized by magnetic domain formation (Fig. 10.2). While neighboring atomic moments are nearly parallel, the moments 100 nm away may all point in a different direction. The magnetostatic energy reduction through the formation of flux closure pathways exceeds the increase in energy due to the domain walls, where neighboring atomic moments are not exactly parallel.

442

Properties

Figure 10.2 Exchange coupling in a ferromagnetic material (a) leads to parallel orientation of the magnetic moments of neighboring atoms, whereas dipolar coupling (b) can lead to antiparallel alignment. The combination of short-range exchange and long-range magnetostatic interactions is responsible for magnetic domain formation in bulk ferromagnets (c).

In the absence of an applied field, the direction of the moment is determined by the anisotropy energy density, or anisotropy, K. At low temperature the moment would lie along a particular crystallographic direction called the easy axis. The magnitude of K is determined by the energy required to rotate the magnetic moment away from the easy axis. Magnetocrystalline effects due to spin–orbit coupling, as well as shape and stress contribute to the overall value of K. Cubic crystal structures tend to have lower values of magnetocrystalline anisotropy than those with tetragonal or hexagonal lattices, where the moment must rotate 180° between easy axis directions. The latter materials are said to be uniaxial, and are important for applications with a stable magnetization, such as permanent magnets or magnetic recording media. The behavior of a ferromagnet is often reported in terms of three technical parameters. The saturation magnetization, Ms, is the magnetic moment per unit volume when all atomic moments are aligned parallel to an external field. After saturating a sample, if the field is reduced to zero there will be a remanent magnetization, Mr. In order to reduce the magnetization to zero, an applied field in the opposite direction equal to the coercivity, Hc, must be applied. Together, these three parameters describe the basic features of a hysteresis loop, which is a plot of the magnetization between positive and negative saturation, and back again (Fig. 10.3).

10.2.2 Magnetic Clusters Magnetism in small clusters is described in a review by Bucher[28] and only a few details will be discussed here. A single gas-phase Fe atom has a magnetic moment of 5 Bohr magnetons (μB), while in bulk Fe, the moment per atom is only 2.2 μB. The difference is understood in terms of the filling of the energy bands in the bulk solid,[29–30] but this raises the

10: Magnetic Nanoparticles and Applications, Majetich

443

Figure 10.3 Hysteresis loop showing the behavior of the material magnetization M in response to an applied magnetic field H.

question of whether magnetic nanoparticles have higher moments than the bulk. The evolution of the magnetic moment with increasing particle size has been studied using cluster beam mass spectroscopy methods.[31–32] When a beam of monodisperse clusters is passed through an inhomogeneous magnetic field, there is a deflecting force proportional to the magnitude of the cluster moment. Using this technique, the moment as a function of the cluster size has been determined.[31] The work of de Heer and co-workers showed that clusters of 500–700 atoms or more have the same moment per atom as in the bulk. The change is not monatomic but oscillates with increasing size, suggesting periodic changes in the electronic structure as an additional layer of atoms are the cluster surface is completed. The results suggest that particles with diameters greater than will have the same moment per atom as in the bulk. In small clusters all atoms are part of the surface, and there is greater magnetic anisotropy than in the bulk.[33]

10.2.3 Molecular Magnetism Collective magnetic behavior has been observed in small molecules containing Mn or Fe.[34–38] Here the “magnetic atoms” are bonded indirectly to each other through bridging oxygen atoms, and outside the metal oxide cluster there are surface ligands that isolate one cluster from another. The relative orientation of the moments may be complex. For example, the [Mn12O12(CH3COO)16]·2CH3COOH·4H2O molecule, more commonly known as Mn12 acetate or just “Mn12,” contains eight Mn3+ ions with spins oriented in one direction plus four Mn4+ ions with their moments in the

Properties

444

opposite direction. Unlike larger magnetic nanoparticles, molecular magnets can be made in macroscopic quantities (milligrams or more) with true monodispersity: exactly the same number of atoms. Molecular magnets have been a fascinating system for testing quantum-mechanical predictions, because the number of spins involved is small. Because the clusters are small, low temperatures (<2 K) are needed to minimize thermal fluctuation effects. Since the molecules are identical and have negligible interactions with each other, sharp jumps are observed in the magnetization as a function of the applied field when the spin state of the cluster changes. Molecular magnets are being investigated as a test system for quantum computing because it is possible to prepare them in a superposition of states.[37–39]

10.2.4 Ideal Monodomain Particles Made from Ferromagnetic Materials Below a critical size, it is not energetically favorable to form a domain wall, and the particle is said to be monodomain. The concept of a magnetic domain was first postulated by Frenkel and Dorfman.[40] The maximum monodomain size dcr is predicted using the relation 72( AK ) dcr = Ms2

12

Eq. (10-2)[41]

In spherical particles, it ranges from ∼20 to several hundred nanometers. As particles are reduced in size, relative to the bulk, the coercivity increases, reaching the maximum value at dcr[42] There have been numerous studies of size-dependent magnetic properties in monodomain particles.[43–48] In the absence of an applied magnetic field, a bulk ferromagnet may have no net magnetization, since the contributions from different domains will cancel. A monodomain particle, however, will always act as a giant moment. In the Stoner–Wohlfarth model for an ellipsoidal monodomain particle of a uniaxial material, all spins within the particle are aligned, and magnetization reversal occurs by coherent rotation.[49–51] Here the atomic spins remain parallel as they rotate to a new magnetic moment direction. The energy of an isolated monodomain particle with volume V is the sum of the anisotropy and Zeeman energy terms, Ept = KV sin2 θ − MsHV cos(φ − θ)

Eq. (10-3)

10: Magnetic Nanoparticles and Applications, Majetich

445

Here θ is the angle between the particle magnetic moment, which has a magnitude m = MsV, and the easy axis; φ is the angle between the easy axis and the applied field H (Fig. 10.4). When particles are free to rotate, as in a ferrofluids, the anisotropy and Zeeman terms can be minimized simultaneously, but for fixed particles, the moment direction is determined by the competition between the interactions, and by the significance of thermal fluctuations. Figure 10.4 Schematic of a The net magnetization of an ensemble of monodomain particle, describing the angles used in identical, noninteracting, uniaxial particles determining the overall can be predicted in terms of an energy energy in Eq. (10-3). θ is the barrier model. Suppose the field is applied angle between the magnetic parallel to the easy axis direction, so that φ = moment and the easy axis, 0. Figure 10.5 shows that the energy minima and φ is the angle between occur when θ = 0° and 180°, and that there is the applied field and the easy axis. an energy maximum in between. When an external field is applied, one minimum is lower in energy. Classically, the rate of particles reversing their magnetization depends on the barrier height, relative to the thermal energy.[52] It also depends on the attempt frequency, which is comparable to the Larmor precession frequency of the moment (τ0−1 ≈ 109 Hz). Particles in the local minimum reverse their magnetization directions with a rate τ−1 = τ0−1exp(ΔE/kT) Eq. (10-4) where the barrier height is given by MH ΔE = KV ⎡⎢1 − s ⎤⎥ 2K ⎦ ⎣

2

= KV [1 − H HK ]

2

Figure 10.5 Schematic of the energy barrier for a uniaxial particle with the field aligned along the easy axis.

Eq. (10-5)

HK is known as the anisotropy field, and it represents the highest possible switching field for a material. As the applied field is increased, the

Properties

446

magnitude of the energy barrier is reduced. The coercivity of the ensemble depends on the measurement time, τmeas. It is given by ⎡ ln( τ meas τ 0 )kT ⎞ Hc = Hk ⎢1 − ⎛ ⎝ ⎠ KV ⎣

12

⎤ ⎥ ⎦

Eq. (10-6)[53]

When the measurement time, τmeas, is much greater than the characteristic relaxation time, τ, the sample will have approached equilibrium, and no coercivity will be observed. The rate of magnetic relaxation is also very sensitive to the temperature. If kT << ΔE, a nonequilibrium magnetization is measured and this gives rise to hysteresis. When kT >> ΔE, thermal fluctuations will tend to demagnetize the sample. When the coercivity is zero due to thermal fluctuations, the sample is said to be superparamagnetic. The threshold temperature for superparamagnetism is called the blocking temperature, TB, which is given by TB = KVln(τmeas/τ0)/k

Eq. (10-7)

Below this temperature the particles are said to be blocked. The blocking temperature depends on the measurement time, and therefore on the type of experiment. For DC superconducting quantum interference device (SQUID) magnetometer or vibrating sample magnetometer (VSM) measurements, τmeas ≈ 102 s. For AC susceptometers, τmeas is equal to the inverse of the excitation field frequency. In Mössbauer spectrometry, τmeas ≈ 10−9 s, and in neutron scattering measurements, τmeas ≈ 10−12–10−7 s. The blocking temperature is not a critical temperature like TC; rather, it is a convenient measurement for the estimation of the effective anisotropy, K. With DC magnetometry, TB is often determined by measuring the zero field-cooled (zfc) and field-cooled (fc) magnetization as a function of temperature. Mzfc(T) is determined by cooling a sample of nanoparticles with H = 0 to low temperature, so that the magnetic moments of the particles have random orientations. A small, constant field (typically ∼200 Oe or less) is applied so that there is a measurable magnetization as the temperature is increased. Mzfc increases as the thermal energy is raised, and there is sufficient energy to align the particle moments parallel to the field. Mzfc drops again at high temperature when thermal fluctuations are able to demagnetize the sample. The field-cooled magnetization Mfc(T) is

10: Magnetic Nanoparticles and Applications, Majetich

447

measured by first applying a small field at room temperature. As the sample is cooled, the magnetization rises as thermal fluctuations become less important. Unlike Mzfc(T), the field-cooled magnetization saturates at low temperature. TB is defined as the temperature point of deviation between Mfc and Mzfc, which is usually quite close to the peak temperature of the zero field-cooled curve. In the Néel model for the rate of equilibration of the magnetization[54] the magnetization, relative to its value at zero temperature is a Langevin function, M 1 = L( x ) = coth( x ) − M (0) x

Eq. (10-8)

where x = μH/kT. Because the Langevin function depends on the ratio H/T, data collected at different temperatures will scale to a universal curve.[55–56] The magnetization curve can be fit to extract values for the average particle moment μ,[57] and if the magnetization of the material is assumed to be constant, it can be used to determine the particle size distribution.[58] An ensemble of identical, noninteracting particles will have a magnetization that decays exponentially with time, M(t) = M0 exp(−t/τ)

Eq. (10-9)

where M0 is the magnetization at zero time, t0.[59–60] This has been verified in microSQUID measurements on single particles.[61–62] However, if there is a distribution of energy barriers, due to variations in the particle size, crystallographic orientation, or anisotropy, or to magnetostatic interactions among the particles, then there will be a range of values for τ, and the contributions must be integrated. Street and Woolley have shown that a flat distribution of finite width yields a magnetization that decays logarithmically in time, M(t) = M0 − S ln(t/t0)

Eq. (10-10)

where S is the magnetic viscosity.[63] While this distribution is a crude approximation, in a wide variety of nanoparticle systems, logarithmic decay is observed over the timescale ranging from 102 to 105 seconds. At shorter time scales, deviations are evident,[64] but the viscosity model is

Properties

448

useful for predicting long-time behavior, such as the thermal demagnetization of magnetic recording media. For submicrosecond timescales, the precession of a magnetic moment about an applied field cannot be ignored. With only a Zeeman energy term, in equilibrium a magnetic moment will be parallel to the applied field. When the field is first applied, however, the moment will precess counterFigure 10.6 Schematic of clockwise around it at the Larmor precesthe magnetization vector sion frequency, circling around and precessing about an applied field, showing both gradually decaying inward (Fig. 10.6). Two longitudinal and transverse different but equivalent approaches are relaxation. commonly used. In the nuclear magnetic resonance community, the Bloch–Bloembergen equations are standard when determining spin–lattice and spin–spin relaxation times. Here   ∂Mz M − Mz = − γ ( M × H )z − s T1 ∂t

Eq. (10-11)

where γ is the gyromagnetic ratio and T1 is the longitudinal or spin–lattice relaxation time. The transverse components of the magnetization are described by the equation   ∂M x ,y M − M x ,y = − γ ( M × H ) x ,y − s ∂t T2

Eq. (10-12)

and T2 is the transverse relaxation time associated with the dephasing of different spins. T2 is generally shorter than T1. In the magnetic recording community, the components are not generally separated, and physicists refer to the Gilbert damping constant α rather than to the relaxation times. Numerical solutions to the Landau–Lifschitz–Gilbert equation are used in micromagnetics to model magnetic moment distribution within a sample of known geometry as a function of time:     α  dM dM = −γ (M × H) + M × dt M dt

Eq. (10-13)[65]

10: Magnetic Nanoparticles and Applications, Majetich 449   The −( M × H ) term is in the direction of the precessing magnetization,   dM   and the M × term pushes M so that it aligns parallel to H . dt

10.2.5 Macroscopic Quantum Tunneling While most of the magnetic behavior of monodomain nanoparticles can be explained in terms of thermal activation and the energy barrier model of section 10.2.4, quantum-mechanical tunneling through the barrier can also occur.[66–67] Magnetic nanoparticles and clusters have been important systems for the study of macroscopic quantum tunneling (MQT) because their thermally activated magnetization rates can be predicted quantitatively, enabling the tunneling contributions to be extracted from experimental results. Tunneling effects are noticeable only at very low temperatures where there is a very low rate for thermally activated reversal. The experimental signature of MQT is a magnetic relaxation rate or viscosity S that is independent of temperature. This has been observed in a number of nanoparticle systems.[68–73] However, the interpretation in terms of MQT remains controversial because energy barrier distributions that can arise from a variety of sources can also lead to temperature-independent viscosities.[74] Magnetic tunneling has been most clearly verified in molecular magnets.[39,75]

10.2.6 Surface and Interface Effects in Nanoparticles The Stoner–Wohlfarth model yields an exact solution for ideal monodomain particles in which all of the atomic spins are parallel, and it ignores more complex interactions within and between the particles. Within a single particle, the atomic moments at the surface will be exchange-coupled to fewer nearest neighbors than those within the particle core. Just as ultrathin atomic films have reduced Curie temperatures,[76] the surface spins of a nanoparticle will have reduced exchange energies. At the same time, symmetry is broken at a surface or interface, and nanoparticles of cubic compounds will have enhanced surface anisotropy.[77] Because the core and surface spins are still exchangecoupled, this leads to increased effective values of K and higher switching fields. In pure nickel ferrite nanoparticles, open hysteresis loops have

450

Properties

been observed up to 17 tesla, and this result has been interpreted in terms of spin canting at the nanoparticle surface.[78] Large surface anisotropies in nanoparticles have been reported by many groups.[79–82] Core–shell nanoparticles of two different materials, usually a metallic core and metal oxide shell made via partial oxidation, show many related effects. In metallic corrosion, oxygen dissociates on the metal surface, and metal ions migrate to the surface to react.[83] There are multiple nucleation sites, and therefore many tiny nanocrystals of the metal oxide phase.[84] These nanocrystals grow until they impinge on each other, and then further oxidation proceeds much more slowly. At this stage the particles are said to be passivated, though strictly speaking oxidation still occurs. The strong exchange coupling between the nanoparticle core and oxide shell leads to a shift in the field-cooled hysteresis loop, which was first demonstrated for Co/CoO nanoparticles.[85] The higher reversal field for antiferromagnetic CoO prevents the exchange-coupled Co from switching at lower fields. The effect is asymmetric because the direction of the applied field during cooling selects an easy magnetization direction. CoO has nearly zero net magnetization, yet it effectively increases the coercivity of the Co nanoparticle core. This phenomenon is referred to as pinning in thin-film multilayers. Small-angle neutron scattering (SANS) studies of Fe/Fe oxide nanoparticles showed that at moderate fields the spins in the ferromagnetic shell had a net perpendicular magnetization.[86] This is similar to the behavior of an antiferromagnet or ferrimagnet in a spin-flop phase, except that it is stabilized at fields of up to 5 tesla, presumably due to higher surface anisotropy. Core–shell nanoparticles have been proposed as a possible solution to the superparamagnetic limit problem in magnetic data storage.[87,88] Core–shell nanoparticles with passivating surfaces have additional interest because they enable particles to have higher magnetic moments than pure oxides. This attribute would be useful in ferrofluids, for MRI contrast agents, and also for high-frequency nanocomposite inductors.[89] There has been a report of true epitaxial oxide coating on nanoparticles prepared by high-temperature gas-phase methods, and then slowly oxidized.[90] High-resolution transmission electron microscopy (TEM) showed epitaxial conformation, with significant strain in the curved regions. The passivation layer was 4–5 nm thick. Unfortunately the particles involved were ∼80 nm in diameter, and it is unclear whether the same type of passivation could be achieved in small particles with much greater curvature.

10: Magnetic Nanoparticles and Applications, Majetich

451

10.2.7 Magnetostatic Interactions between Nanoparticles Monodisperse monodomain magnetic nanoparticles act like giant magnetic moments, and the fields that they generate will perturb the behavior of nearby particles. Because dipolar forces are long-range, the effects are noticeable with only 1 vol% of particles. Compared with exchange interactions within a bulk ferromagnet, the magnetostatic interactions between particles are weak. It is possible to have a purely dipolar ferromagnet,[91–93] but the equivalent of a Curie temperature would be far below that of a bulk magnet of the same material. There have been numerous experimental studies of magnetostatic effects in monodisperse particles both in dilute,[94–113] and concentrated[86,114–133] dispersions. The upper limit to the volume fraction is limited only by the thickness of the particle coating and the size of the particles. A typical dilute assembly is a frozen ferrofluid with 5 vol% or less. A typical concentrated sample is a self-assembled array with nearly close-packed surfactant-coated particles and a volume fraction of between 25 and 45%. The particles used in these studied are superparamagnetic at room temperature, and are typically less than 10 nm in diameter. The broadening of the zero field-cooled magnetization curve and shifting to higher temperature is the most common signature of collective behavior. The coercivity and remanence show more complex changes. The most sensitive magnetic properties to magnetostatic interactions are the time-dependent magnetic relaxation and the frequency-dependent susceptibility, χAC. With a small AC driving field HAC, the susceptibility will have in-phase (χ′) and out-of-phase (χ″) components, χ AC =

M = χ ′ − iχ ′′ HAC

Eq. (10-14)

The temperature dependence of χ″ can be related to the energy barriers present in the system, and can help to experimentally quantify the volume fraction of particles that are coupled and to determine the coupling energy[101,127] The temperature of a peak in χ″ reflects the blocking temperature for the coupled particles, and its frequency dependence reveals the timescale over which the magnetization of the coupled nanoparticle clusters is stable. While DC measurements of the zero field-cooled magnetization tend to broaden and shift to higher temperatures, the AC

452

Properties

out-of-phase susceptibility shows much sharper temperature-dependent features.[127] Dilute frozen ferrofluids have also been shown to behave like dipolar spin glasses.[102] However, more concentrated assemblies do not have a sharp cusp in the magnetization versus temperature curve, but they likely contain clusters of coupled particle moments over a range of sizes. This makes them “spin glass-like” rather than a true spin glass, where scaling relations and critical exponents apply. This is sometimes called a magnetic cluster glass or mictomagnet, which has large interacting clusters but no long-range behavior.[134] At low enough temperatures and over short enough timescales, the assemblies can be considered to be magnetostatic ferromagnets, or superferromagnets. The length scale of the magnetic correlations can be as large as 70 nm,[86] but it is unclear whether magnetic domains are formed. In some nanoparticle assemblies, strong antiferromagnetic correlations are observed.[123] Different groups have reported conflicting results on whether the coercivity and remanent magnetization go up or down with increasing particle concentration. There have been a variety of theoretical approaches.[135–142] Kechrakos and Trohidou have used Monte Carlo methods to simulate the coercivity and remanence at different temperatures, as a function of the ratio of the dipolar and magnetocrystalline anisotropy energies.[140] The magnetostatic energy in different systems was compared using values of the energy per pair of aligned dipoles, given by Ems = μpt2 /4πa3, where μpt = MsV is the particle moment and a is the centerto-center separation. The anisotropy energy was given by EK = KV. In dilute dispersions, magnetostatic coupling introduces a preference for being magnetized in a particular direction at zero applied field, which is an effective anisotropy. In concentrated samples it generates a Lorentz cavity field within the assembly.[140] In a simplified picture of dilute dispersions with weak interactions, the effect of the dipolar field is to generate an effective anisotropy in clusters with small numbers of particles. They are strongly coupled to each other and weakly coupled to other clusters. In a dense assembly such as the self-assembled nanoparticle arrays, however, magnetostatic interactions between neighboring clusters will affect the local demagnetizing field. It is feasible that in intermediate applied fields a domain-like structure develops within the arrays that contains flux closure pathways and therefore stabilizes some of the particle moments against switching. A useful analytical tool for quantifying the interactions among magnetic nanoparticles is known as a ΔM plot. Two types of magnetization curves are used to determine ΔM. The remanent or DC demagnetization,

10: Magnetic Nanoparticles and Applications, Majetich

453

Mr(H) = Mdcd(H), is found by first saturating and then measuring the remanence after applying different values of H. The isothermal remanent magnetization, Mirm(H), is the remanence measured after starting with a demagnetized sample at H = 0, then applying a field H and measuring the remanence. By convention, the sign of the field is opposite for Mdcd(H) and Mirm(H). The portion of the magnetization curve of interest is the region where particles are changing their magnetization directions. For noninteracting particles, Mdcd(H) = M(∞) − 2Mirm(H)

Eq. (10-15)[136]

The deviation from ideal behavior due to interactions, ΔM, is defined as ΔM ( H ) =

Mdcd( H ) ⎡ M (H) ⎤ − 1 − 2 irm M (∞) ⎢⎣ M (∞) ⎥⎦

Eq. (10-16)[137]

Where ΔM is positive the net interactions are magnetizing, and where it is negative they are demagnetizing[138] If interactions couple particle moments together ferromagnetically, higher fields will be needed for reversal, compared with those for noninteracting particles. If the coupling is antiferromagnetic, the particles will switch at lower fields. Interactions also lead to a sharper switching field distribution. ΔM reflects the overall behavior, and reveals nothing about the sizes and size distributions of the magnetically coupled particle clusters. However, it is a straightforward measurement, and has been widely used to characterize nanoparticle powders and magnetic tape media.

10.2.8 Exchange Interactions between Nanoparticles Most prior work on interactions of monodisperse nanoparticles has dealt solely with magnetostatics, since the nanoparticles in arrays and ferrofluids were in organic matrices. Magnetostatic interactions alone are sufficient to favor a ferromagnetic ground state,[91–93] though the ordering temperature, TCdip, will be relatively low.[92] It is predicted to depend on the magnetic moment and the distance between moments. In an array with particle moment μpt and center-to-center particle distance s + d, this model suggests that

Properties

454 T dip C ∝

μ 2pt

(s + d )

3

Eq. (10-17)

Nanocrystalline metals,[143–144] and alloys for high-frequency applications[14–16] have significant exchange interactions between magnetic grains. This coupling leads to much larger interaction domains, which can be micrometers in size.[145] When the soft-magnetic alloys are heated above the Curie temperature of the amorphous matrix phase, the coercivity rises sharply because the grains become magnetically decoupled. Granular giant magnetoresistive nanocomposites[146–147] also show some exchange effects. While the magnetic behavior has been modeled in terms of strongly interacting superparamagnets,[148] magnetostatic interactions alone are insufficient for quantitative agreement with experimental results.[149–150] There have also been hints of exchange interactions in twodimensional assemblies of Fe islands that showed surprisingly high blocking temperatures.[151–152] Several theoretical papers have made predictions concerning the effect of size, separation, and matrix material on the strength of Rudiger–Kittel–Kasuya–Yoshida (RKKY) exchange interactions between pairs of nanoparticles.[153–157] Up until now the ability to relate these intriguing experimental results to theoretical models has been limited by the difficulty in controlling the particle size and spacing. The exchange strength Jpt between nanoparticles is predicted to depend on the probability of electron tunneling, and therefore on the interparticle separation s and a barrier height φ: J pt ~ (1 s 2 ) exp[ − 2 m* φ s ]

Eq. (10-18)

where m* is the electron effective mass and h is Planck’s constant.[153] In a typical scanning tunneling microscopy (STM) experiment, the barrier is vacuum, and the barrier height is the work function of the metal tip. In that case the exponential factor ( 2m *φ ) h is on the order of 1 Å−1. Here we are interested in cases where the electrons pass through materials more conducting than vacuum, where the relevant barrier height is the energy needed to generate a conduction electron. Here the inverse decay lengths can be significantly longer. If the barrier height is fixed at ∼3 eV, as might be expected for an alkane chain surfactant, exchange starts to become noticeable only at an edge-to-edge particle spacing of 1.5 nm, and significant for a spacing below 0.5 nm. However, if the barrier height φ is

10: Magnetic Nanoparticles and Applications, Majetich

455

changed by replacing the surfactant with a more conductive matrix, then significant exchange effects are feasible. Exchange interactions are expected to occur over larger length scales under a number of conditions.[155] Near phase transitions, the correlation lengths are large. In a ferromagnetic-to-paramagnetic phase transition, there will still be sizable spin correlations just above the Curie temperature. Near an electronic percolation transition, there will be a significant amount of electron mobility between conducting grains even before the conducting network spans the entire sample.[156] If the matrix contains resonant states, the likelihood of tunneling through these states will be enhanced.[157] Magnetic impurities could be added to the matrix to create localized nearly resonant states. It is even possible to have exchange through an insulating matrix.[157] In a bulk exchange-coupled ferromagnet, the thermal energy at the Curie temperature is comparable to the product of the exchange stiffness A and the moment separation. Using an approximate formula for the exchange stiffness[158] and the edge-to-edge spacing s instead of the moment separation, this model predicts that TCex ∝ μ 2pt s 2 J = μ 2pt exp[ − 2 m* φ s ]

Eq. (10-19)

It should be possible to differentiate exchange and magnetostatic effects in a nanoparticle assembly from the dependence of the ordering or blocking temperature, the type of material between the particles, and the edgeto-edge interparticle separation.

10.3 Applications of Monodomain Magnets 10.3.1 Ferrofluids Ferrofluids are magnetic liquids made of dispersed magnetic nanoparticles that respond collectively to a magnet. When a permanent magnet is brought close to a droplet of a ferrofluid, the whole droplet moves, as would be expected for a magnetic liquid (Fig. 10.7). If the particles move within the droplet and collect near the magnet, then the dispersion is not a true ferrofluid. The particles in a ferrofluid (most commonly iron oxide, with an average core diameter of 6–10 nm) have much higher densities than the surrounding carrier liquid (∼5.4 g/cm3 for Fe3O4 versus slightly less than 1 g/cm3 for most organic solvents used, such as kerosene). For

456

Properties

Figure 10.7 A ferrofluid responds collectively to a magnet, rather than having magnetic particles move separately from the surrounding liquid.[163]

this size range, the nanoparticles will be superparamagnetic, so that the moments will quickly randomize after the removal of an applied field. The most critical feature of magnetic nanoparticles in ferrofluids is the surface coating, which leads the fluid to act like a homogeneous magnetic liquid. For an organic solvent, surfactants such as oleic acid are used to form bonds between the polar head groups and the charged particle surface, while extending the alkane chain tails into the nonpolar solvent. The long alkane chains are important for stabilizing the particles with respect to sedimentation, both because of the entropy associated with different bond orientations,[159–160] and because the surfactant shell reduces the overall particle density and makes it closer to that of the solvent. For aqueous ferrofluids more complex surfactant and counterion combinations are used, so that the outer surfaces of the composite particles are charged and are therefore compatible with the surrounding water.[161–162] In either case, the surface coating reduces the magnetic force between particles, which tends to line up the dipole moments head-to-tail. If large assemblies were formed, Brownian motion would be insufficient to prevent sedimentation. The surface coating also provides a symmetrical repulsive force between particles, due to steric interactions in organicbased ferrofluids, and due to charge repulsion in aqueous ferrofluids. For a stable dispersion of magnetic nanoparticles to act as a ferrofluid, the particle concentration must be sufficiently high that the particles inter-

10: Magnetic Nanoparticles and Applications, Majetich

457

act with each other and start to form chains. Cryo-transmission electron microscopy (cryo-TEM) has been used to show evidence of nanoparticle chains in a ferrofluid that was flash frozen.[164] Typical ferrofluids have ∼5 vol% particles, based on the average core size. With much lower concentrations the particles are too far away from each other to induce collective movement of the fluid droplet. At much higher concentrations, the dispersion becomes unstable. The surface coatings must also prevent the particles from sticking together, so that chains constantly fall apart and reform. In the absence of a magnetic field the chains have random orientations, but if an external field is applied, they align along the field direction. If there is a field gradient, the liquid moves collectively, and the magnetic forces can exceed gravitational forces.[165–166] The applications of ferrofluids[165,167–168] rely on different properties, but the most central is the magnetic force, which can either move the ferrofluid or hold it in place. One of the simplest is a variation of the Bitter method to determine the location of magnetic domain walls. The magnetic fringe field outside a bulk ferromagnetic sample is greatest near domain walls, and its intensity decays rapidly with distance. The field gradient leads to a magnetic force on the monodomain particles in the ferrofluid, which then move over the sample surface until they are next to the domain wall. Ferrofluids can be used as magnetic inks or toners,[169] using magnetic field gradients to direct the particles to desired regions on a surface. Several applications take advantage of ferrofluids to form frictionless seals. Here a ferrofluid in the field of a permanent magnet is used to form an impenetrable barrier. Exclusion seals keep out dust particles and unwanted gases in semiconductor processing chambers, and also in magnetic disc drives. Ferrofluids are also used as liquid O-rings to make rotary vacuum seals. An external magnet holds the ferrofluid in place around a rotating shaft, which can then transfer torque to an object in a vacuum chamber without admitting air. Ferrofluids used in vacuum seals are made with high-boiling-point, low-vapor-pressure solvents. Because a ferrofluid has a higher overall density than most liquids, it can help in dissipating heat. The voice coils of audio speakers experience resistive heating from the driving current. When the coils are immersed in a ferrofluid, which is held in place by permanent magnets, the heat can be carried away more efficiently than by air cooling, and electrical shorts associated with water leaks are avoided. The density of the fluid also modifies the oscillation frequency spectrum, helping to damp out some of the lower frequencies and unwanted resonances. This damping ability is also used in actuators for lens assemblies of CD and DVD players.

458

Properties

Long-term stability is the biggest challenge in making better ferrofluids. The nanoparticles tend to flocculate and sediment, because the surface coating is not stable, especially in aqueous ferrofluids. For exclusion seals there are also needs for ferrofluids that can operate at higher temperatures, or form barriers to more strongly reactive gases.

10.3.2 Biomedical Applications There are three main groups of biomedical applications that use magnetic nanoparticles: (1) magnetic separation and drug delivery, (2) contrast agents for magnetic resonance imaging (MRI), and (3) magnetic hyperthermia in cancer treatment. There have been several excellent reviews discussing the preparation and applications of magnetic nanoparticles in biomedicine.[170–174] Compared with the other applications, magnetic nanoparticles for biomedicine[173–182] have the additional requirement of biocompatibility. Iron oxide nanoparticles are considered to be biocompatible, and in fact the body stores iron in the form of ferritin, which is a 7 nm iron oxyhydroxide nanoparticle encased in a protein shell. Ferritin is superparamagnetic above a few kelvins and does not have a large enough magnetic moment to be useful for applications in its natural form. However, it is possible to remove the cores and replace them with other magnetic materials[71,175] While Co and Ni magnetic nanoparticles are being tested for some particularly interesting applications,[177–178] it remains to be seen whether the toxicity of these materials will make them unsuitable for in vivo functions. Particle mobility is also essential. For biocompatibility, the particles must form a stable dispersion in ∼0.2 M saline solution; this rules out the most stable ferrofluids, which use organic solvents. If the nanoparticles are dispersed directly in liquids, then they must be small to avoid sedimentation. However, if the particles are smaller than ∼70 nm, they may be viewed as foreign by the body, and will be rapidly taken up by cells via endocytosis. To delay nonspecific cell uptake, the particles may be coated with a biologically inert material such as polyethylene glycol (PEG). For separation and hyperthermia applications, the magnetic nanoparticles are usually incorporated into larger composite particles with polystyrene, starch, or dextran. While the composite particles are typically hundreds of nanometers or even micrometers in diameter, because they are formed from low-density materials, they can form stable dispersions in biological media. Furthermore, the polystyrene surface can be conveniently mod-

10: Magnetic Nanoparticles and Applications, Majetich

459

ified to selectively attach selectively binding bioconjugate species. A magnetic nanoparticle for these biomedical applications is therefore a composite containing monodomain magnetic nanoparticles plus surface groups for selective binding to biological species. Detection schemes using magnetic nanoparticles to detect biological species (DNA, cells, proteins, etc.) require specificity and high sensitivity, must be readily modifiable to detect different species, need to differentiate tagged cells from unbound particles, and must be robust, portable, and mass producible. Through molecular recognition, streptavidin and biotin bioconjugates will bind to each other, but not to other molecules. To demonstrate binding specificity, a surface is often coated with streptavidin, and then exposed to a liquid dispersion containing biotinylated composite particles. Other types of bioconjugates include single-strand DNA, monoclonal antibodies and their antigens, and synthetic glycoproteins and distinct types of saccharides at cell surfaces.[183] A modification of the magnetically guided drug delivery is the use of composite magnetic particles with radionuclides.[184] Most magnetic separation strategies use the composite magnetic particles in solution with mixtures of biological species.[185–187] After selective reaction is allowed to occur, the bound species are separated magnetically. When biological assays such as ELISA (enzyme-linked immunosorbent assay) are used for detection, it is immaterial how many bound species   are on a given particle. Here the force Fm = (μ pt ⋅ ∇) B must  magnetic  F = 6πη Δ ν exceed the viscous drag force , where η is the solvent visd  cosity and Δν is the difference in velocities of the solvent and the particle plus its bound species. In magnetically guided drug delivery,[188] field gradients are used to concentrate composite particles containing a drug in a localized area. The advantage of this approach is that drug side-effects could be minimized, since the drug would not be distributed throughout the body. Anticancer treatments are now being tested using attached radioisotopes,[184] and chemotherapy drugs.[189] One type uses hyperthermia to release the drug from a hydrogel surrounding the particles.[190] There are ongoing clinical trials in humans using magnetic drug delivery.[171] In in-vitro studies, magnetically guided separation techniques have been able to detect single composite particles in vitro.[191–192] Magnetoelectronic detection has also been used for biosensing.[193–197] Here giant magnetoresistance spin valve devices have been able to detect single micrometer-sized composite particles,[193] but detection of single monodomain nanoparticles has not yet been achieved.

Properties

460

Magnetic nanoparticles have been used for magnetic heating of cells for hyperthermic cancer treatment.[198–203] For treatment of humans without introducing adverse side-effects, the excitation frequency can range from 50 kHz to 1.2 MHz, and the magnetic field amplitude must be below 200 Oe.[171] AC magnetic field heating depends on the power absorbed from the field and dissipated by the magnetic nanoparticles. With sufficient heating, the temperature can be raised near the particles, which makes cancer cells more susceptible to radiation or chemotherapy (hyperthermia, >43°C) or kills them directly (thermal ablation, >50°C). There are two mechanisms by which energy can be transferred to a material via an AC magnetic field and dissipated as heat. In any conducting material, eddy currents are induced at high frequencies, and the power loss will be proportional to f 2. For the iron oxide nanoparticles studied for hyperthermia, this is not a significant factor, and hysteresis losses will dominate. Although ferromagnetic materials would potentially have larger hysteresis losses than superparamagnets, in the allowed field range they cannot be saturated. In practice, superparamagnetic particles currently have greater power losses, PSP = μ0πfχ″H 2

Eq. (10-20)

where μ0 is the permittivity of free space, f is the frequency, is defined as in Eq. (10-14), and H is the magnetic field amplitude.[172] Optimum power dissipation is predicted for 14 nm magnetite particles, but unfortunately stable ferrofluids cannot be formed with this particle size.[172] The metric used to compare the effectiveness of magnetic fluid heating is the specific absorption rate (SAR) of the magnetic field power, which is proportional to the predicted change in temperature ΔT: P=

( SAR)m V

=

cΔT Δt

Eq. (10-21)

Here m is the mass of Fe, V is the sample volume, c is the average specific heat of the sample, and Δt is the time over which the power P is dissipated. From Eq. (10-20), the power delivered is expected to depend on the frequency and applied field amplitude; SAR values of 180 W/g are feasible[200] within the safe field and frequency limits. Nuclear magnetic resonance (NMR) involves nearly resonant precession of nuclear magnetic moments (1H nuclei, or protons, in water molecules for contrast reduction). Magnetic nanoparticles can have a strong influence on the resonant frequency and the rate of magnetization decay

10: Magnetic Nanoparticles and Applications, Majetich

461

if the nuclei are within a diffusion length of the particle during the measurement time. The inhomogeneous field generated by the particles affects both the spin–lattice relaxation time T1 and the spin–spin relaxation or dephasing time T2 of the protons. The effectiveness of a particular contrast agent is gauged by the value of its relaxivity, R, which depends on the Fe molar concentration.[204] There are separate relaxivities for each relaxation time, and R1 is defined by the formula 1 1 = + R1C T1 T1,0

Eq. (10-22)

where T1,0 is the value of T1 with no contrast agent, C is the Fe molar concentration, and R has typical units of mmol−1 s−1. R2 = 1/T2 and has units of s−1. The addition of an iron oxide contrast agent can change the value of T1 by a factor of over 3000, and can change T2 by a factor of over 70.[205] Currently a paramagnetic gadolinium chelate compound known as GdDTPA (gadolinium diethylenetriaminepentaacetic acid) is most commonly used as a contrast agent. Here each molecule has a magnetic moment of 7 μB. In comparison, a ferrimagnetic iron oxide nanoparticle may have a moment up to one hundred times as large, and will therefore create a stronger and longer-range inhomogeneous magnetic field. Here composite particles are not used; the iron oxide nanoparticles ∼10 nm in diameter are broken down by the body and excreted through the liver. A type of iron oxide nanoparticle coated with dextran has recently been approved by the US Food and Drug Administration for use in humans.[171] In addition to general use in MRI imaging.[206] these magnetic nanoparticles have been used to detect specific cells and their movement.[207–215] The practical limit of magnetic nanoparticle detection based on T *2 measurements can be estimated using the results obtained in prior experiments with NMR micro-coils.[216] Lauterbur and co-workers estimated[217] that a single magnetic particle is detectable in a volume 50 times its diameter. A technique for detecting the resonance signal from the particles themselves, as opposed to that from protons in the surrounding molecules, has recently been reported.[218] This method will not only increase the MRI signal strength, but may enable more sophisticated in-vivo tracing experiments using particles functionalized for specific binding affinities.

10.3.3 Imaging with Magnetic Nanoparticles The MRI technique has also been extended for use in non-biomedical imaging in a technique known as magnetic resonance force microscopy

462

Properties

(MRFM). Here, just as in MRI, a region of the sample is brought into resonance through the combination of a large DC magnetic field and a small AC magnetic field generated by a coil.[219] If the sample is slightly offresonance, a small region may be brought into resonance by placing a magnetic nanoparticle nearby.[220–223] If the particle is attached to a scanning probe tip like that used in atomic force microscopy (AFM), then this technique can be used to study the behavior of flat surfaces. MFRM has been used to detect single spin defects in Si surfaces, and it therefore has the potential to be far more sensitive than magnetic force microscopy (MFM), which has an optimal resolution of ∼30 nm and a typical resolution of ∼100 nm.

10.3.4 Data Storage Media Magnetic tapes and floppy discs are made with thin films of needleshaped magnetic nanoparticles embedded in a polymer matrix. The typical lengths of the particles are several hundred nanometers, with aspect ratios on the order of 10 : 1. Originally the particles were made of γ-iron oxide, but more recent media have used metallic particles. While these particles exceed the monodomain threshold, dcr, based on the magnetocrystalline anisotropy, their anisotropic shape strongly favors magnetization along the long direction. Although magnetic tape is a relatively mature technology, it remains an economical and reliable method for storing large quantities of information, particularly information that does not need to be recorded or retrieved quickly. Magnetic hard disc media, in contrast, have been developing at breakneck speed.[224–226] The growth in the number of bits per square inch has until recently followed or even exceeded Moore’s law, according to which the device density doubles every 18 months. The current state-of-the-art density is on the order of 130 Gbits/in2. This medium is not made of nanoparticles but of a CoCrPt alloy film with an average grain size of ∼8.5 nm. The grain boundaries are Cr-rich, which helps to decouple the moments of neighboring grains. The grains are Pt-rich, with alternating atomic layers of Co and Pt. The medium is prepared with preferential orientation of these layers, which leads to high magnetocrystalline anisotropy and high coercivities. While the original paper describing the superparamagnetic limit stipulated a threshold of KV/kT ≈ 40 for usable media,[227] values on the order of 60 are being investigated for the future.[225] The materials with the highest magnetocrystalline anisotropy include Fe and Co platinum alloys, and rare-earth transition metal alloys such as SmCo5.

10: Magnetic Nanoparticles and Applications, Majetich

463

The latter group of alloys is not currently under investigation because of their extreme oxidation sensitivity. In FePt, bulk values of K predict that a 2.5 nm particle would be magnetically stable long enough to be used in recording media. If larger particles of bulk-like FePt were used, it would be difficult to obtain the magnetic fields of 10 kOe or greater from the write head that are needed to switch the magnetic bits. Ideally, a bit would consist of a single grain or particle, and all of the bits on the disc would have identical switching fields and magnetic stability. Variations in size, shape, crystallographic orientation, and chemical composition between grains or particles make them magnetically nonuniform, which is a major source of noise. For reasonable signal-tonoise levels, with noise attenuation of less than 20 dB, current media use the combined magnetic signal of ∼100 grains. In the near term, magnetic thin-film recording media will shift from longitudinal to perpendicular magnetization orientation, relative to the plane of the disc. The storage layer will be on top of a soft magnetic underlayer, so that there is a flux closure pathway going from the write head through the bit, through the underlayer, and back to the far side of the write head pole piece. The flux that is part of the magnetic circuit will be greater than the leakage flux of the fringe field from a bit magnetized inplane, enabling higher storage densities. There is currently a considerable amount of research on sputtered nanocomposite media for magnetic recording.[228–230] Here we will focus on work directed toward future media in which nanopatterning is used to decouple the particles from each other, and where the uniformity is sufficient for a single particle to be used to store a bit of information. While not at the level of commercial production, there has been intense interest in the possibility of patterned media, in which arrays of monodomain particles could be used for single-particle-per-bit recording. The wavelength-dependent diffraction limit prevents the extension of current lithographic methods to pattern features smaller than ∼100 nm. A variety of different techniques for making patterned media are currently being explored. Electron beam lithography has been used to pattern arrays of 35 nm holes.[231] To do this, a polymer resist coating is exposed in certain areas to a focused electron beam, which breaks chemical bonds and makes the residual material soluble, so that it can be washed away (Fig. 10.8(a)). The holes are then filled with Ni, forming pillars that are magnetized perpendicular to the plane of the substrate due to their shape anisotropy. A density of 65 Gbits/in2 has been demonstrated, and 400 Gbits/in2 is possible with this approach. Numerous groups have a similar lithographic

464

Properties

Figure 10.8 Schematic of different self-assembly techniques. (a) Electron beam lithography, in which a focused electron beam damages parts of a polymer resist, which is then removed. (b) Nanoimprint lithography, in which a mold is pressed into a thermoset polymer to transfer a pattern. (c) Anodic alumina templates, where aluminum is etched in acid to create an ordered network of cylindrical pores. (d) Block co-polymer templates, in which thermally-induced phase separation generates a hexagonal array of cylindrical regions rich in a UV-sensitive polymer, which is then removed. In (a)–(d) the magnetic material is introduced by a final electrodeposition stage. (e) In self-assembling nanoparticle arrays, the magnetic particles are preformed and pack together because of their uniform size and their surfactant coating.

10: Magnetic Nanoparticles and Applications, Majetich

465

approach to prepare samples and study the fundamental physics of interaction effects between particles,[232–233] domain structures,[234] and the distribution of switching fields.[235] Because the marks are written serially by the focused electron beam, such an approach would not be suitable for mass production of data storage media. Other serial writing processes such as focused ion beam lithography and scanning tunneling microscopebased lithography[236] share this limitation, and have similar strengths. They can be used to generate patterns with more complex shapes than the other approaches, and the patterns can be written in a square lattice or in an arbitrary sequence, if desired. One possible extension of these lithographic techniques that could enable large-scale production is nanoimprint lithography (Fig. 10.8(b)).[237] Here a master formed by electron beam writing is used as a mold for a polymer. The sample is pressurized above the glass transition temperature, then cooled, and the mold is removed. Long-range ordering over disc-sized dimensions can be achieved.[238] This technique has not yet been used to make magnetic patterned media, possibly because of variations in the shapes of the patterns formed, which would lead to a range of switching fields. There has been considerable interest in self-assembly methods because they are faster and of lower cost. Important issues that remain to be resolved depend on the synthetic approach. In all cases, it is possible to achieve “single-grain” periodic nanostructures over a length scale of a micrometer, but none is perfect over the millimeter scale. This would require either the use of multiple particles per bit, or a more refined servomechanism to correct for abrupt shifts in the particle positions across a dislocation or grain boundary. A smooth piece of aluminum can be anodized in strong acid to form an array of hexagonal pores lined with a thin (∼10 nm) Al2O3 layer (Fig. 108(c)).[239–240] Electrodeposition can be used to fill the pores with a magnetic material such as Co, Ni, or Fe.[241–242] The diameter of the nanowires can be tuned between 5 and 200 nm. The length scale of ordering is on the order of 1.5 μm[241] and the potential bit density is 65 Gbits/in2. Similar filling methods have been used to grow magnetic nanowires in the pores created by etching ion tracks in polycarbonate membranes,[243–244] but these wires do not form ordered arrays and have much lower densities. An alternative to anodic alumina is to use a phase-segregating block copolymer to form a hexagonal pore network (Fig. 10.8(d)). Here a very monodisperse copolymer of polystyrene (PS) and poly(methyl methacrylate) (PMMA) is coated onto a substrate and heated above the polymer glass transition temperatures to phase segregate the polymer blocks, forming ordered cylinders of PMMA in a PS matrix. The minimum

466

Properties

diameter of the cylinders is 25–30 nm and the minimum spacing is ∼45 nm.[245–247] The cylinders can be oriented perpendicular to the substrate by electric fields or, in the case of thin films (∼40 nm), they will orient themselves spontaneously.[248] Exposure to ultraviolet light degrades the PMMA, making it soluble, and cross-links the PS, making it more robust. Electrodeposition can be used to fill the cylindrical pores left behind after the PMMA is removed. Because the pores are hydrophobic, the electrolytic bath contains methanol as well as water. Co nanopillar arrays with a bit density of 1.25 Tbit/in2 have been prepared in this manner. The use of block copolymers with better etch selectivity[249–253] and improved casting techniques[254] has shown promise in developing this technique for making usable recording media.[255] An array of oriented nanopillars can have a large areal density of bits. The diameter of each particle can be small without being superparamagnetic because of shape anisotropy. At high densities, however, the moments of particles in adjacent pores are strongly interacting.[241] For data storage media based on anodic alumina, either the pore separation would have to be increased, or a material with lower magnetization would have to be used in filling, so that particles could be switched independently. The smallest particle size (3 nm) and spacing (7 nm) have been achieved with surfactant-coated FePt nanoparticles that self-assemble into arrays (Fig. 10.8(e)).[256–270] This would correspond to a single-particle-per-bit density of 6.6 Tbits/in2. Numerous groups have investigated gas-phase high-vacuum methods to grow FePt nanoparticles,[271–273] but these do not have the same degree of monodispersity and do not self-assemble into arrays. Monodisperse particles have been grown in a cluster beam, and show short range ordering when deposited on a surfactant-coated surface.[274] However, the production rate for cluster beam sources is too low to be useful in manufacturing. The surfactant-coated nanoparticle arrays have a serious disadvantage, relative to the other techniques: they must be annealed to transform the low-coercivity face-centered cubic or A1 phase into the high coercivity face-centered tetragonal or L10 phase. The high annealing temperature decomposes the surfactant and the particles sinter together, destroying the order. Sintering can be prevented,[263, 266] but the phase transformation rate is extremely slow, since a separate nucleation event must occur in each particle.[267] Progress has been made in directly synthesizing FePt-based particles that have high coercivity after reduced annealing temperatures.[261, 264] Although sufficient monodispersity for highly ordered arrays has not yet been achieved, this work offers a potential solution.

10: Magnetic Nanoparticles and Applications, Majetich

467

Orientation is another requirement for magnetic uniformity, and here again the various self-assembly approaches differ. For perpendicular media, bits should have magnetization directions exactly perpendicular to the plane of the substrate, either “spin-up” or “spin-down.” The filled anodized alumina templates, where shape anisotropy dominates magnetocrystalline anisotropy, offer the highest degree of orientation. Other approaches rely on magnetocrystalline anisotropy for high coercivity, but at the same time there is no reason for preferential crystallographic orientation based on self-assembly. Self-assembly in an applied magnetic field has shown modest effects,[261, 275] but significant improvements are still needed.

10.3.5 Magnetoresistive Devices Today’s commercial electronics technology is based on the charge on an electron, and the ability to manipulate it. In the rapidly growing field of spintronics, the spin of an electron is manipulated in nanoscale magnetic materials. Spintronic devices are now in the read heads of computer disc drives, and they have the potential to be used as magnetic random access memory (MRAM) as well. These devices, and indeed much of the research on spintronics is based on patterned thin-film structures that may be nanoscale only in their thickness. There have been a number of excellent reviews in this area.[276–277] The subject of spintronics is included in this review because the underlying physics is based on single magnetic domains, and because future research will likely focus on smaller structures. The electron spin angular momentum gives rise to the atomic magnetic moment and—in materials like Fe, Co, and Ni—to ferromagnetism. Atoms within a domain of a ferromagnet are coupled by an exchange field,  H ex, and electrical transport through a ferromagnet is different for “spinup” and “spin-down” electrons. According to the rigid band model of a ferromagnet, the density of states for spin-up and spin-down electrons is the same, except for an offset by the exchange energy (Fig. 10.9). A material is ferromagnetic when the Fermi energy intersects at least one of these offset bands, because there are more occupied states with one type of electrons than the other. A half-metallic ferromagnet such as Fe3O4 has the Fermi energy below the threshold of the minority spin band, and so it can ideally have a 100% spin-polarized current. In comparison, the spin polarization in Fe and Co is or the order of 40%.[278] Consider the effect of the band structure on the electrical transport between a diamagnetic metal such as Cu, and a ferromagnet such as Co.

468

Properties

Figure 10.9 Schematic of the density of states (DOS) for (a) a nonferromagnetic metal, where there are equal numbers of spin-up and spin-down electrons, and therefore there is no net magnetization, and for (b) a ferromagnetic metal, where the spin-up and spin-down bands are split by the exchange energy Eex, and the excess of one type of electron leads to a net magnetization. EF is the Fermi energy.

In the Cu there are equal numbers of spin-up and spin-down electrons, and their density of states is identical. Suppose that the Co is monodomain and magnetized so that spin-up electrons have lower energy. Electrons can only pass through if there are available empty states, and the density of spin-up states will be greater than the density of spin-down states. The electrical conductivity of spin-up electrons through the Cu—Co junction will be greater than that for spin-down electrons. If the magnetization direction is reversed, spin-down electrons have a higher conductivity channel. The spin-flip diffusion length in Co is 5.5 nm for majority carriers, which have spins parallel to the magnetization, but only 0.6 nm for minority carriers.[279] In semiconductors the spin diffusion length can be micrometers, which is one of the reasons why magnetic semiconductors are of great interest for spintronics. Transport through a multidomain magnet will tend to average out the spin-dependent effects, but a monodomain ferromagnetic particle will act as a spin filter, enabling the preparation of spin currents for spin-dependent electronics, or “spintronics.” Giant magnetoresistance (GMR) was first reported in Fe/Cr multilayer thin films.[280] If the magnetization directions of the ferromagnetic layers are parallel, then one type of electron will have high conductivity through both monodomain ferromagnetic layers (Fig. 10.10). Just as with a large and a small resistor in parallel, there is overall a low resistance. If the magnetization directions are antiparallel, then one layer will have low conductivity for spin-up electrons and the other for spin-down. Overall both spin-up and spin-down channels will have moderate resistance. The

10: Magnetic Nanoparticles and Applications, Majetich

469

Figure 10.10 Differences in scattering of electron spins, depending on the magnetization of the material they pass through. In (a) an unpolarized current passes from left to right through a metal trilayer. The first and third layers are ferromagnetic (FM) and have upward magnetization, and the middle layer is nonmagnetic (NM). Electrons with a spin that is opposite to the magnetization direction are preferentially reflected at the NM–FM interfaces and scattered within the ferromagnet. The net effect of passing through the multilayer is to generate a polarized spin current. Here spin-up electrons have low resistance while spin-down electrons have high resistance. In (b) the FM layers have opposite magnetization directions. Spin-up and spin-down electrons are preferentially transmitted through different layers, but the net effect is that the current does not become polarized, and both spin-up and spin-down channels have moderate resistance.

magnetoresistance is the relative difference in electrical resistance with (R(H)) and without (R(0)) an applied field, MR( H ) =

R( H ) − R(0) × 100% R(0)

Eq. (10-23)

In bulk materials, the largest magnetoresistance is on the order of 2%, but in nanostructured materials with monodomain ferromagnets, it can exceed 25% at room temperature. In the GMR read heads and magnetic field sensors it is more typically 5–10%.[277] Control of the domain structure in magnetic thin films and minimization of nonmagnetic scattering at the

470

Properties

interfaces have been important in fabricating devices based on the GMR effect. This is now done successfully, and the read heads of magnetic disc drives now contain GMR multilayers, which are several times more sensitive than the read heads used previously. The first studies of magnetoresistance in nanoparticles were reported in co-sputtered AgCo thin films with a granular structure.[146–147] A giant Hall effect[281–282] and a giant magnetoimpedance effect at high frequency[283] have been observed in related materials. In thermodynamic equilibrium, Ag and Co are immiscible. When the mixture of Ag and Co made by sputtering is annealed, the elements phase-segregate, and with the right composition this leads to Co-rich nanoparticles in a silver-rich matrix. The average particle size and composition change with the annealing time. In the absence of an applied field, the particle moments are disordered, and there is considerable scattering of both spin-up and spin-down electrons. When the applied field aligns the particle moments, there is a low resistance channel for one type of spins, and the overall resistance drops. Maximum magnetoresistance is observed in composites just below the magnetic percolation threshold, at which they become ferromagnetic. With larger or more strongly interacting superparamagnetic nanoparticles, the magnetization curve described by Eq. (10-8) becomes sharper and the particles align at lower fields. These granular GMR materials required higher fields than the thin films for maximum magnetoresistance. GMR read heads are complex multilayer structures designed so that they switch from a low to a high resistance state at ∼10–30 Oe. Thin-film growth technology is far more advanced than nanoscale synthesis, and it is difficult to make such complex architectures on the nanoscale in all three dimensions. It is also a nontrivial task to perform more controlled electrical measurements on small numbers of magnetic nanoparticles. Anodized alumina templates and related polycarbonate track-etched membranes already have one electrode attached to the particles for the electrolytic deposition used in filling, and they were the first systems studied, showing similar GMR behavior to that of the thin films.[243–244] In self-assembled nanoparticle arrays, the organic surfactant coating leads to very high electrical resistance, but with low-temperature annealing, the particle separation can be reduced sufficiently that electron hopping between particles has been observed.[284–287] The charging energy associated with the hopping barrier is on the order of 10 meV, so that nonlinear I–V curves are observed only at cryogenic temperatures, where kT is less than the charging energy. While the degree of spin polarization may be reduced near the surface or interface of a ferromagnet,[288] evidence of some spin polarization at low

10: Magnetic Nanoparticles and Applications, Majetich

471

temperature was found for both 10 nm Co[284] and 5.5 nm Fe3O4[287] nanoparticles. A breakthrough in the characterization of single particles was made on lithographically patterned thin-film multilayers.[289–293] Here the multilayer stack was ∼130 nm across and contained a thicker, magnetically “pinned” Co layer separated by a Cu spacer from a thinner “free” Co layer. The thickness of the layers is much less than the spin diffusion length for these materials. At extremely high current densities (107–108 A/cm2), the current itself can switch the magnetization direction of the free layer. The interaction mechanism is itinerant exchange between the conduction electrons and the 3d electrons of the ferromagnet. While a current generates a magnetic field, it is not the self-field alone that reverses the magnetization direction, because the magnetization–current hysteresis loop is asymmetric. The experimental results are interpreted in terms of a spin torque exerted by the electrons on the magnetic moment of the free layer. The magnetic switching mechanism exploited here is due to spin transfer,[294–295] illustrated in Fig. 10.11. The Landau–Lifschitz–Gilbert equation given in Eq. (10-13) must be modified to include a driving term from the spin current density,     α  dM dM P0 J   ˆ = − γM × H + γ M × M0 × M + M × eMδ M dt dt

Eq. (10-24)

where P0 is the polarization of the current, J is the current  density, e is the unit charge, δ is the thickness of the free layer, and M o is the effective magnetization of the spin current.[296] When a current flows from the free layer to the pinned layer, the electrons flow in the opposite direction. The conduction electrons passing through the pinned layer become spin-polarized parallel to the direction of its magnetization. When this spin-polarized current reaches the free layer, it creates a torque on the local magnetic moment. If the spinpolarized current density exceeds the switching threshold of the free layer, the magnetization will switch to lie parallel to that of the pinned layer. If the current direction is reversed, even if the spin current is unpolarized and the free layer is small, preferential reflection of electron spins antiparallel to the magnetization of the reference layer will deliver a spin torque. The current becomes spin-polarized through transfer of angular momentum from the free layer. If the current density is high enough, the free layer magnetization can be switched from parallel to antiparallel. These differences make the critical current densities for switching asymmetric.

472

Properties

Figure 10.11 Schematic showing how the spin torque of electrons can rotate the magnetic moment of a nanomagnet. In (a), current flows to the left, so electrons move toward the right. They are incident on a trilayer consisting of a thick layer of a ferromagnetic metal (the pinned layer), then a layer of a nonferromagnetic metal, and then a thin layer of a ferromagnetic metal (the free layer). If the pinned layer has a magnetization pointing upward, then spin-down electrons will be preferentially reflected at the interface, and the current becomes spin-polarized before entering the free layer. Conduction electrons can transfer angular momentum to the ferromagnet through spin-flip scattering; for a large amount of angular momentum transfer in a small magnet, this can be sufficient to switch the magnetization direction. During the transfer the free layer magnetization direction rotates and the current becomes less spin-polarized. In (b) the current flows in the opposite direction, so that an unpolarized current is incident first on the free layer.

Spin torque transfer changes the direction of the atomic moments in the ferromagnet, which can lead either to switching (complete magnetization reversal), or to the excitation of spin waves, which have frequencies in the microwave region. A DC current through a nanopillar has been demonstrated to generate microwaves,[289] and current through a spin valve can excite 1/f noise and create spin waves.[296] As the size of a monodomain magnet becomes smaller, the maximum wavelength for a spin wave is reduced. Theories have predicted the spin wave spectrum in spherical particles,[297] and spin wave quantization effects have been observed in lith-

10: Magnetic Nanoparticles and Applications, Majetich

473

ographically patterned one-dimensional nanomagnets.[298] Nanomagnets may someday be used to generate microwaves in specific frequency ranges. While GMR has been investigated in metallic multilayers, tunneling magnetoresistance (TMR)[299–300] has been studied in pairs of ferromagnetic metal layers separated by a very thin (<1 nm) insulating spacer, typically AlOx. The oxide barrier greatly increases the resistance of the multilayer overall, and 20–50% changes are observed.[277, 301–302] Thin-film TMR structures are soon to be used in read heads for magnetic disc drives. They are also currently being developed for use in magnetic random access memory (MRAM).[303–304] In a memory bit, the TMR junction is connected to a pair of wires that are perpendicular to the junction axis and to each other. The relative magnetization directions of the pinned and free layers are “read” by measuring the resistance between the two wires across the barrier. The bit is “written” by passing a large current through the wire in contact with the free layer; the direction of the current determines the direction of the magnetic field in the free layer. MRAM has potential advantages over flash memory and electrically erasable programmable read only memory (EEPROM) because it will not wear out after multiple write cycles and it requires less energy for writing.[277] There has so far been little research on TMR in magnetic nanoparticles, but this will likely change in the future. For MRAM-like behavior, the particles would have to be large enough, or the temperature low enough, to avoid superparamagnetic behavior. Spin torque effects would be difficult to observe because the high resistance would make it impossible to achieve the high current densities required.

10.4 Conclusions There are many exciting new developments in the field of monodomain magnetic nanoparticles, concerning rich and subtle physics and important applications. With the ability to prepare highly monodisperse particles, more details concerning size- and surface-dependent effects are being revealed. The ability to stabilize magnetic nanoparticles in liquids and to control their motion is starting to have a significant impact in biomedicine, which is likely to expand in the future. Magnetic data storage is rapidly approaching the superparamagnetic limit, and will require a paradigm shift to patterned media to continue increasing the information storage density. Electronically connected monodomain “particles” with large magnetoresistance have transformed from a laboratory demonstration

474

Properties

to a component in every magnetic disc drive in less than 15 years. Further advances may lead to magnetic RAM and perhaps to magnetic transistors as well. We have focused here on nanoparticles based on traditional ferromagnetic materials, but the next nanoparticles of interest may be made from ferromagnetic semiconductors for spintronic application, or of multiferroics, where an applied electric field can switch the magnetization direction in an insulating material.

Acknowledgments Support from the National Science Foundation through grants CTS0227645 and ECS-0304453 is gratefully acknowledged.

References 1. Dennis, C.L., Borges, R.P., Buda, L.D., Ebels, U., Gregg, J.F., Hehn, M., Jougelet, E., Ounadjela, K., Petej, I., Prejbeanu, I.L., and Thornton, M.J., J. Phys. C: Condens. Matter., 14:R1175-R1262 (2002). 2. Martin, J.L., Nogues, J., Liu, K., Vincent, J.L., and Schuller, I.K., J. Magn. Magn. Mater., 256:449–501 (2003). 3. Fruchart, O., C.R. Physique, 6:61–73 (2005). 4. Himpsel, F.J., Ortega, J.E., Mankey, G.J., and Willis, R.F., Adv. Phys., 47:511–597 (1998). 5. Leslie-Pelecky, D.L., and Rieke, R.D., Chem. Mater., 8:1770 (1996). 6. Skomski, R., J. Phys.: Condens. Matter, 15:R841-R896 (2003). 7. Willard, M.A., Kurihara, L.K., Carpenter, E.E., Calvin, S., and Harris, V.G., Int. Mater. Rev., 49:125–170 (2004). 8. Moeser, G.D., Roach, K.A., Green, W.H., Laibnis, P.E., and Hatton, T.A., Ind. Eng. Chem. Res., 41:4739–3749 (2002). 9. Ponder, S.M., Darab, J.G., Bucher, J., Caulder, D., Craig, I., Davis, L., Edelstein, N., Lukens, W., Nitsche, H., Rao, L., Shuh, D.K., and Mallouk, T.E., Chem. Mater., 13:479–486 (2001). 10. Elliott, D.W., and Zhang, W.-X., Environ. Sci. Technol., 35:3922–4926 (2001). 11. Liu, Y., Majetich, S.A., Tilton, R.D., Sholl, D.S., and Lowry, G.V., Environ. Sci. Technol., 39:1338–1345 (2005). 12. Phulé, P.P., and Ginder, J.M., Int. J. Mod. Physics B, 13:2019–2027 (1999). 13. Poddar, P., Wilson, J.L., Srikanth, H., Yoo, Y.-H., Wereley, N.M., Kotha, S., Barghouty, L., and Radhakrishnan, R., J. Nanosci. Nanotechnol., 4:192 (2004). 14. Makino, A., Inoue, A., Hatani, T., and Bitoh, T., Mater. Sci. Forum, 235–238:723–728 (1997). 15. Herzer, G., IEEE Trans. Magn., 25:3327 (1989).

10: Magnetic Nanoparticles and Applications, Majetich

475

16. Yoshizawa, Y., Oguma, S., and Yamauchi, K., J. Appl. Phys., 64:6044 (1988). 17. Skomski, R., and D. Coey, J.M., Phys. Rev. B, 48:15812–15816 (1993). 18. Zeng, H., Li, J., Liu, J.P., Wang, Z.L., and Sun, S., Nature, 420:398–398 (2002). 19. Xu, X.L., Majetich, S.A., and Asher, S.A., J. Am. Chem. Soc., 124:13864–13868 (2002). 20. Xu, X., Friedman, G., Humfeld, K.D., Majetich, S.A., and Asher, S.A., Adv. Mater., 13:1681 (2001). 21. Mamedov, A., and Kotov, N.A., Langmuir, 16:5530–5533 (2000). 22. Aliev, F.G., Correa-Duarte, M.A., Ostrander, J.W., Giesberg, M., LizMarzan, L.M., and Kotov, N.A., Adv. Mater., 11:1006–1010 (1999). 23. Gates, B., and Xia, Y., Adv. Mater., 13:1605–1608 (2001). 24. McMichael, R.D., Shull, R.D., Swartzentruber, L.J., Bennett, L.H., and Watson, R.E., J. Magn. Magn. Mater., 111:29–33 (1992). 25. Pecharsky, V.K., and Gschneidner, K.A., Phys. Rev. B, 78:4494–4497 (1997). 26. Provenzano, V., Shapiro, A.J., and Shull, R.D., Nature, 429:853–857 (2004). 27. McCurrie, R.A., Ferromagnetic Materials. Structure and Properties, p. 251, Academic Press San Diego (1994). 28. Bucher, J.P., in: Quantum Phenomena in Clusters and Nanostructures (S.N. Khanna, and A.W. Castleman, Jr., eds.), pp. 83–137, Springer, Berlin (2003). 29. Liu, F., Khanna, S.N., and Jena, P., J. Appl. Phys., 67:4484 (1990). 30. Gatteschi, D., Caneschi, A., Pardi, L., and Sessoli, R., Science, 265:1054 (1994). 31. Billas, I.M.L., Chatelain, A., and de Heer, W.A., Science, 265:1682–1684 (1994). 32. Apsel, S.E., Emmert, J., Deng, J., and Bloomfield, L., Phys. Rev. Lett., 76:1441–1444 (1996). 33. Gambardella, P., Rusponi, S., Veronese, M., Dhesi, S.S., Grazioli, C., Dallmeyer, A., Cabria, I., Zeller, R., Diederichs, P.H., Kern, K., Carbone, C., and Brune, H., Science, 200:1130–1133 (2003). 34. Friedman, J.R., Sarachik, M.P., Tejada, J., Maciejewski, J., and Ziolo, R., J. Appl. Phys., 79:6031 (1996). 35. Mertes, K.M., Suzuki, Y., Sarachik, M.P., Paltiel, Y., Shtrikman, H., Zeldov, E., Rumberger, E., Hendrickson, D.N., and Christou, G., Phys. Rev. Lett., 87:227205 (2001). 36. del Barco, E., Kent, A.D., Rumberger, E.M., Hendrickson, D.N., and Christou, G., Phys. Rev. Lett., 91:047203 (2003). 37. Wernsdorfer, W., and Sessoli, R., Science, 284:133–135 (1999). 38. Chudnovsky, E.M., Science, 274:938–939 (1996). 39. Leuenberger, M.N., and Loss, D., Nature, 410:789–793 (2001). 40. Frenkel, J., and Dorfman, J., Nature, 126:274 (1930). 41. McCurrie, R.A., Ferromagnetic Materials. Structure and Properties, pp. 13, 232, Academic Press, San Diego (1994). 42. Luborsky, F.E., J. Appl. Phys., 32:171S (1961).

476

Properties

43. Berkowitz, A.E., Science and Technology of Nanostructured Materials, vol. 11, Plenum Press, New York (1991). 44. Zhang, D., Klabunde, K.J., Sorensen, C.M., and Hadjipanayis, G.C., Phys. Rev. B, 58:14167 (1998). 45. Kachkachi, H., Nogues, M., Tronc, E., and Garanin, D., J. Magn. Magn. Mater., 221:158 (2000). 46. Christou, G., Gatteschi, D., Hendrickson, D.N., and Sessoli, R., MRS Bull., 25:66 (2000). 47. Suber, L., Zysler, R., Santiago, A.G., Fiorani, D., Angiolini, M., Montone, A., and Dormann, J.L., Mater. Sci. Forum, 269–272:937 (1998). 48. Farrell, D., Majetich, S.A., and Wilcoxon, J.P., J. Phys. Chem. B, 107:11022 (2003). 49. Stoner, E.C., and Wohlfarth, E.P., Trans. Roy. Soc (London) A, 240:599–642 (1948). 50. Jacobs, I.S., and Bean, C.P., in: Magnetism (G.T. Rado, and H. Suhl, eds.), 3:271, Academic Press, New York (1963). 51. Kneller, E., in: Magnetism and Metallurgy (A.E. Berkowitz, and E. Kneller, eds.), 1:365–371, Academic Press, New York (1969). 52. Néel, L., J. de Physique et le Radium, 20:215–221 (1959). 53. Sharrock, M.P., IEEE Trans. Magn., 26:193–197 (1990). 54. Néel, L., Ann. Géophys., 5:99–136 (1949). 55. Bean, C.P., and Jacobs, I.S., J. Appl. Phys., 27:1448 (1956). 56. Bean, C.P., and Livingston, J.D., J. Appl. Phys., 30:120S (1959). 57. McHenry, M.E., Majetich, S.A., De Graef, M., Artman, J.O., and Staley, S.W., Phys. Rev. B, 49:11358 (1994). 58. Chantrell, R.W., Poppelwell, J., and Charles, S.W., IEEE Trans. Magn., 14:975 (1978). 59. Brown, W.F., Jr., Phys. Rev., 130:1677 (1963). 60. Brown, W.F., Jr., Micromagnetics, Wiley-Interscience, New York (1963). 61. Wernsdorfer, W., Orozco, E.B., Hasselbach, K., Benoit, A., Barbara, B., Demoncy, N., Loiseau, A., Pascard, H., and Mailly, D., Phys. Rev. Lett., 78:1791 (1997). 62. Wernsdorfer, W., in: Advances in Chemical Physics (I. Progogine, and S.A. Rice, eds.), 118:99–190 (2001). 63. Street, R., and Woolley, J.C., Proc. Roy. Soc. A, 62:562–572 (1949). 64. Chamberlin, R.V., and Scheinfein, M.R., Science, 260:1098–1101 (1993). 65. Mallinson, J.C., IEEE Trans. Magn., 23:2003 (1987). 66. Chudnovsky, E.M., and Gunther, L., Phys. Rev. Lett., 60:661–664 (1988). 67. Chudnovsky, E.M., Phys. Rev. Lett., 72:3433 (1994). 68. Tejada, J., Hernandez, J.M., and de Barco, E., J. Magn. Magn. Mater., 196–197:552 (1999). 69. Sappey, V., Vincent, E., Ocio, M., and Hamman, J., J. Magn. Magn. Mater., 221:87 (2000). 70. Awschalom, D.D., McCord, M.A., and Grinstein, G., Phys. Rev. Lett., 65:783 (1990). 71. Gider, S., Awschalom, D.D., Douglas, T., Mann, S., and Charapala, M., Science, 268:1–4 (1995). 72. Wernsdorfer, W., Orozco, E.B., Hasselbach, K., Benoit, A., Mailly, D., Kubo, O., Nakano, H., and Barbara, B., Phys. Rev. Lett., 79:4014 (1997).

10: Magnetic Nanoparticles and Applications, Majetich

477

73. Tejada, J., Zhang, X.X., de Barco, E., Hernandez, J.M., and Chudnovsky, E.M., Phys. Rev. Lett., 79:1754 (1997). 74. Battle, X., Garcia del Muro, M., and Labarta, A., Phys. Rev. B, 55:6440 (1997). 75. Thomas, L., Lionti, F., Ballou, R., Gatteschi, D., Sessoli, R., and Barbara, B., Nature, 383:145 (1996). 76. Shen, J., Ohresser, P., V Mohan, Ch., Klaua, M., Barthel, J., and Kirshney, J., Phys. Rev. Lett., 80:1980 (1998). 77. Arrott, A.S., Templeton, T.L., and Yoshida, Y., IEEE Trans. Magn., 29:2622 (1993). 78. Kodama, R., and Berkowitz, A.E., Phys. Rev. Lett., 77:394–397 (1996). 79. De Biasi, E., Ramos, C.A., Zysler, R.D., and Romero, H., Phys. Rev. B, 65:144416 (2002). 80. Wilcoxon, J.P., Venturini, E.L., and Provencio, P., Phys. Rev. B, 69:172402 (2004). 81. Pankhurst, Q.A., Yedra Martinez, A., and Fernandez Barquin, L., Phys. Rev. B, 69:212401 (2004). 82. Respaud, M., Broto, J.M., Rakoto, H., Fert, A.R., Thomas, L., Barbara, B., Verelst, M., Snoeck, E., Lecante, P., Mosset, A., Osuna, J., Ould Ely, T., Amiens, C., and Chaudret, B., Phys. Rev. B, 57:2925–2935 (1998). 83. Cabrera, N., and Mott, N., Rep. Prog. Phys., 12:163–184 (1948). 84. Gangopadhyay, S., Hadjipanayis, G.C., Dale, B., Sorensen, C.M., Klabunde, K.J., Papaefthymiou, V., and Kostikos, A., Phys. Rev. B, 45:9778 (1992). 85. Meiklejohn, W.H., and Bean, C.P., Phys. Rev., 105:904 (1957). 86. Ijiri, Y., Kelly, C.V., Borchers, J.A., Rhyne, J.J., Farrell, D.F., Majetich, S.A., Appl. Phys. Lett., 86:243102–243104 (2005). 87. Skumryev, V., Stoyanov, S., Zhang, Y., Hadjipanayis, G.C., Givord, D., Nature, 423:850 (2003). 88. Givord, D., Skumryev, V., and Nogues, J., J. Magn. Magn. Mater., 294:111–116 (2005). 89. Ohnuma, M., Hono, K., Onodera, H., Ohnuma, S., Fujimori, H., and Pedersen, J.S., J. Appl. Phys., 87, 817–823 (2000). 90. Kwok, Y.S., Zhang, X.X., Qin, B., and Fung, K.K., Appl. Phys. Lett., 77:3971–3973 (2000). 91. Luttinger, J.M., and Tisza, L., Phys. Rev., 70:954–964 (1946). 92. Fernandez, J.F., and Alonso, J.J., Phys. Rev. B, 62:53–56 (2000). 93. Mukhopadhyay, G., Apell, P., and Hanson, M., J. Magn. Magn. Mater., 203:286 (1999). 94. Hess, P.H., and Parker, P.H., P.H., Jr., J. Appl. Polym. Sci., 10:1915–1927 (1966). 95. Griffith, C.H., O’Horo, M.P., and Smith, T.W., J. Appl. Phys., 50:7108 (1979). 96. Grinstaff, M.W., Salamon, M.B., and Suslick, K.S., Phys. Rev. B, 48:269 (1993). 97. Park, S.J., Kim, S., Lee, S., Zhim, Z.G., Char, K., and Hyeon, T., J. Am. Chem. Soc., 122:8581–8582 (2000). 98. Luo, W., Nagel, S.R., Rosenbaum, T.F., and Rosensweig, R.E., Phys. Rev. Lett., 67:2721–2724 (1992).

478

Properties

99. Chantrell, R.W., Fearon, M., and Wohlfarth, E.P., Phys. Stat. Sol. (a), 97:213 (1986). 100. O’Grady, K., El-Hilo, M., and Chantrell, R.W., IEEE Trans. Magn., 29:2608 (1993). 101. Zhang, J., Boyd, C., and Luo, W., Phys. Rev. Lett., 77:390–393 (1996). 102. Djurberg, C., Swedlindh, P., Nordblad, P., Hansen, M.F., Bødker, F., and Mørup, S., Phys. Rev. Lett., 79:5154 (1997). 103. Humfeld, K.D., Giri, A.K., Majetich, S.A., and Venturini, E.L., IEEE Trans. Magn., 37:2194–2196 (2001). 104. Chamberlin, R.V., Humfeld, K.D., Farrell, D., Yamamuro, S., Ijiri, Y., and Majetich, S.A., J. Appl. Phys. 91:6961–6963 (2001). 105. Chamberlin, R.V., Hemberger, J., Loidl, A., Humfeld, K.D., Farrell, D., Yamamuro, S., Ijiri, Y., and Majetich, S.A., Phys. Rev. B, 66:172403 (2002). 106. Mørup, S., Madsen, M.B., Frank, J., Villadsen, J., and W. Koch, C.J., J. Magn. Magn. Mater., 40:163 (1983). 107. Jonsson, T., Nordblad, P., and Svedlindh, P., Phys. Rev. B, 57:497 (1998). 108. Dormann, J., Bessais, L., and Fiorani, D., J. Phys. C, 21:2015 (1988). 109. Hendriksen, P.V., Mørup, S., Christiansen, G., and Jacobsen, K.W., in: Science and Technology of Nanostructured Magnetic Materials (G. Hadjipanayis, and G.A. Prinz, eds.), p. 573, Plenum Press, New York (1992). 110. Mørup, S., and Tronc, E., Phys. Rev. Lett,. 72:3278 (1994). 111. Mørup, S., and Frandsen, C., Phys. Rev. Lett., 92:217201 (2004). 112. Hansen, M.F., and Mørup, S., J. Magn. Magn. Mater., 184:262–274 (1998). 113. Dormann, J.L., Fiorani, D., and Tronc, E., J. Magn. Magn. Mater., 202:251–267 (1999). 114. Puntes, V.F., Krishnan, K.M., and Alivisatos, P., Appl. Phys. Lett., 78:2187 (2001). 115. Puntes, V.F., Gorostiza, P., Aruguete, D.M., Bastus, N.G., and Alivisatos, A.P., Nature Mater., 3:263 (2004). 116. Bao, Y., Pakhomov, A.B., and Krishnan, K.M., J. Appl. Phys., 97:101317 (2005). 117. Sun, S., and Murray, C.B., J. Appl. Phys., 85:4325–4330 (1999). 118. Held, G.A., Grinstein, G., Doyle, H., Sun, S., and Murray, C.B., Phys. Rev. B, 64:012408 (2001). 119. Poddar, P., Telem-Shafir, T., Fried, T., and Markovich, G., Phys. Rev. B, 66:060403 (2002). 120. Petit, C., Russier, V., and Pileni, M.P., J. Phys. Chem. B, 107:10333–10336 (2003). 121. Russier, V., Petit, C., Legrand, J., and Pileni, M.P., Phys. Rev. B, 62:3910 (2000). 122. Lalatonne, Y., Motte, L., Russier, V., Ngo, A.T., Bonville, P., and Pileni, M.P., J. Phys. Chem. B, 108:1848–1854 (2004). 123. Lalatonne, Y., Richardi, J., and Pileni, M.P., Nature Mater., 3:121–123 (2004). 124. Maillard, M., Motte, L., Ngo, A.T., and Pileni, M.P., J. Phys. Chem. B, 104:11871–11877 (104). 125. Pileni, M.P., Appl. Surf. Sci., 171:1–14 (2001).

10: Magnetic Nanoparticles and Applications, Majetich

479

126. Garcia-Otero, J., Porto, M., Rivas, J., and Bunde, A., Phys. Rev. Lett,. 84:167–170 (2000). 127. Farrell, D., Cheng, Y., McCallum, R.W., and Majetich, S.A., J. Phys. Chem. B, 109:13409–13419 (2005). 128. Farrell, D., Ding, Y., Sanchez-Hanke, C., Kao, C.-C., and Majetich, S.A., J. Appl. Phys., 95:6636 (2004). 129. Farrell, D., Cheng, Y., Ding, Y., Yamamuro, S., Sanchez-Hanke, C., Kao, C.-C., and Majetich, S.A., J. Magn. Magn. Mater., 282:1–5 (2004). 130. Poddar, P., Wilson, J.L., Srikanth, H., Farrell, D.F., and Majetich, S.A., Phys. Rev. B, 68:214409 (2003). 131. Giersig, M., and Hilgendorff, M., J. Phys. D: Appl. Phys., 32:L111–L113 (1999). 132. Bellouard, C., Mirebeau, I., and Hennion, M., Phys. Rev. B, 53:5570–5578 (1996). 133. Kortright, J.B., Hellwig, O., Chesnel, K., Sun, S., and Fullerton, E.E., Phys. Rev. B, 71:012402 (2005). 134. Mydosh, J.A., Spin Glasses: An Experimental Introduction, p. 36, Taylor and Francis, London, Washington, DC (1993). 135. Schultess, T.C., Benakli, M., Visscher, P.B., Sorge, K.D., Thompson, J.R., Modine, F.A., Haynes, T.E., Boatner, L.A., Stocks, G.M., and Butler, W.H., J. Appl. Phys., 89:7594–7596 (2001). 136. Wohlfarth, E.P., J. Appl. Phys., 29:595 (1958). 137. Kelly, P.E., O’Grady, K., Mayo, P.I., and Chantrell, R.W., IEEE Trans. Magn., 25:3881 (1989). 138. Henkel, O., Phys. Stat. Sol., 7:919 (1964). 139. Chantrell, R.W., Walmsley, N., Gore, J., and Maylin, M., Phys. Rev. B, 63:024410 (2000). 140. Kechrakos, D., and Trohidou, K.N., Phys. Rev. B, 58:12169–12177. (1998). 141. Denisov, S.I., Lyutyy, T.V., and Trohidou, K.N., Phys. Rev. B, 67:014411 (2003). 142. Zhang, H., and Widom, M., Phys. Rev. B, 51:8591–8597 (1995). 143. Löffler, J., Kostorz, G., Widenmann, A., and Wagner, W., Physica B, 241–243:603–605 (1998). 144. Löffler, J.F., Braun, H.-B., and Wagner, W., Phys. Rev. Lett., 85:1990–1993 (2000). 145. Schäfer, R., J. Magn. Magn. Mater., 215–216:652–663 (2000). 146. Berkowitz, A.E., Mitchell, J.R., Carey, M.J., Young, A.P., Zhang, S., Spada, F.E., Parker, F.T., Hutten, A., and Thomas, G., Phys. Rev. Lett., 68:3745–3748 (1992). 147. Xiao, J.Q., Jiang, J.S., and Chien, C.L., Phys. Rev. Lett., 68:3749–3752 (1992). 148. Allia, P., Coisson, M., Knobel, M., Tiberto, P., and Vinui, F., Phys. Rev. B, 60:12207 (1999). 149. Ucko, D.H., Pankhurst, Q.A., Fernandez Barquin, L., and Rodriguez Fernandez, J., Phys. Rev. B., 64:104433 (2001). 150. Pankhurst, Q.A., Ucko, D.H., Fernandez, L., Barquin, R., Garcia Calderon, J. Magn. Magn. Mater., 266:131–141 (2003).

480

Properties

151. Scheinfein, M.R., Schmidt, K.E., Heim, K.R., and Hembree, G.G., Phys. Rev. Lett., 76:1541 (1996). 152. Pierce, J.P., Torija, M.A., Gai, Z., Shi, J., Schulthess, T.C., Farnan, G.A., Wendelen, J.F., Plummer, E.W., and Shen, J., Phys. Rev. Lett., 96:237201 (2004). 153. Kondratyev, V.N., and Lutz, H.O., Phys. Rev. Lett., 81:4508 (1998). 154. Genkin, G.M., and Sapozhnikov, M.V., Appl. Phys. Lett., 64:794 (1994). 155. Vargas, P., and Altbir, D., J. Magn. Magn. Mater., 167:161 (1997). 156. Skomski, R., Europhys. Lett., 48:455–460 (1999). 157. Skomski, R., Kashyap, A., Qiang, Y., and Sellmyer, D.J., J. Appl. Phys., 93:6477 (2003). 158. O’Handley, R.C., Modern Magnetic Materials, p. 122, Wiley, New York, (2000). 159. Yamamuro, S., Farrell, D., and Majetich, S.A., Phys. Rev. B, 65:224431 (2002). 160. Korgel, B.A., and Fitzmaurice, D., Phys. Rev. B, 59:14191 (1999). 161. Berger, P., Adelman, N., Beckman, K.J., Campbell, D.J., Ellis, A.B., and Lisensky, G.C., J. Chem. Educ., 76:943–948 (1999). 162. Bacri, J.-C., Perzynski, R., Salin, D., Cabuil, V., and Massart, R., J. Magn. Magn. Mater., 85:27–32 (1990). 163. Farrell, D.F., thesis, Ph.D., Carnegie Mellon University, 2003. 164. Butter, K., H. Bomans, P.H., Frederik, P.M., Vroege, G.J., and Philipse, A.P., Nature Mater., 2:88–91 (2003). 165. Rosensweig, R.E., Ferrohydrodynamics, Cambridge University Press, Cambridge, UK (1995). 166. O’Grady, K., Stewardson, H., Chantrell, R., Fletcher, D., Unwin, D., and Parker, M., IEEE Trans. Magn., 22:1134–1136 (1986). 167. Raj, K., and Moskowitz, R., Magn. Magn. Mater., 85:233–245 (1990). 168. Raj, K., Moskowitz, R., and Casciari, R., J. Magn. Magn. Mater., 149:174–180 (1995). 169. Ziolo, R.F., Giannelis, E.P., Weinstein, B., O’Horo, M.P., Ganguly, B.N., Mehrotra, V., Russell, M.W., and Huffman, D.R., Science, 257:219 (1992). 170. Berry, C.C., and G. Curtis, A.S., J. Phys. D.: Appl. Phys., 36:R198–R206 (2003). 171. Pankhurst, Q.A., Connolly, J., Jones, S.K., and Dobson, J., J. Phys. D: Appl. Phys., 36:R167–R181 (2003). 172. Rosensweig, R.E., J. Magn. Magn. Mater., 252:370–374 (2002). 173. Tartaj, P., del Puerto Morales, M., Veintemillas-Verdaguer, S., GonzalezCarreno, T., and Serna, C.J., J. Phys. D.: Appl. Phys., 36:R182–R197 (2003). 174. Gilles, C., Bonville, P., Rakoto, H., Broto, J.M., W. Wong, K.K., and Mann, S., J. Magn. Magn. Mater., 241:430–440 (2002). 175. Roath, S., J. Magn. Magn. Mater., 122:329–340 (1993). 176. Molday, R.S., and Mackenzie, D., J. Immunol. Methods. 32, 353–367 (1982). 177. Reich, D.H., Bauer, L.A., Tanase, M., Hultgren, A., Chen, C.S., and Meyer, G.J., J. Appl. Phys., 93:7275 (2003). 178. Vadala, M.L., Rutnakornpituk, M., Zalich, M.A., St. Pierre, T.G., and Riffle, J.S., Polymer, 45:7449–7461 (2004).

10: Magnetic Nanoparticles and Applications, Majetich

481

179. Gu, H., Zheng, R., Zhang, X.X., and Xu, B., J. Am. Chem. Soc., 126:5664–5665 (2004). 180. Robinson, D.B., J. Persson, H.H., Zeng, H., Li, G., Pourmand, N., Sun, S., and Wang, S.X., Langmuir, 21:3096–3103 (2005). 181. Tedesco, A.C., Oliveira, D.M., M. Lacava, Z.G., Azevedo, R.B., D. Lima, E.C., Gansau, C., Buske, N., and Morais, P.C., J. Appl. Phys., 93:6704 (2003). 182. Tada, M., Hatanaka, S., Sanbonsugi, H., Matsushita, N., and Abe, M., J. Appl. Phys., 93:7566 (2003). 183. Bertozzi, C.R., and Kiessling, L.L., Science, 291:2357–2364 (2001). 184. Cao, J., Wang, Y., Yu, J., Xia, J., Wang, C., Yin, D., and Hafeli, U.O., J. Magn. Magn. Mater., 277:165–174 (2004). 185. Zborowski, M., and Chalmers, J.J., Methods Mol. Biol., 295:291–300 (2005). 186. McCloskey, K.E., Chalmers, J.J., and Zborowski, M., Anal. Chem., 75:6868–6874 (2003). 187. McCloskey, K.E., Moore, L.R., Hoyos, M., Rodriguez, A., Chalmers, J.J., and Zborowski, M., Biotechnol. Prog., 19:899–907 (2003). 188. Ruuge, E.K., Rusetski, A.N., J. Magn. Magn. Mater., 122:335 (1993). 189. Jain, T.K., Morales, M.A., Sahoo, S.K., Leslie-Pelecky, D.L., and Labhasetwar, V., Mol. Pharmaceut., 2:194–205 (2005). 190. Lao, L.L., and Ramanujan, R.V., J. Mater. Sci.: Mater Med., 15:1061–1064 (2004). 191. Yellen, B., and Friedman, G., Adv. Mater., 16:111 (2004). 192. Gunnarsson, K., Roy, P.E., Felton, S., Pihl, J., Svedlindh, P., Berner, S., Lindbaum, H., and Oscarson, S., Adv. Mater., 17:1730–1734 (2005). 193. Miller, M.M., Sheehan, P.E., Edelstein, R.L., Tamanaha, C.R., Zhong, L., Bounnak, S., Whitman, L.J., and Colton, R.J., J. Magn. Magn. Mater., 25:138 (2001). 194. Baselt, D.R., Lee, G.U., Natesan, M., Metzger, S.W., Sheehan, P.E., and Colton, R.J., Biosensors and Bioelectronics, 13:731 (1998). 195. Tondra, M., Popple, A., Jander, A., Millen, R.L., Pekas, N., and Porter, M.D., J. Magn. Magn. Mater., 293:725–730 (2005). 196. Grancharov, S.G., Zeng, H., Sun, S., Wang, S.X., O’Brien, S., Murray, C.B., Kirtley, J.R., and Held, G.A., J. Phys. Chem. B, 109:13030–13035 (2005). 197. Ferreira, H.A., Feliciano, N., Graham, D.L., and Freitas, P.P., J. Appl. Phys., 97:10Q904 (2005). 198. Andrä, W., d’Ambly, C.G., Hergt, R., Hilger, I., and Kaiser, W.A., J. Magn. Magn. Mater., 194:197–203 (1999). 199. C. Chan, D.F., Kirpotin, D.B., and Bunn Jr., P.A., J. Magn. Magn. Mater., 122:374–378 (1993). 200. Hergt, R., Andrä, W., d’Ambly, C.G., Hilger, I., Kaiser, W.A., Richter, U., and Schmidt, H.G., IEEE Trans. Magn., 34:3745–3754 (1998). 201. Jordan, A., Wust, P., Scholtz, R., Tesche, B., Fähling, H., and Mitrovics, T., Int. J. Hyperthermia, 12:705–722 (1996). 202. Jordan, A., Scholz, R., Wust, P., Fähling, H., and Felix, R., J. Magn. Magn. Mater., 201:413–419 (1999). 203. Pardoe, H., Clark, P.R., St. Pierre, T.G., Moroz, P., and Jones, S.K., Magn. Reson. Imaging, 21:483–488 (2003).

482

Properties

204. Romanus, E., Hückel, M., Gross, C., Glöckl, G., Weber, P., and Weischies, W., J. Magn. Magn. Mater., 252:387–389 (2002). 205. Josephson, L., Jacobs, P., Hahn, P.F., and Stark, D., Magn. Reson. Imaging, 6, 647–653 (1988). 206. Toth, E., and Merbach, A.E., The Chemistry of Contrast Agents in Medical Magnetic Resonance Imaging, Wiley, Chichester, UK (2001). 207. Dodd, S.J., Williams, M., Suhan, J.P., Williams, D.S., Koretsky, A.P., and Ho, C., Biophys J., 76:103–109 (1999). 208. M. Bulte, J.W., Hoekstra, Y., Kamman, R.L., Magin, R.L., Webb, A.G., Briggs, R.W., Go, K.G., Hulstaert, C.E., Miltenyi, S., The, T.H., and de Leij, L., Magn. Reson. Med., 25:148–157 (1992). 209. Pouliquen, D., Perdrisot, R., Ermias, A., Akoka, S., Jallet, P., and Le Jeune, J.J., Magn. Reson. Imaging, 7:619 (1989). 210. Weissleder, R., Lee, A.S., Khaw, B.A., Shen, T., and Brady, T.J., Radiology, 182:381–385 (1992). 211. Weissleder, R., Cheng, H., Bogdanova, A., and Bogdanov, A., J. Magn. Reson. Imaging, 7:258–263 (1996). 212. Yeh, T.-C., Zhang, W., Ildstad, S.T., and Ho, C., Magn. Reson. Med,. 30:617–625 (1993). 213. Yeh, T.-C., Zhang, W., Ildstad, S.T., and Ho, C., Magn. Reson. Med., 33:200–208(1995). 214. Lange, J., Kötitz, R., Haller, A., Trahms, L., Semmler, W., and Weitschies, W., J. Magn. Magn. Mater., 252:381–383 (2002). 215. Schoepf, U., Marecos, E.M., Melder, R.J., Jain, R.K., and Weissleder, R., BioTechniques, 24:642–651(1998). 216. Grant, S.C., Aiken, N.R., Plant, H.D., Gibbs, S., Mareci, T.H., Webb, A.G., and Blackband, S.J., Magn. Reson. Med., 44:19–22 (2000). 217. Lauterbur, P.C., Bernardo, M.L., Jr., Mendonca Dias, M.H., and Hedges, L.K., “Microscopic NMR imaging of the magnetic field around magnetic particles”, in: Proc. of the Society of Magnetic Resonance in Medicine, 5th Annual Meeting, Montreal, Quebec, Canada, pp. 229–230 (1986). 218. Gleich, B., and Weizenecker, J., Nature, 435:1214–1217 (2005). 219. Stipe, B.C., Mamin, H.J., Stowe, T.D., Kenny, T.W., and Rugar, D., Phys. Rev. Lett., 86:2874–2877 (2001). 220. Bruland, K.J., Dougherty, W.M., Garbini, J.L., Sidles, J.A., and Chao, S.H., Appl. Phys. Lett., 73:3159–3161 (1998). 221. Barbic, M., and Scherer, A., J. Appl. Phys., 95:3598–3600 (2004). 222. Barbic, M., and Scherer, A., Nano Lett., 5, 187–190 (2005). 223. Zhang, Z., Hammel, P.C., and Wigen, P.E., Appl. Phys. Lett., 68:2005–2007 (1996). 224. Speliotis, D.E., J. Magn. Magn. Mater., 193:2935 (1999). 225. Weller, D., Moser, A., Folks, L., Best, M.E., Lee, W., Toney, M.F., Schwickert, M., Thiele, J.-U., and Doerner, M.F., IEEE Trans. Magn., 36:10–15 (2000). 226. Victora, R.H., IEEE Trans. Magn., 38:1886–1891 (2002). 227. Lu, P., and Charap, S., IEEE Trans. Magn., 31:2767 (1995). 228. Ouchi, K., and Honda, N., IEEE Trans. Magn., 36:16–22 (2000).

10: Magnetic Nanoparticles and Applications, Majetich

483

229. Kang, K., Suzuki, T., Zhang, Z.G., and Papusoi, C., J. Appl. Phys., 95:7273–7275 (2004). 230. Yan, M.L., Sabiryanov, R.F., Xu, Y.F., Li, X.Z., and Sellmyer, D.J., IEEE Trans. Magn., 40:2470–2472 (2004). 231. Chou, S.Y., Proc. IEEE, 85:652–671 (1997). 232. Allwood, D.A., Xiong, G., Cooke, M.D., Faulkner, C.C., Atkinson, D., Vernier, N., and Cowburn, R.P., Science, 296:2003–2006 (2002). 233. Raabe, J., Pulwey, R., Sattler, R., Schweinbeck, T., Zweck, J., and Weiss, D., J. Appl. Phys., 88:4437–4439 (2001). 234. Hehn, M., Ounadjela, K., Bucher, J., Dusseaux, F.R., Decanini, D., Bartenliane, B., and Chappert, C., Science, 272:1782–1785 (1996). 235. Kirk, K.J., Chapman, J.N., and Wilkinson, C.D., J. Appl. Phys., 85:5237–5242 (1999). 236. Kent, A.D., Shaw, T.M., von Molnár, S., and Awschalom, D.D., Science, 262:1249 (1993). 237. Chou, S.Y., Krauss, P.R., and Renstron, P.J., Science, 272:85–87 (1996). 238. Gourgon, C., Perret, C., Tallal, J., Lazzarmo, F., Landis, S., Joubert, O., and Pelzer, R., J. Phys. D: Appl. Phys., 38:70–73 (2005). 239. Masuda, H., and Fukuda, K., Science, 268:1466–1468 (1995). 240. Jessensky, O., Müller, F., and Gösele, U., J. Electrochem. Soc., 145:2473–2480 (1998). 241. Metzger, R.M., Konovolov, W., Sun, M., Xu, T., Zangari, G., Xu, B., Benakli, M., and Doyle, W.D., IEEE Trans. Magn., 36:30–35 (2000). 242. Piraux, L., Dubois, S., Ferain, E., Legras, R., Ounadjela, K., George, J.M., Maurice, J.L., and Fert, A., J. Magn. Magn. Mater., 165:352 (1997). 243. Liu, K., Nagodawithana, K., Searson, P.C., and Chien, C.L., Phys. Rev. B, 51:7381–7384 (1995). 244. Wegrowe, J.E., Gilbert, S.E., Kelly, D., Doudin, B., and Ansermet, J.-Ph., IEEE Trans. Magn., 34:903 (1998). 245. Park, M., Harrison, C., Chaikin, P.M., Register, R.A., and Adamson, D.H., Science, 276:1401–1404 (1997). 246. Thurn-Albrecht, T., Schotter, J., Kästle, G.A., Emley, N., Shiauchi, T., Krusin-Elbaum, L., Guarini, K., Black, C.T., Tuominen, M.T., and Russell, T.P., Science, 290:2126–2129 (2000). 247. Black, C.T., Guarmi, K.W., Milkove, K.R., Baker, S.M., Russell, T.P., and Tuominen, M.T., Appl. Phys. Lett., 79:409–411 (2001). 248. Thurn-Albrecht, T., Steiner, R., De Rouchey, J., Stafford, C.M., Huang, E., Bal, M., Tuominen, M., Hawker, C.J., and Russell, T.P., Adv. Mater.,12:787 (2000). 249. H. Lammertink, R.G., Hempenius, M.A., H. Chan, V.Z., Thomas, E.L., and Vancso, G.J., Chem. Mater., 13:429–434 (2001). 250. Cheng, J.Y., Mayes, A.M., and Ross, C.A., Nature Mater., 3:823–828 (2004). 251. Shima, M., Hwang, M., and Ross, C.A., J. Appl. Phys., 93:3440 (2003). 252. Abraham, M.C., Schmidt, H., Savas, T.A., Smith, H.I., Ross, C.A., and Ram, R.J., J. Appl. Phys., 89:5667–5670 (2001). 253. Abes, J.I., Cohen, R.E., and Ross, C.A., Mater. Sci. Eng. C, 23:641–650 (2003).

484

Properties

254. Kimura, M., Misner, M.J., Xu, T., Kim, S.H., and Russell, T.P., Langmuir, 19:9910–9913 (2003). 255. Naito, K., Hieda, H., Sakurai, M., Kamata, Y., and Asakawa, K., IEEE Trans. Magn., 38:1949–1951 (2002). 256. Sun, S., Murray, C.B., Weller, D., Folks, L., and Moser, A., Science, 287:1989 (2000). 257. Liu, C., Wu, X., Klemmer, T., Shukla, N., Yang, X., Weller, D., Roy, A.G., Tanase, M., and Laughlin, D., J. Phys. Chem. B, 108:6121–6123 (2004). 258. Liu, C., Wu, X., Klemmer, T., Shukla, N., Weller, D., Roy, A., Tanase, M., and Laughlin, D., Chem. Mater., 17:620–625 (2005). 259. Chen, M., and Nikles, D.E., Nano Lett., 2:211 (2002). 260. Kang, S., Harrell, J.W., and Nikles, D.E., Nano Lett., 2:1033 (2002). 261. Kang, S., Jia, Z., Shi, S., Nikles, D.E., and Harrell, J.W., Appl. Phys. Lett., 86:062503 (2005). 262. Chen, M., Nikles, D.E., Yin, H., Wang, S., Harrell, J.W., and Majetich, S.A., J. Magn. Magn. Mater., 266: 8–11 (2003). 263. C. Yu, A.C., Mizuno, M., Sasaki, Y., Inoue, M., Kondo, H., Ohta, I., Djayaprawira, D., and Takahashi, M., Appl. Phys. Lett., 82:4352 (2003). 264. Jeyadevan, B., Urakawa, K., Hobo, A., Chinnasamy, N., Shinoda, K., Tohji, K., J. Djayaprawira, D.D., Tsunoda, M., and Takahashi, M., Jpn. J. Appl. Phys., 42:L350–L352 (2003). 265. Ding, Y., Farrell, D., Yamamuro, S., and Majetich, S.A., J. Appl. Phys., 93:7411 (2003). 266. Ding, Y., Majetich, S.A., Kim, J., Barmak, K., Rollins, H., and Sides, P., J. Magn. Magn. Mater., 284:336–341 (2004). 267. Ding, Y., and Majetich, S.A., Appl. Phys. Lett., 87:022508 (2005). 268. Teng, X., and Yang, H., J. Am. Chem. Soc., 125:14559–14563 (2003). 269. Chen, M., Liu, J.P., and Sun, S., J. Am. Chem. Soc., 126: 8394–8395 (2004). 270. Thompson, T., Toney, M.F., Raoux, S., Lee, S.L., Sun, S., Murray, C.B., and Terris, B.D., J. Appl. Phys., 96:1197–1199 (2004). 271. Bian, B., Laughlin, D.E., Sato, K., and Hirotsu, Y., IEEE Trans. Magn., 36:3021–3023 (2000). 272. Takahashi, Y., Ohkubo, T., Ohnuma, M., and Hono, K., J. Appl. Phys., 93:7166–7168 (2003). 273. Gai, Z., Howe, J.Y., Guo, J., Plummer, E.W., and Shen, J., Appl. Phys. Lett., 86:023107 (2005). 274. Terheiden, A., Mayer, C., Moh, K., Stahlmecke, B., Stappert, S., Acet, M., and Rellinghaus, B., Appl. Phys. Lett., 84:3891–3893 (2004). 275. Kan, S., Sachan, M., Kirchhoff, J., and Majetich, S.A., IEEE Trans. Magn., 41:3370–3372 (2005). 276. Bobo, J.F., Gabillet, L., and Bibes, M., J. Phys. Condens. Matter, 16:S471–S496 (2004). 277. Wolf, S.A., Awschalom, D.D., Buhrman, R.A., Daughton, J.M., von Molnar, S., L Roukes, M., Ylchtchelkanova, A., and Treger, D.M., Science, 294:1488–1495 (2001). 278. Meservey, P., and Tedrow, P.M., Phys. Rep., 238:173 (1994). 279. Gurney, B.A., Speriosu, V.S., Nozieres, J.P., Lefakis, H., Wilhoit, D.R., and Need, O.U., Phys. Rev. Lett., 71:4023 (1993).

10: Magnetic Nanoparticles and Applications, Majetich

485

280. Baibich, M.N., Broto, J.M., Fert, A., Nguyen Van Dau, F., Petroff, F., Etienne, P., Creuzet, G., Friederich, A., and Chazelas, J., Phys. Rev. Lett., 61:2472 (1988). 281. Zhang, X.X., Wan, C., Liu, H., Li, Z.Q., Sheng, P., and Lin, J.J., Phys. Rev. Lett., 86:5562–5565 (2001). 282. Denardin, J.C., Pakhomov, A.B., Brandl, A.L., Socolovsky, L.M., Knobel, M., and Zhang, X.X., Appl. Phys. Lett., 82:763–765 (2003). 283. Velazquez, J., Vazquez, M., Chen, D.-X., and Hernando, A., Phys. Rev. B, 50:16737–16740 (1994). 284. Black, C.T., Murray, C.B., Sandstrom, R.L., and Sun, S., Science, 290:1131 (2000). 285. Beecher, P., Quinn, A.J., Shevchenko, E.V., Weller, H., and Redmond, G., Nano Lett., 4:1289–1293 (2004). 286. Beecher, P., Quinn, A.J., Shevchenko, E.V., Weller, H., and Redmond, G., J. Phys. Chem. B, 108:9564–9567 (2004). 287. Poddar, P., Fried, T., and Markovich, G., Phys. Rev. B, 65:172405 (2002). 288. Zhang, S., Phys. Rev. Lett., 83:640–643 (1999). 289. Kiselev, S.I., Sankey, J.C., Krivorotov, I.N., Emley, N.C., Schoelkopf, R.J., Buhrman, R.A., and Ralph, D.C., Nature, 425:380–383 (2003). 290. Albert, F.J., Katine, J.A., Buhrman, R.A., and Ralph, D.C., Appl. Phys. Lett., 77:3809–3811 (2000). 291. Katine, J.A., Albert, F.J., Buhrman, R.A., Myers, E.B., and Ralph, D.C., Phys. Rev. Lett., 84:3149–3152 (2000). 292. Grollier, J., Cros, V., Hamzic, A., George, J.M., Jaffres, H., Fert, A., Faini, G., Ben Youssef, J., and Legall, H., Appl. Phys. Lett., 78:3663–3665 (2001). 293. AlHajDarwish, M., Kurt, H., Urazhdin, S., Fert, A., Lolee, R., Pratt, W.P., Jr., and Bass, J., Phys. Rev. Lett., 93:157203 (2004). 294. Slonczewski, J.C., J. Magn. Magn Mater., 159:L1–7 (1996). 295. Berger, L., J. Magn. Magn Mater., 162:155–61 (1996). 296. Zhu, J.-G., Kim, N., Zhou, Y., Zheng, Y., Chang, J., Ju, K., Zhu, X., and White, R.M., IEEE Trans. Magn., 40: 2325–2327 (2004). 297. Hendriksen, P.V., Linderoth, S., and Lindgard, P.-A., Phys, Rev. B, 48:7259–7273 (1993). 298. Jorzick, J., Krämer, C., Demokritov, S.O., Hillebrands, B., Bartenlian, B., Chappert, C., Decanini, D., Rousseaux, F., Cambril, E., Søndergard, E., Bailleul, M., and Fermon, C., J. Appl. Phys., 89:7091–7095 (2001). 299. Julliere, M., Phys. Lett., A54:225 (1975). 300. Tedrow, P.M., and Merservey, P., Phys. Rev. Lett., 26:192 (1971). 301. Moodera, J.S., Kinder, L.R., Wong, T.M., and Meservey, R., Phys. Rev. Lett., 74:3273–3276 (1995). 302. Miyazaki, T., and Tezuka, N., J. Magn. Magn. Mater., 139:L231 (1995). 303. Daughton, J., Thin Solid Films, 216:162–168 (1992). 304. Tehrani, S., Chen, E., Durlam, M., De Herrera, M., Slaughter, J.M., Shi, J., and Kerszykowski, G., J. Appl. Phys., 85:5882–5887 (1999).

11 Magnetic Properties of Nanocrystalline Materials Akihisa Inoue1, Akihiro Makino1 and Teruo Bitoh2 1

2

Advanced Research Center of Metallic Glasses, Institute for Materials Research, Tohoku University, Japan

Department of Machine Intelligence and Systems Engineering, Faculty of Systems Science and Technology, Akita Prefectural University, Japan

11.1 Introduction It was reported in 1976 that an amorphous phase in Pd—Si, Fe—P—C and (Fe, Co, Ni)—Si—B systems is useful as a precursor to prepare a nanocrystalline structure upon crystallization.[1] For the last two decades, a great effort has been devoted to the development of new types of high-strength or high-function materials utilizing the formation of crystallization-induced nanostructure. In the field of magnetic materials, good hard magnetic properties have been obtained in the crystallized structure of Nd2Fe14B and amorphous phases obtained from amorphous Fe— Nd—B alloys.[2] In addition for soft magnetic alloys, it has been found that crystallization of Fe—Si—B amorphous alloys containing Nb and Cu causes the formation of a nanoscale bcc structure and that the bcc alloys exhibit good soft magnetic properties with saturation magnetic flux density (Bs) of 1.2 to 1.4 T and effective permeability (μe) at 1 kHz of 105.[3] These results indicate that crystallization-induced nanostructure is useful in creating hard or soft magnetic properties. Although good soft magnetic properties are obtained for nanoscale bcc Fe73.5Si13.5B9Nb3Cu1, the relatively low Fe concentration leads to Bs being limitated to less than 1.4 T. Hence, the development of a new soft magnetic alloy with high Bs (above 1.5 T) and high μe (above 105) at 1 kHz has been a strongly desired goal because the simultaneous achievement of both properties enables the extension of application fields to various kinds of power transformers. It has subsequently been reported[4] that Fe—C thin films consisting of nanocrystalline bcc and TaC phases obtained by crystallization of a sputtered amorphous phase exhibit good soft magnetic properties (i.e., Bs of 1.54 T and μe of 5.3 Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 487–536 © 2007 William Andrew, Inc.

487

488

Properties

× 103 at 1 MHz), though the nanostructure cannot be synthesized by the melt spinning technique. We have previously reported that Fe-rich amorphous alloys of 90 at% Fe are formed in Fe—Zr—B and Fe—Hf—B systems by rapid solidification.[5] If a nanostructure is formed in the Fe-rich alloys subjected to crystallization treatment, the nanostructured alloys are expected to exhibit high Bs exceeding 1.5 T. Based on this concept, we have examined[6–7] the possibility of synthesizing a nanocrystalline structure in Fe—M—B (M = Zr, Hf or Nb) alloys and of obtaining good soft magnetic properties and we have succeeded[8–11] in developing nanocrystalline Fe— Zr—Nb—B—Cu alloys with excellent soft magnetic properties (1.57 T for Bs and 16 × 104 for μe at 1 kHz) which had not been obtained before for any kind of soft magnetic material. It has been subsequently found[12–15] that the Cu-free quaternary Fe85.5Zr2Nb4B8.5 alloy exhibits high Bs of 1.64 T and high μe of 6.0 × 104 at 1 kHz, simultaneously. Furthermore, the Fe84.9Nb6B8P1Cu0.1 alloy combines a Bs of more than 1.6 T with good soft magnetic properties and good productivity has been developed.[15–17] On the other hand, it has been found[18,19] that the dissolution of a large amount of oxygen in the remaining amorphous phase is effective in achieving a drastic increase in electrical resistivity of sputtered Fe—Hf—O and Fe— Zr—O films, leading to the achievement of good high-frequency permeability in the range 1 to 100 MHz. This paper reviews our recent results in the formation of a nanogranular bcc and amorphous structures, the development of excellent soft magnetic material by nanocrystallization in the Fe—Zr—Nb—B—Cu, Fe—Zr—Nb—B, Fe—Nb—B—P—Cu and Fe— Hf—O systems and their engineering applications.

11.2 Fe–M–B (M = Zr, Hf or Nb) Amorphous Alloys and their Crystallization-Induced Nanostructure Iron-based amorphous alloys have been reported to be formed in a number of alloy systems such as Fe—(B, C, Si, P, Ge),[20] Fe—(Zr, Hf),[21] Fe—Re (Re = rare earth metal)[22] and Fe—(Zr, Hf, Nb, Ta)—B.[6,7] If we pay attention to a maximum Fe concentration for formation of an amorphous phase by the melt spinning method, the previous data show that the Fe concentration increases in the order Fe—(Zr, Hf) > Fe—Re > Fe—(Zr, Hf, Nb, Ta)—B > Fe—(B, C, Si, P, Ge). This indicates the possibility that the Bs values of the resulting nanocrystalline phases also increase in the same order. Hence, the relation between the formation tendency of nanocrystalline structure and soft magnetic properties was systematically

11: Magnetic Properties, Inoue et al.

489

examined for Fe—(Zr, Hf), Fe—Re and Fe—(Zr, Hf, Nb, Ta)—B amorphous ribbons prepared by melt spinning. The crystallized structure of the Fe90Zr10, Fe90Hf10 and Fe90Nd10 amorphous alloys consists of bcc and amorphous phases at the first-stage, and α-Fe and compound phases at the second stage, but the grain size of the bcc and α-Fe phases are above 40 nm and 80 nm, respectively, which are too large to obtain good soft magnetic properties.[23] Therefore, it is concluded that the binary Fe-based alloys cannot be regarded as an appropriate system leading to the desired nanocrystalline structure, though high Bs above 1.5 T is obtained. Figure 11.1 summarizes the composition ranges in which an amorphous phase is formed in melt-spun Fe—Zr—B, Fe—Hf—B and Fe—Nb—B alloys containing more than 75 at% Fe and the subsequent crystallized structure consists of a nanoscale bcc-Fe and amorphous structure.[24] The maximum Fe concentration for formation of an amorphous phase is about 92 at% for the Fe—Zr—B and Fe—Hf—B alloys and about 86 at% Fe for the Fe—Nb—B alloy. Besides, it is seen that a nanogranular bcc-Fe phase surrounded by the residual amorphous phase is formed in the Ferich concentration range above 86 at% Fe for Fe—Zr—B and Fe—Hf— B alloys. Here, it is important to describe the criteria for formation of nanoscale bcc-Fe and amorphous mixed structure in the limited Fe-rich composition ranges. Figure 11.2 shows differential thermal analysis (DTA) curves of the Fe90Zr7B3 and Fe89Hf7B4 amorphous alloys. Two exothermic peaks are seen on the DTA curves, indicating that the crystallization takes place through two stages. From X-ray diffraction (XRD) analysis and transmission electron microscopic (TEM) observation, the first-stage exothermic reaction is due to the precipitation of bcc-Fe phase and the second exothermic peak results from the precipitation of α-Fe and Fe2Zr or Fe2Hf phases from the bcc and remaining amorphous phases. Notice that the temperature interval between the first and second exothermic peaks is as large as 150 K, indicating that the bcc-Fe and amorphous phases have a high metastability. Figure 11.3 shows a bright field electron micrograph and selected-area electron diffraction (SAED) pattern of the Fe90Zr7B3 alloy annealed for 3.6 ks at 923 K, a temperature between the first and second exothermic peaks. The bcc phase has spherical equiaxed grains with a size of about 15 nm and each grain has a random orientation. Furthermore, the diffraction pattern reveals the existence of a residual amorphous phase and the absence of the second crystalline phase. Further heating to a temperature above the second exothermic peak caused the precipitation of α-Fe and Fe2Zr phases, accompanied by the complete disappearance of the remaining amorphous phase and significant grain growth of the α-Fe phase. Borides of any kind are not observed and hence

490

Properties

Figure 11.1 Compositional dependence of structure for Fe—Zr—B, Fe—Hf—B and Fe—Nb—B alloys in as-quenched and annealed states.

11: Magnetic Properties, Inoue et al.

491

Figure 11.2 Differential thermal analysis (DTA) curves of amorphous Fe90Zr7B3 and Fe89Hf7B4 alloys.

Figure 11.3 (a) Bright field transmission electron microscopy (TEM) image and (b) selected-area electron diffraction pattern of Fe90Zr7B3 alloy annealed at 923 K for 3.6 ks.

492

Properties

the B element is thought to be preferentially dissolved into the Fe2Zr phase because of the similarity of the crystal structure and chemical formula between Fe2Zr and Fe2B. The reason for the high thermal stability of the metastable bcc-Fe and amorphous phases is discussed in Section 11.3 on the basis of high-resolution TEM images, nanobeam electron diffraction and atom-probe field ion microscopic data.

11.3 Soft Magnetic Properties and Structural Analyses of Fe –M–B (M = Zr, Hf or Nb) Nanocrystalline Ternary Alloys Figure 11.4 shows the composition dependence of as-quenched phase, Bs, and μe at 1 kHz for the melt-spun Fe—Zr—B and Fe—Nb—B alloy ribbons subjected to optimum annealing treatments for 3.6 ks at 873 and 923 K. The amorphous phase is formed in the wide composition range up to 92 at% Fe for the Fe—Zr—B alloys and 84 at% Fe for the Fe—Nb— B alloys. The annealed alloys exhibit good soft magnetic properties of high Bs above 1.6 T and high μe exceeding 104 around the composition of 90 at% Fe in the former system and 84 at% Fe in the latter system. It is also seen in Fig. 11.4 that the best soft magnetic properties are obtained at the upper limit of Fe concentration where the amorphous single phase is obtained in the melt-spun state. The good correspondence is because nanoscale bcc structure is obtained only from the amorphous single phase in the Fe-rich composition range. Besides, one can notice that the composition range in which the best soft magnetic properties are obtained for the Fe—Nb—B alloys shifts to the higher B concentration in comparison with that for the Fe—Zr—B alloys. The deviation is due to the extension of the formation range of the nanoscale bcc-Fe structure resulting from a decrease in the precipitation tendency of compounds, reflecting the weaker bonding nature of Nb—Fe and Nb—B pairs as compared with Zr—Fe and Zr—B pairs. Similar data to those for the Fe—Zr—B nanocrystalline alloys have been obtained in Fe—Hf—B system.[9,24] Figure 11.5 shows the changes with annealing temperature (Ta) in the structure, Bs, μe, mean grain size of the bcc phase (D) and saturation magnetostriction constant (λs) for the melt-spun Fe90Zr7B3, Fe89Hf7B4 and Fe84Nb7B9 alloys. As Ta increases, the amorphous phase changes to a mostly single bcc phase in the Ta range of 750 to 930 K and α-Fe plus compound in the Ta range above 930 K. The D value is as small as 10 to 17 nm and increases rapidly upon the phase transition to α-Fe plus compound. The Bs and μe have nearly zero values in the amorphous single-

11: Magnetic Properties, Inoue et al.

493

Figure 11.4 Composition dependence of saturation magnetic flux density (Bs) and effective premaebility (μe) at 1 kHz for Fe—M—B (M = Zr or Nb) amorphous alloys annealed for 3.6 ks at 873 and 923 K. The data for as-quenched phase are also shown for reference.

phase state because of the Invar effect. The precipitation of bcc phase caused by a further increase in Ta gives rise to significant increases in Bs and μe, followed by maximum values of 1.4 to 1.7 T and 2 × 104 to 3 × 104, respectively, in the Ta range of 873 to 923 K just before the phase decomposition of the bcc phase. The parameter λs has positive values in the amorphous single-phase state for the three alloys and changes to negative values for the Zr- and Hf-containing bcc alloys, and to slightly positive or nearly zero values for the Nb-containing bcc alloy. Considering the correspondence between structure and magnetic properties, the best soft magnetic properties are obtained in a partially crystallized structure consisting of nanoscale bcc and amorphous phases. Therefore, it is concluded that the residual existence of the amorphous phase plays an

494

Properties

Figure 11.5 Changes in the structure, (a) saturation magnetic flux density (Bs), (b) effective permeability (μe), (c) mean grain size (D) and (d) saturation magnetostriction constant (λs) with annealing temperature (Ta) for amorphous Fe90Zr7B3, Fe89Hf7B4 and Fe84Nb7B9 alloys.

important role in the achievement of good soft magnetic properties as a result of the formation of a nanoscale mixed structure. Here, it is important for the understanding of the formation mechanism of the nanostructure to confirm the residual existence of the amorphous phase and to examine the solute concentrations of the bcc and amorphous phases. Figure 11.6 shows a high-resolution TEM image of Fe88Hf10B2

11: Magnetic Properties, Inoue et al.

495

Figure 11.6 (a) High-resolution transmission electron microscopy (TEM) image, (b) and (c) nanobeam diffraction patterns and (d) and (e) energy-dispersive X-ray spectroscopy (EDX) profiles taken from the small regions (1) and (2), respectively, with a diameter of 0.6 nm for amorphous Fe88Hf10B2 alloy annealed for 3.6 ks at 873 K.

alloy annealed for 3.6 ks at 873 K, together with data from nanobeam diffraction patterns and energy-dispersive X-ray spectroscopy (EDX) profiles taken from regions 1 and 2. From the fringe contrast in the TEM image and the diffraction patterns, the regions 1 and 2 are identified as bcc and amorphous phases respectively. Thus, the nanoscale bcc phase with a grain size of about 10 nm is surrounded by the remaining amorphous phase. The EDX profiles also indicate that the Hf content is

496

Properties

Figure 11.7 (a) High-resolution transmission electron microscopy (TEM) image, (b) and (c) energy-dispersive X-ray spectroscopy (EDX) profile and (d) and (e) electron energy loss spectroscopy (EELS) profiles taken from the small regions (1) and (2), respectively, with a diameter of 0.6 nm (EDX) and 3 nm (EELS) for amorphous Fe84Nb7B9 alloy annealed for 3.6 ks at 923 K.

enriched in the remaining amorphous phase. Figure 11.7 shows a highresolution TEM image, EDX and electron energy loss spectroscopy (EELS) profiles for the Fe84Nb7B9 amorphous alloy annealed for 3.6 ks at 923 K. This structural feature (that the nanogranular bcc-Fe particles with a size of about 12 nm are surrounded by the amorphous phase) is the same as that for the Fe—B alloy shown in Fig. 11.6. Similarly, the Nb is enriched in the remaining amorphous phase and no appreciable Nb is detected in the bcc particle. Furthermore, the EELS data indicate that B is also enriched in the amorphous phase. These results allow us to conclude that the structure consists of nanogranular bcc-Fe particles sur-

11: Magnetic Properties, Inoue et al.

497

rounded by the remaining amorphous phase and the the solute elements are significantly enriched in the amorphous phase. The enrichment is presumed to cause the increase in the thermal stability of the residual amorphous phase, leading to the maintenance of the nanogranular bcc-Fe mixed structure even in the high Ta range. The redistribution of the Nb and B elements is due to an extremely low solid solubility limit of these elements in the bcc-Fe phase. With the aim of clarifying the redistribution of the solute elements in more detail, the nanostructure was examined using atom-probe field ion microscopy techniques.[25] Figure 11.8(a) shows the change in the numbers of detected ions across the interface between bcc-Fe and amorphous phases for Fe90Zr7B3 alloy annealed for 3.6 ks at 723 K. Although the change near the interface is slight for Fe, the change in the numbers of detected Zr and B ions is significant, indicating that the concentrations of Zr and B differ significantly near the interface between the bcc-Fe and amorphous phases. Figure 11.8(b) shows the concentration profiles of Fe and Zr near the interface on the basis of the results shown in Fig. 11.8(a). It is seen that the bcc phase has higher Fe and lower Zr concentrations as compared with their nominal concentrations, while the remaining amorphous phase has lower Fe and higher Zr concentrations. Furthermore, notice that Zr is significantly enriched in the amorphous phase near the interface and has a steep concentration gradient. Because of the significant enrichment of Zr in the remaining amorphous phase near the interface, the increase in thermal stability becomes most significant in the amorphous phase region near the interface. As a result, preferential heterogeneous nucleation at the interface is suppressed, leading to the achievement of the nanoscale bcc structure. Thus, segregation of the element leading to an increase in thermal stability of the remaining amorphous phase is essential to the maintenance of the nanocrystalline structure in the high Ta range. It has been pointed out that the elements leading to the segregation satisfy the following three criteria:[25] (1) Large atomic size and high melting temperature with lower diffusivity (2) Low solid solubility limit in main constituent metal (3) Large negative heat of mixing compared to other constituent elements Makino et al. proposed[8–10] a mechanism for appearance of good soft magnetic properties for nanoscale bcc Fe—M—B alloys. Here, it is important to describe the mechanism because the information is thought

498

Properties

Figure 11.8 (a) Fractions of the numbers of Fe, Zr and B ions in the total numbers of detected ions and (b) concentration profiles of Fe and Zr elements in the region across the interface between amorphous and bcc-Fe phases for amorphous Fe90Zr7B3 alloy annealed for 3.6 ks at 723 K.

11: Magnetic Properties, Inoue et al.

499

to be useful in understanding a method for the improvement of soft magnetic properties. Good soft magnetic properties have been thought[6–11] to result from the simultaneous satisfaction of the following factors: (1) The achievement of high Bs resulting from magnetic coupling between the nanoscale bcc particles via the ferromagnetic amorphous phase (2) The ease of reversion of magnetization due to achievement of magnetic homogeneity resulting from the width of magnetic domain walls which are larger than the grain size of the bcc-Fe phase (3) The retainment of nanoscale bcc structure resulting from the existence of a residual amorphous phase where the solute elements are enriched and the thermal stability increases (4) The reduction of λs resulting from the redistribution of solute elements between bcc-Fe and the remaining amorphous phase. If this mechanism is correct, soft magnetic properties are expected to be further improved by modifications of the following four factors: (1) An increase in the Curie temperature (Tc) for the remaining amorphous phase (2) A decrease in the bcc grain size and an increase in the volume fraction of the bcc-Fe phase (3) An increase in the thermal stability of the remaining amorphous phase (4) An achievement of a nearly zero λs value.

11.4 Improvement of Soft Magnetic Properties by the Addition of Small Amounts of Solute Elements Improvement of the soft magnetic properties of Fe—M—B alloys has been tried by modifying the above-described four factors by adding small amounts of various solute elements. The addition of Co has been reported to be the most effective among VIa to VIII group transition metals. Figure 11.9 shows Bs, μe, Tc for the remaining amorphous phase, and D values as a function of Ta for (Fe0.985Co0.015)90Zr7B3 alloy, together with the data

500

Properties

Figure 11.9 Changes in the structure, (a) saturation magnetic flux density (Bs), (b) effective permeability (μe), (c) Curie temperature (Tc) of amorphous phase and (d) mean grain size (D) with annealing temperature (Ta) for amorphous (Fe0.985Co0.015)90Zr7B3 and Fe90Zr7B3 alloys.

for Fe90Zr7B3 alloy. Although no appreciable changes in crystallization behavior and D value are seen with the addition of Co, significant increases in Bs and μe occur, as well as the extension of the annealing temperature range leading to high Bs and μe. Notice also that high μe values above 2 × 104 are obtained in the wide Ta range between 823 and 923 K. The extension of the Ta range is also important from an engineering point

11: Magnetic Properties, Inoue et al.

501

Figure 11.10 Changes in (a) mean grain size (D), (b) effective permeability (μe) and coercivity (Hc) with heating rare (α) for amorphous Fe90Zr7B3, Fe89Hf7B4 and Fe84Nb7B9 alloys annealed for 3.6 ks at 923 K.

of view because the use of wide thermal treatment condition is possible. Considering that Tc for the amorphous phase increases significantly for the Co-containing alloy, the improvement of soft magnetic properties seems to result from an increase in the degree of magnetic coupling between bcc particles as a result of the increase in the magnetization of the residual amorphous phase containing Co.[9] The effect of grain size on the soft magnetic properties for the nanoscale bcc alloys has also been examined. Figure 11.10 shows changes in D and μe at 1 kHz and coercivity (Hc) with heating rate (α) up to 923 K for

502

Properties

Fe90Zr7B3, Fe89Hf7B4 and Fe84Nb7B9 amorphous alloys. With increasing α, μe increases by 2 to 10 times and Hc decreases by about 1.4 to 4 times, accompanied by a significant decrease in D. It is, therefore, concluded that decreasing the D value is effective for the improvement of soft magnetic properties. The decrease in D with increasing α has been interpreted[8] to originate from the increase in the number of nucleation sites and the suppression of grain growth. Rapid heating is a useful technique for the production of a nanocrystalline alloy with better soft magnetic properties. Besides, the use of a rapid heating technique gives us a unique opportunity to produce a nanocrystalline structure with significantly different D values. In the relation between the logarithms of Hc and D, the slope (n value) for the present bcc alloys has been measured to be 5.2,[8,9] which is nearly the same as the theoretically evaluated value of 6.0.[26] The agreement allows us to consider that the decrease in Hc occurs as a result of the increase in the ease of reversion of magnetization resulting from the decrease in D. Figure 11.11 shows Bs, Hc, μe, λs and D values as a function of Cu content for the bcc Fe90−xZr7B3Cux and Fe84−xNb7B9Cux alloys.[27] As Cu content increases from 0 to 2 at%, the Bs, Hc and D values decrease for both alloys and λs tends to increase. However, no systematic change in μe with Cu content is seen. From these changes, it is seen that the addition of only 1 at% Cu causes a decrease in D, accompanying a decrease in Hc. Similar decreases in D and Hc have also been recognized for Fe—Zr—B alloys containing 1 at% Pd.[8] From the atom probe field ion microscopic data, it has been confirmed[28] that the decrease in D resulting from the dissolution of Cu is due to easy heterogeneous nucleation of bcc phase in the Cu-rich region of the amorphous matrix. Furthermore, it is seen that λs is negative for the Fe—Zr—B—Cu alloys and positive for the Fe—Nb—B—Cu alloys. The opposite λs values indicate the possibility that the coexistence of Zr and Nb might causes nearly zero λs, leading to an improvement of soft magnetic properties. Figure 11.12 shows the relation between D, λs and μe or Hc for the bcc Fe—M—B and Fe—M—B—Cu (M = Zr and/or Nb) alloys. One can see a tendency for the highest μe and the lowest Hc values to be obtained around the slightly positive value of λs of 0.3 × 10−6. Also notice that the Fe—Zr—Nb—B—Cu alloys have a grain size of 7 nm, slightly positive λs of 0.3 × 10−6, high μe of more than 105 at 1 kHz and low Hc of less than 2 A/m. The value of Bs is also as high as 1.53 T. The simultaneous achievement of high μe and Bs, low Hc and nearly zero λs is the main result and exceeds those for all soft magnetic materials including Feand Co-based amorphous alloys and nanocrystalline Fe—Si—B—Nb— Cu[3] and Fe—P—C—Ga—Cu[29] alloys reported hitherto. The excellent soft magnetic properties are deduced to result from the combination of the

11: Magnetic Properties, Inoue et al.

503

Figure 11.11 Changes in (a) magnetic flux density (Bs), (b) coercivity (Hc), (c) effective permeability (μe), (d) saturation magnetostriction constant (λs) and (e) mean grain size (D) with Cu content for amorphous Fe90−xZr7B3Cux and Fe84−xNb7B9Cux alloys annealed for 1.8 ks at 723–923 K.

grain size refinement as a result of the addition of Cu and nearly zero λs as a result of dissolution of Zr and Nb.

11.5 Soft Magnetic Properties and Structure of Cu-free Quaternary Fe–Zr–Nb–B Alloys In the nanocrystalline ternary Fe—Zr—B and Fe—Nb—B alloys, the compositional range where μe shows a maximum does not strictly

504

Properties

Figure 11.12 Relation between mean grain size (D), saturation magnetostriction constant (λs) and (a) effective permeability (μe) or (b) coercivity (Hc) for bcc Fe—M—B(—Cu) (M = Zr and/or Nb) alloys.

coincide with the zero-λs line.[9] The best soft magnetic properties are obtained around the compositions of Fe90Zr7B3 and Fe84Nb7B9. Since λs changes from negative to positive with increasing B content for the alloys, λs of the Fe90Zr7B3 alloy is negative whereas that of the Fe84Nb7B9 alloy is positive. It is expected that the soft magnetic properties of the Fe—M—B alloys can be improved further by achieving zero-λs. The zero-magnetostrictive Fe—Zr—Nb—B alloys with mixed composition of Fe—Zr—B with negative λs and the Fe—Nb—B alloys with positive λs were studied. First, Zr, Nb and B concentrations were investigated by choosing Fe90Zr7B3 and Fe84Nb7B9 alloys as basic constituents and mixing them in various ratios.[12]–[14] Figure 11.13 shows the compositional dependence of Bs, μe, D and λs for the (Fe90Zr7B3)1−x(Fe84Nb7B9)x alloys as a function of x. The saturation magnetic flux density, D, and λs

11: Magnetic Properties, Inoue et al.

505

Figure 11.13 Compositional dependence of (a) saturation induction (Bs), (b) permeability (μe), (c) mean grain size (D) and (d) saturation magnetostriction constant (λs) for nanocrystalline (Fe90Zr7B3)1−x(Fe84Nb7B9)x alloys after annealing at the optimum conditions.

of the (Fe90Zr7B3)1−x(Fe84Nb7B9)x alloys show intermediate values between those of the Fe90Zr7B3 and the Fe84Nb7B9 alloys. However, μe for the (Fe90Zr7B3)1−x(Fe84Nb7B9)x alloys are inferior to those for the Fe90Zr7B3 and the Fe84Nb7B9 alloys. It is noted that μe shows a minimum around x = 0.8 where the alloy exhibits zero λs. The alloy with x = 0.83 results in strongly {001} textured microstructure in the free surface of the ribbon due to surface crystallization.[30] This causes inferior soft magnetic properties even though zero λs is obtained with this composition, because the magnetocrystalline anisotropy cannot be averaged out without random orientation of α-Fe nanocrystals.

506

Properties

Next, the effect of Zr + Nb amount on the soft magnetic properties was studied.[12]–[15] The best soft magnetic properties were obtained at Zr + Nb = 6 at%. Figure 11.14 shows the pseudo-ternary diagram of μe, Bs and λs for Fe–6%(Zr, Nb)—B alloys crystallized under optimum conditions, where the Zr + Nb amount was constant at 6 at%. A small grain size of 10–11 nm has been obtained in a compositional range of 0–3 at% Zr and 6–9 at% B. The permeability reaches a maximum value of 6.0 × 104 for the Fe85.5Zr2Nb4B8.5 alloy, which shows zero λs. Figure 11.15 shows the pseudo-ternary diagram of core loss (W) at 1.4 T and 50 Hz and Hc for Fe–6%(Zr, Nb)—B alloys crystallized under optimum conditions. The gray region indicates where the high μe values more than 5.0 × 104 have been obtained. Extremely low W (less than 0.09 W/kg) has been obtained in a compositional range of 1.5–2.2 at% Zr and 8–9 at% B. The compositional range where W exhibits the minimum extends to lower B content from the region where the best μe and Hc values have been obtained. This is due to increase of Bs with decreasing B content. Since W increases rapidly near magnetic saturation, the higher Bs is favorable to obtaining low W.

Figure 11.14 Pseudo-ternary diagram of permeability (μe), saturation magnetic flux density (Bs) and saturation magnetostriction constant (λs) for nanocrystalline Fe–6%(Zr, Nb)—B alloys after annealing at the optimum conditions.

11: Magnetic Properties, Inoue et al.

507

Figure 11.15 Pseudo-ternary diagram of core loss (W) and coercivity (Hc) for nanocrystalline Fe–6%(Zr, Nb)—B alloys after annealing at the optimum conditions. The gray area indicates the compositional range giving high permeability of more than 5 × 104.

11.6 Soft Magnetic Properties and Structure of Fe—Nb—B—P—Cu Alloys Produced in Air 11.6.1 Structure and Soft Magnetic Properties As described above, typical compositions of nanocrystalline Fe—M—B type alloys contain 7 at% M elements or 2 at% Zr and are produced by a melt-spinning technique in a controlled atmosphere (a vacuum or inert gases) because the M elements have high oxidation activity. The vacuum chamber surrounding the melt-spinning apparatus is necessary to produce typical Fe—M—B type alloys. This is a great disadvantage of the Fe—M—B type alloys for industrial applications. The production of Fe—Nb—B(—Cu) alloys is easier than for Fe—Zr(—Nb)—B(—Cu) alloys. However, Fe—Nb—B alloys exhibit the lowest Bs of about 1.5 T among typical Fe—M—B(—Cu) alloys. Therefore, the development of a new nanocrystalline soft magnetic alloy with higher Bs more than 1.5 T and with good productivity has been strongly desired because the simultaneous

508

Properties

achievement of both these properties enables the extension of application fields. We tried to synthesize new Fe—Nb—B-based nanocrystalline soft magnetic alloys with high Bs as well as good productivity. Melt-pinning the Fe—Nb—B alloys with various Nb contents in air has been tried. Ductile ribbons with metallic luster were obtained for alloys with 6.5 at% Nb or less.[15] Figure 11.16 shows the as-quenched structure and the compositional dependence of μe in the crystallized state of Fe— Nb—B(—Cu) alloys with 6 at% Nb as a function of B content.[15] For all the alloys, a B content of more than 10–11 at% is necessary to obtain a single amorphous structure in the as-quenched state. When the B content is less than this value, a mixed structure of α-Fe and amorphous phases is formed in the as-quenched state. The addition of Cu scarcely changes the glass-forming ability of the alloys. Figure 11.17 shows the XRD patterns taken from the free surface in the as-quenched state of Fe85−xNb6B9Cux (x = 0 − 1) alloys. The diffraction peak corresponding to α-Fe 110 accompanied by a halo from an amorphous phase is observed in the profile around 2θ ≈ 52° for all the alloys. The values for the mean grain size of the α-Fe phase (Dq) roughly estimated from the α-Fe diffraction peak are about 45, 25 and 50 nm for the Cu-free, 0.5 at% Cu-

Figure 11.16 Compositional dependence of effective permeability (μe) in crystallized state of Fe—Nb—B(—Cu) alloys with 6 at% Nb as a function of B content. The as-quenched structure evaluated from the X-ray diffraction patterns is also shown.

11: Magnetic Properties, Inoue et al.

509

Figure 11.17 X-ray diffraction patterns taken from free surface in as-quenched state of Fe85−xNb6B9Cux alloys. Mean grain size (Dq) of α-Fe phase estimated by broadening of the diffraction peak is also shown.

added and 1 at% Cu-added alloy, respectively. On the other hand, Dq of 0.1 at% Cu alloy is not estimated because the diffraction peak is too broad and unclear. Therefore, we can assume that Dq for the 0.1 at% Cu alloy is extremely small. This result indicates that the addition of 0.1 at% Cu to the Fe85Nb6B9 alloy should suppress coarsening of the α-Fe grains in the as-quenched state around the glass-formation limit. As shown in Fig. 11.16, a maximum value of μe for crystalline alloys without Cu and with 1 at% Cu is obtained at the compositions where a single amorphous structure in the as-quenched state is formed. On the other hand, for 0.1 at% Cu-added alloy, the maximum μe of 2.8 × 104 in the crystalline state is obtained at 9 at% B where a structure of an amorphous phase and extremely small α-Fe grains is formed in the as-quenched state. Therefore, it can be said that the addition of 0.1 at% Cu changes the as-quenched structure, and results in the expansion of the compositional range where good soft magnetic properties are obtained in the crystalline state to high Fe content. Next, the addition of P to Fe—Nb—B(—Cu) alloys with 6 at% Nb has been studied to improve the soft magnetic properties of the alloys.[15] Figure 11.18 shows the as-quenched structure and μe in the crystalline state of Fe—Nb—B(—P—Cu) alloys as a function of (B + P) content.

510

Properties

Figure 11.18 Compositional dependence of effective permeability (μe) in crystallized state of Fe—Nb—B(—P—Cu) alloys with 6 at% Nb as a function of B + P content. The as-quenched structure evaluated from the X-ray diffraction patterns is also shown.

The effective permeability is improved by replacing B by 1 at% P around 9–10 at% (B + P). The maximum value of μe is 4.1 × 104 obtained with crystalline Fe84.9Nb6B8P1Cu0.1 alloy. This alloy also exhibits a high Bs of 1.61 T, which is considerably higher than that of the Fe84Nb7B9 alloy. Figures 11.19 and 11.20 show the changes in the XRD patterns taken from the free surface in the as-quenched state for Fe85−xNb6B8P1Cux and Fe84.9Nb6B9−yPyCu0.1 alloys, respectively.[15,16] The diffraction peak corresponding to α-Fe 110 is observed in the profile for the alloys except for Fe84.9Nb6B8P1Cu0.1 alloy. The values of Dq are estimated to be about 20– 60 nm for the alloys. On the other hand, the as-quenched structure of Fe84.9Nb6B8P1Cu0.1 alloy evaluated from the XRD pattern seemed to be a single amorphous phase. Figure 11.21 shows (a) the TEM image, (b) the SAED patterns, (c) a high-resolution TEM image and nanobeam electron diffraction pattern of the as-quenched Fe84.9Nb6B8P1Cu0.1 alloy taken from the inner part of the ribbon. A large number of the nanoscale α-Fe grains are observed in all the TEM images.[16,17] The simultaneous addition of P and Cu to the Fe85Nb6B9 alloy makes a drastic change in the as-quenched structure including the nanoscale α-Fe grains. From TEM observation of the crystallized Fe84.9Nb6B8P1Cu0.1 alloy, it is clear that the nanocrystalline

11: Magnetic Properties, Inoue et al.

511

Figure 11.19 Changes of X-ray diffraction patterns taken from free surface in the as-quenched state of Fe85−xNb6B8P1Cux alloys as a function of Cu content. Values of mean grain size (Dq) of α-Fe phase estimated by the broadening of the diffraction peak are also shown.

Figure 11.20 Changes of X-ray diffraction patterns taken from free surface in the as-quenched state of Fe84.9Nb6B9−yPyCu0.1 alloys as a function of P content. Values of mean grain size (Dq) of α-Fe phase estimated by the broadening of the diffraction peak are also shown.

512

Properties

Figure 11.21 (a), (b) Transmission electron microscopy (TEM) images and selected-area electron diffraction patterns, (c) high-resolution TEM image and nanobeam electron diffraction pattern of as-quenched Fe84.9Nb6B8P1Cu0.1 alloy ribbon taken from the inner part of the ribbon.

structure of the alloy does not include the coarse α-Fe grains and is as uniform as that of the typical Fe84Nb7B9 alloy obtained by crystallizing a fully amorphous phase produced in a controlled atmosphere. Figure 11.22 shows the relation between μe in a crystallized state and Dq in an as-quenched state of the Fe85−xNb6B9−yPyCux (x = 0 − 1, y = 0 − 2) alloys.[16] The permeability in a crystallized state increases with decreasing

11: Magnetic Properties, Inoue et al.

513

Figure 11.22 Relation between effective permeability (μe) in a crystallized state and mean α-Fe grain size (Dq) in as-quenched state for Fe85−xNb6B9−yPyCux (x = 0 − 1, y = 0 − 2) alloys.

Figure 11.23 Transmission electron microscopy (TEM) images and selectedarea electron diffraction patterns taken from inner part of the ribbon of asquenched and crystallized (923 K for 300s) Fe85Nb6B9 alloy.

Dq. This result indicates that the precursor to the uniform nanostructure with high μe is not always a fully amorphous phase, and there is a possibility of realizing higher Bs materials with higher Fe contents.

11.6.2 Effect of Grain-Size Distribution and Curie Temperature of Intergranular Amorphous Phase on Soft Magnetic Properties Figure 11.23 shows TEM images and SAED patterns taken from the inner part of the ribbon of as-quenched and crystallized Fe85Nb6B9 alloy.[16, 31] The Fe85Nb6B9 alloy has a mixed as-quenched structure composed of an amorphous phase and α-Fe grains with 10–25 nm in size as

Properties

514

shown in Section 11.6.1. These grains grow to a size of about 40 nm during the annealing treatment, and remain in the nanostructure after crystallization as shown in Fig. 11.23(b). It is considered that such structural inhomogeneities will result in remarkable effects on the soft magnetic properties. Here, the effect of the grain-size distribution on the magnetic softness of nanocrystalline soft magnetic alloys is discussed based on the random anisotropy model (RAM).[26, 32] For simplicity, let us consider that the maximum grain size (Dm) does not exceed the exchange correlation length (Lex). Then the fluctuating part of the magnetocrystalline anisotropy (〈K1〉) is given by using a distribution function of the grain size ( f(D)), as[31] 2

K ⎧ Dm ⎫ K1 = (1 − v a ) 16 ⎨ ∫ D3 f ( D)dD⎬ , K0 ⎩ 0 ⎭ 2

Eq. (11-1)

where va is the volume fraction of the intergranular amorphous phase, K1 is the intrinsic magnetocrystalline anisotropy constant and L0 is the intrinsic exchange correlation length.[32] We further assume a log-normal distribution function as the grain-size distribution: f ( Dr ) =

2 1 ⎛ ln Dr ⎞ exp − , ⎝ 2σ 2D ⎠ 2 πσ D Dr

Eq. (11-2)

where Dr = D/D0 is the reduced grain size, D0 is the median and σD is the geometric standard deviation. If f(Dr) is negligibly small at Dr > Dm, then Dm can be regarded as infinity and we obtain Lex = L40 /{(1 − va) 〈D〉3exp(3σD2)} and 6

D⎞ K1 = (1 − v a ) K1⎛ exp(6σ 2D ), ⎝ L0 ⎠ 2

Eq. (11-3)

where 〈D〉 = D0exp(σD2/2) is the mean grain size. Then eqn (11.3) indicates that 〈K1〉 increases with increasing σD, i.e., soft magnetic properties of nanocrystalline alloys deteriorate with increasing σD, even if 〈D〉 is constant. It should be noted that this result is essentially established in other distribution functions.Naturally, the effective anisotropy in nanocrystalline alloys may have contributions from induced anisotropies such as magnetoelastic anisotropy other than the random magnetocrystalline anisotropy and hence the effective anisotropy constant in actual materials is more correctly K = K u2 + K1 2 ,[33] where Ku is the induced uniaxial anisotropy constant and K1 = (1 − v a ) K1

N = K1 (1 − v a )∫ D3 f ( D) dD L3ex

11: Magnetic Properties, Inoue et al.

515

(N is the number of grains in a magnetically coupled volume). Most of the nanocrystalline Fe—M—B (M = Zr, Hf, Nb) alloys with good soft magnetic properties exhibit a remanence ratio (Jr/Js) of around 0.5. This means that the magnetization process of the alloys is mostly governed by the induced anisotropies, i.e., Ku > 〈K1〉. In a limiting condition of K 2u >> 〈K1〉2 enables us to arrive at Lex = L0 K1 K u and K ≈ K u + (1 − v a )

3

Ku K1 ⎛ D ⎞ exp(3σ 2D ). 2 ⎝ L0 ⎠

Eq. (11-4)

The D3 behavior of Hc is observed for nanocrystalline Fe—Zr—B(–Cu) alloys with sufficiently small D.[33] The grain-size distribution evaluated by counting α-Fe grains in TEM images is shown in Fig. 11.24.[31] The log-normal distribution function reproduces well the observed grain-size distribution. In order to reproduce the grain-size distribution of the Fe85Nb6B9 alloy more accurately, we consider the bimodal distribution function ( fb(D)) expressed by superimposing the two log-normal distribution functions with different medians (D0 and dbD0, db > 1), the geometric standard deviations (σD and σb) and the ratio of the distribution function for the large grains to that of the small grains (rb).[31] When the induced anisotropies are dominant, the coercivity is given as Hc = 0.64(〈K 〉 − Ku)/Js,[34] where Js is the saturation magnetization. The calculated results are also shown in Fig. 11.24.[31] Here, va values are determined by X-ray diffraction profiles.[31] The calculated Hc with the unimodal log-normal distribution function for the Fe84Nb7B9 and Fe84.9Nb6B8P1Cu0.1 alloys are in good agreement with the measured ones. On the other hand, the calculated Hc for the Fe85Nb6B9 alloy is 2/3 as large as the experimental one. This large difference originates in disregarding the existence of the coarse grains with about 40 nm in size. The calculated Hc is consistent with the experimental one when the existence of the coarse grains is considered. Therefore, it can be said that our model explains well the dependence of Hc on the grain-size distribution for the nanocrystalline Fe—Nb—B(—P—Cu) alloys. These results also suggest that one should pay attention not only to the mean grain size but also to the grain-size distribution since inhomogeneity of the grain size increases Hc. vs. temperature plots for the crystallized (at Figure 11.25 shows J1/β s 823 K) Fe84Nb7B9 and the Fe84.9Nb6B8P1Cu0.1 alloys. Here, β (= 0.36) is the critical exponent for spontaneous magnetization.[35] It should be noted that the Fe84.9Nb6B8P1Cu0.1 alloy exhibits the higher Curie temperature for the intergranular amorphous phase (T cam) than the Fe84Nb7B9 alloy. In order

516

Properties

Figure 11.24 Grain-size distribution of Fe—Nb—B(—P—Cu) alloys. The ¯ ) were obtained from transmission electron histograms and mean grain size (D microscopy (TEM) images. The solid lines indicate fitting (a), (b) and (d) unimodal or (c) bimodal log-normal distribution functions. The inset in (c) is an enlarged view of 30–50 nm grain size. The fitting parameters and calculated results are also shown.

to discuss the effect of va and Tcam on the soft magnetic properties, the extended random anisotropy model (RAM) proposed by Löffler et al.[36] was applied. The model takes into account the reduction of the ferromagnetic exchange interaction between adjacent grains. Here, we assume that the crystalline and inter-granular amorphous phases have different exchange stiffness constants, i.e., Ac and Aa, respectively. Let us consider that Lex is determined by Ku instead of 〈K1〉, i.e., Lex = ϕ Aeff / K u , where ϕ is a parameter which reflects both the sym-

11: Magnetic Properties, Inoue et al.

517

Figure 11.25 J 1/β s vs. temperature plot for crystallized (at 823 K) Fe84Nb7B9 and Fe84.9Nb6B8P1Cu0.1 alloy, where J is the magnetization and β is the critical exponent for the spontaneous magnetization. The arrows indicate the Curie temperature of the inter-granular amorphous phase (T am c ).

metry of K1 and the total spin rotation angle over the exchange-correlated coupling chain,[37]

Aeff

AΛ Λ = Ac⎛1 + ⎞ ⎛1 + c ⎞ ⎝ ⎠ ⎝ D Aa D ⎠

−1

⎡ ⎫⎤ A ⎧ 1 ≈ − 1⎬⎥ 1+ c ⎨ 13 ⎢ 13 (1 − va ) ⎣ Aa ⎩ (1 − v a ) ⎭⎦ Ac

−1

Eq. (11-5)

is the effective exchange stiffness constant proposed by Löffler et al.[36] and Λ ≈ D{(1 − va)−1/3 − 1} is the thickness of the amorphous matrix.[38] Equations (11.4) and (11.5) yield K ≈ K u + (1 − v a )

32

Ku K12 D3 2ϕ 3 Ac3 2 32

⎡ ⎫⎤ Ac ⎧ 1 − 1⎬⎥ exp(3σ 2D ). ⎨ ⎢1 + 13 Aa ⎩ (1 − v a ) ⎣ ⎭⎦

Eq. (11-6)

518

Properties

Figure 11.26 Calculated coercivity (Hc) as a function of Curie temperature of the inter-granular amorphous phase (T cam). The solid circles show the relative value of observed Hc for Fe84Nb7B9 and Fe84.9Nb6B8P1Cu0.1 alloys annealed at 823 K.

Since the exchange stiffness constant is proportional to the Curie temperature and the square of the spin, Ac/Aa can be written as Ac/Aa ≈ Sc2T ccr/(Sa2T cam), where Tccr is the Curie temperature of the crystalline phase, Sc and Sa are the spin of the crystalline and inter-granular amorphous phases, respectively. Figure 11.26 shows the calculated Hc (relative value for Hc (va = 0) = 0.64 K u K12 D3 (2ϕ 3 Ac3/2 J s )) as a function of T cam. Here, Sa is assumed to be 1/2 because the magnetic hyperfine field of the intergranular amorphous phase is about 1/2 as large as that of the α-Fe phase.[37,39] The closed circles show the relative values of observed Hc, which are corrected for the difference of D and Js, for the Fe84Nb7B9 and Fe84.9Nb6B8P1Cu0.1 alloys annealed at 823 K. The coercivity increases with increasing va and decreasing T cam. Especially, the effect of T cam on Hc is considerably greater for the alloys with large va. Furthermore, the calculated results are in good agreement with the observed values. Therfore, it can be concluded that the small va and high Tcam of the Fe84.9Nb6B8P1Cu0.1 alloy are the main factors contributing to good soft magnetic properties even in the low Ta region around 800 K. Table 11.1 summarizes soft magnetic properties of Bs, μe, Hc, λs, electrical resistivity (ρ) and core losses (W) for the bcc Fe—M—B and Fe— M—B—Cu alloys, together with the data of sample thickness (t) and D, in comparison with those for amorphous Fe—Si—B[40] and Co—

18 20 20 19 22 21 20 18 22 10 6 19 14 24 19 20 19 18 19 22 20 21 20 18 1.48 1.47 1.52 1.61 1.61 1.57 1.53 1.64 1.56 0.88 1.28 1.24

10 9 8 10 9 8 8 11

10

13 13 10

1.67 1.63 1.64 1.64 1.70 1.62 1.63 1.59 1.50 1.55

Bs (T)

16 16 12 16 12

D (nm)

me*

29 000 49 000 41 000 110 000 160 000 120 000 60 000 10 000 70 000 85 000 100 000

38 000

27 000 29 000 34 000 27 000 48 000 34 000 30 000 32 000 36 000 30 000

* f = 1 kHz, ** Wa/b is the core loss at a × 10−1 T and b Hz, *** Ref. (3).

Fe91Zr7B2 Fe90Zr7B3 Fe89Zr7B3Cu1 (Fe0.985Co0.015)90Zr7B3 (Fe0.98Co0.02)90Zr7B2Cu1 (Fe0.995Ni0.005)90Zr7B3 Fe89Zr7B3Pd1 Fe89Hf7B4 Fe84Nb7B9 Fe84Nb7B9 Fe84Nb7B9 Fe83Nb7B9Ga1 Fe83Nb7B9Ga1 Fe83Nb7B9Ge1 Fe83Nb7B9Cu1 Fe84.9Nb6B8P1Cu0.1 Fe86Zr3.25Nb3.25B6.5Cu1 Fe85.6Zr3.3Nb3.3B6.8Cu1 Fe84Zr3.5Nb3.5B8Cu1 Fe85.5Zr2Nb4B8.5 Fe78Si9B13 Co70.5Fe4.5Si10B15 Fe73.55Si13.3B9Nb3Cu1 Fe73.55Si13.3B9Nb3Cu1***

t (mm)

5.6 3.8 4.7 2.0 1.2 1.7 3.5 3.5 1.2 1.1 0.5

4.8

5.5 4.2 4.5 4.2 4.2 3.5 3.2 4.5 7.0 7.6

Hc (A/m)

+2.1

0

−0.3 −0.3 +0.3 −0.1

+0.2 +1.1

−1.2 +0.1

−1.1 −1.1 0 −0.1

ls (10-6)

1.15

1.37 1.47

0.56 0.56 0.61

0.69 0.64

0.70

0.48 0.58

0.53

0.44 0.51

r (mWm)

0.05 0.06 0.09 0.28

0.11

0.22

0.14 0.14

0.12 0.08

0.21

W14/50** (W/kg)

166.0 62.0 49.4 39.1

60.0 49.0 58.7

59.0 72.7 27.5 20.8 47.0 35.4 69.2 54.7

79.7 185.4 63.7 80.8

W2/100 k** (W/kg)

Table 11.1 Sample thickness (t), mean grain size (D), electrical resistivity (r), and magnetic properties (Bs: saturation magnetic flux density, me: effective permeability, Hc: coercivity, ls: saturation magnetostriction constant, and W: core loss) for the nanogranular bcc Fe—M—B(—Cu) alloys, Fe—Si—B—Nb—Cu alloys and amorphous alloys

11: Magnetic Properties, Inoue et al. 519

520

Properties

Figure 11.27 Relation between magnetic flux density (Bs) and effective permeability (μe) for bcc Fe—M—B alloys. The data for other soft magnetic materials are also shown for comparison.

Fe—Si—B[40] and nanoscale bcc Fe—Si—B—Nb—Cu[3] alloys. It is confirmed that soft magnetic properties combined with high Bs and low W for the present nanoscale bcc alloys are superior to those for the other magnetic materials. Based on the data summarized in Table 11.1, the relation between Bs and μe for soft magnetic materials is shown in Fig. 11.27, where the data of silicon steels and Mn—Zn ferrites are also presented for comparison. It is seen that the Fe—M—B type alloys possess much better soft magnetic properties which had not been acieved for all other soft magnetic materials.

11.7 Improvement of High-frequency Permeability by the Dissolution of Oxygen in the Surrounding Amorphous Phase 11.7.1 As-sputtered Structure As described above, the permeability (μ′) of the nanocrystalline Fe— M—B alloys (t = 20 μm) is as high as 105 at 1 kHz, but decreases to about 2 × 103 at 1 MHz (7 × 102 at 3 MHz). For the future development of highperformance and miniaturized electronic devices, it is important to

11: Magnetic Properties, Inoue et al.

521

improve the high-frequency permeability characteristics. It is generally known that an increase in electrical resistivity and a decrease in sample thickness are effective for the improvement of high-frequency permeability because these changes cause a decrease in eddy current loss which becomes significant in a high frequency range. As one of the methods to increase the electrical resistivity, the B content in Fe—M—B alloys was increased to about 20 at%. However, the increase in B content causes the disappearance of the nanocrystalline bcc structure by the simultaneous precipitation of bcc and compound, though an amorphous single phase is obtained in the melt-spun state.[40] Subsequently, Hayakawa et al.,[41] examined the effect of the addition of oxygen on the formation and soft magnetic properties of nanogranular bcc alloys for Fe—M base alloys. As an effective method to dissolve a large amount of oxygen and to prepare a thin sample, we used a sputtering method in a mixed atmosphere of argon and oxygen. It has been reported[18,19] that as-deposited Fe46−88Hf2−20O7−41 films have four types of structure which consist of a bcc single phase, mixed bcc and amorphous phases, an amorphous single phase and an oxide phase, depending on the film composition. Useful soft magnetic properties are obtained in the films which have mixed bcc and amorphous phases. Figure 11.28 shows the XRD patterns of as-deposited Fe—M—O [M = Ti, Zr, Hf, V, Nb, Ta, W, rare earth metals (Re)][42] films. In the Zr-, Hf- and Re-containing alloys, a broad peak due to an amorphous phase and diffraction peaks due to a bcc phase are observed. In contrast to these results, only the diffraction peaks corresponding to a bcc phase are observed for the Ti-, V-, Ta-, Nb- and W-containing alloys and no appreciable broad peak is seen. Furthermore, one can notice that the diffraction angle of bcc 110 peak shifts to a lower angle than that of pure bcc-Fe marked with a dashed line in Fig. 11.28, indicating that the bcc phase includes a larger amount of M and oxygen. Figures 11.29 and 11.30 show high-resolution TEM images, nanobeam electron diffraction patterns and EDX profiles for the as-deposited Fe55Hf11O34 and Fe49Hf16O35 films, respectively, together with data of electrical resistivity at room temperature (ρRT). The electron diffraction patterns and the EDX profiles were taken from the points marked in the figure. Both films are composed of very fine-grained crystals less than 10 nm in diameter, which were surrounded by the amorphous phase. Grains with diameter less than 5 nm in size are fewer and the region of amorphous phase becomes larger for Fe49Hf16O35 film compared with Fe55Hf11O34 film. These crystals are identified as bcc-Fe phase supersaturated with Hf and O from the nanobeam diffraction pattern and the EDX profile of the crystal (region 1). The ρRT value increases from 10.5 to 492 μΩm with an increase in the amorphous region of the film.

522

Properties

Figure 11.28 X-ray diffraction patterns of as-deposited Fe—M—O films.

Figure 11.31 shows the X-ray photoelectron spectroscopy (XPS) profiles of Fe2p3/2, Hf4f7/2, Y3d5/2 and Ta4f7/2 for as-deposited Fe55Hf11O34, Fe68Y22O10 and Fe55Ta18O27 films. Each profile is indicated with solid triangles in the figure. In all systems, the binding energy of Fe2p3/2 agrees with that of pure Fe. In contrast, the binding energies of Hf4f7/2 and Y3d5/2 are close to those of HfO2 and Y2O3, respectively. Therefore, O preferentially combines with Hf and Y in the Fe55Hf11O34 and Fe68Y22O10 films consisting of bcc and amorphous phases, as shown in Figs. 11.28 and 11.29. Then, these elements are mainly dissolved in the amorphous phase and probably form M-oxide like structure. In contrast, for the Fe55Ta18O27 film consisting of a mostly single bcc phase, the binding energy of Ta4f7/2 is close to that of metallic Ta and there is no evidence of chemical binding

Figure 11.29 High-resolution transmission electron microscopy (TEM) images, nanobeam electron diffraction patterns and energy-dispersive X-ray spectroscopy (EDX) profiles taken from each microregion for as-deposited Fe55Hf11O34 film.

Figure 11.30 High-resolution transmission electron microscopy (TEM) images, nanobeam electron diffraction patterns and energy-dispersive X-ray spectroscopy (EDX) profiles taken from each microregion for as-deposited Fe49Hf16O35 film.

Properties

524

Figure 11.31 X-ray photoelectron spectroscopy (XPS) profiles for as-deposited Fe55Hf11O34, Fe68Y22O10 and Fe55Ta18O27 films.

between Ta and O elements. They are presumed to be supersaturated in the bcc phase, which results in the diffraction peak shift of bcc 110 as shown in Fig. 11.28. As a result, it is considered that the rapid increase in ρRT results from the highly resistive amorphous region including M—O atomic pairs for the Fe—(Hf, Zr, Re)—O films. The formation of the mixed structure is due to a combination of three factors: (1) A low solid solubility limit of Hf in the bcc-Fe phase (2) Peferential interaction of oxygen with Hf (3) Large solubility of oxygen in an amorphous phase.

11.7.2 Magnetic Properties Figure 11.32 shows the compositional dependence of Bs and Hc of as-deposited Fe—Hf—O films which were sputtered under no applied

11: Magnetic Properties, Inoue et al.

525

Figure 11.32 Compositional dependence of saturation magnetic flux density (Bs) and coercivity (Hc) for as-deposited Fe—Hf—O films.

magnetic field. Open and solid circles represent the single bcc phase and the single amorphous phase, respectively. Double open and half solid circles indicate the oxide phase and the mixed bcc and amorphous phases, respectively. The value of Bs tends to decrease with increasing Hf and O contents and has a ridge around Hf content of 10 to 15 at%. The value of Hc decreases with increasing Hf and O contents and has a valley around the same Hf content as the ridge of Bs. Therefore, the ridge of Bs approximately agrees with the valley of Hc, and there is a region in which Bs above 1.0 T and Hc below 160 A/m are simultaneously obtained. In this region, the real part of the initial permeability (μ′) is about 4 × 102 even in the as-deposited state and the film structure is composed of fine bcc and amorphous phases as indicated by shading. Figure 11.33 shows the temperature dependence of Bs for as-deposited Fe—Hf—O films in comparison with a Fe—Hf amorphous alloy film. The Fe—Hf—O films with the mixed amorphous and bcc structure exhibit two-stage crystallization behavior. The arrows indicate the first crystallization temperature corresponding to grain growth of the bcc phase,

526

Properties

Figure 11.33 Temperature dependence of saturation magnetic flux density (Bs) for as-deposited Fe76Hf24, Fe55Hf11O34 and Fe49Hf16O35 films. The arrows indicate the first crystallization temperature (Tx) of the films.

which was measured by DSC. The value of Bs for both films decreases with increasing temperature with bending points in the course of changes, at 500 and 750 K for the Fe49Hf16O35 and Fe55Hf11O34 films, respectively. An amorphous Fe76Hf24 film has a low Tc (below room temperature) owing to the Invar effect.[43] For the Fe49Hf16O35 film, the bending point is thought to result from the Tc of the amorphous phase not from crystallization, because the temperature is lower than that of the first crystallization. It is difficult to conclude certainly that the bending point is attributed to the Tc of the amorphous phase for the Fe55Hf11O34 film, because the temperature of the bending point is close to the first crystallization temperature. However, there is no bending point below that temperature, and the Tc of the amorphous phase for the Fe55Hf11O34 film can be regarded as higher than 700 K. As a consequence, the Tc of the amorphous phase including Fe and Hf increases with the dissolution of O in the Fe—Hf—O films. Furthermore, we confirmed[44] that the Tc of the amorphous phase for the Fe—Hf—O films increased after annealing. The magnetic properties and ρRT values for as-deposited Fe—M—O (M = group IVa–VIa transition metals and Re) films are summarized in Table 11.2, together with their film structures. The film compositions are the same as those of the films shown in Fig. 11.28. In all systems, Bs above 0.9 T and high ρRT above 4 μΩm are simultaneously obtained. The λs values were 0.1–2.9 × 10−6. Relatively low Hc values below 400 A/m are obtained in M = Hf, Zr and Re systems, which have mixed structure of nanogranular bcc and amorphous phases.

11: Magnetic Properties, Inoue et al.

527

Table 11.2 Magnetic properties (Bs: saturation magnetic flux density, Hc: coercivity), electrical resistivity (r), and film structure for as-deposited Fe—M—O films

Fe56Ti15O29 Fe65Zr9O26 Fe55Hf11O34 Fe59V15O26 Fe64Nb12O24 Fe64Ta13O23 Fe67W13O20 Fe68Y22O10 Fe75Ce15O10 Fe81Nd10O9 Fe67Dy7O26

Bs (T)

Hc (A/m)

r (mWm)

1.2 1.3 1.2 1.2 1.3 1.3 1.3 0.9 1.2 1.4 1.3

1040 211 154

6.4 6.6 9.1 5.6 4.1 1.7 8.3 22.6 5.9 5.2 9.3

168 658 207 116 286 322

Structure bcc amorphous amorphous bcc amorphous bcc bcc amorphous amorphous amorphous amorphous

+ bcc + bcc + bcc

+ bcc + bcc + bcc + bcc

As shown in Fig. 11.28, no amorphous phase is formed in the Ti-, V-, Ta- and W-containing alloys. In these alloys, Hc is relatively large and sufficient soft magnetic properties are not obtained. Therefore, the formation of a mixed structure consisting of the nanoscale bcc and amorphous phases is required for the achievement of good magnetic softness in the as-deposited state. The soft magnetic properties in as-deposited Fe—(Hf, Zr, Re)—O films are probably due to the nanoscale grain size and the intergrain ferromagnetic coupling through the high-Tc amorphous phase as shown in Fig. 11.33, which averages out the magnetocrystalline anisotropy of bcc-Fe phase.[44] Furthermore, it should be noted that good soft magnetic properties are obtained even in the films including rare earth (Re) elements, which have a large magnetocrystalline anisotropy and usually inhibit soft magnetic properties. This is due to the decrease in magnetic interaction between Fe and Re resulting from the preferential binding with O. The soft magnetic properties of the Fe—(Hf, Zr, Re)—O films are improved by sputtering under a uniaxial magnetic field, or uniaxial field annealing (UFA) treatment after deposition under no applied magnetic field. Furthermore, we tried to improve the frequency characteristics by enhancing uniaxial anisotropy (HK) by the addition of Co to the

528

Properties

Figure 11.34 Magnetization curves for as-deposited Co44.3Fe19.1Hf14.5O22.1 film and Fe61Hf13O26 film after uniaxial field annealing (UFA) at 673 K for 10.8 ks.

Fe—M—O films. Figure 11.34 shows the B—H curve for an as-deposited Co44.3Fe19.1Hf14.5O22.1 film, together with the data for the Fe61Hf13O26 film. The former was deposited under a static magnetic field and the latter was UFA treated after deposition under no applied magnetic field. The HK value of 1.15 kA/m for the Fe61Hf13O26 film is the largest of all the Fe— Hf—O films. However, the HK of the Co44.3Fe19.1Hf14.5O22.1 film exhibits 4.8 kA/m which is approximately four times larger than that of the Fe61Hf13O26 film. This film exhibits a high Bs of 1.1 T and a high ρRT of 15.1 μΩm simultaneously, and moreover, the angle dispersion of magnetic anisotropy becomes small. Therefore, excellent high-frequency characteristics are expected to be obtained for the Co44.3Fe19.1Hf14.5O22.1 film owing to its significant high ρRT and large HK values. Figure 11.35 shows the frequency dependence of μ and the quality factor (Q = μ′/μ″) of the Fe—Hf—O and Co—Fe—Hf—O films prepared by various methods, together with the data on other metallic soft magnetic alloy films developed to date. The Q value is a very important factor from

11: Magnetic Properties, Inoue et al.

529

Figure 11.35 Frequency dependence of the real part of initial permeability (μ′) and the quality factor (Q = μ′/μ″) for Fe62Hf11O27 film (as-deposited), Fe61Hf13O26 film (after uniaxial field annealing (UFA) at 673 K for 10.8 ks) and Co44.3Fe19.1Hf14.3O22.1 film (as-deposited) compared with the other soft magnetic films reported.

the viewpoint of applications. Actually, conventional soft magnetic films exhibit high permeability at 1 MHz, in particular, approximately 104 has been obtained for nanocrystalline soft magnetic films such as Fe—Si— Al—Hf—C film.[45] However, μ′ decreases with increasing frequency because of low ρRT values. On the other hand, the μ′ of the Fe—Hf—O films is lower than that of conventional films in the range below 30 MHz, while in the frequency range higher than that, the Fe—Hf—O films exhibit higher and flat μ’ characteristics over 100 MHz because of their high ρRT values and moderate HK. Fe62Hf17O27 film exhibits a Bs of 1.3 T and a high μ′ of 1.4 × 103 at 100 MHz in an as-deposited state. Moreover, the Q values

530

Properties

are also higher than those of the conventional films. The highest Q value of 61 at 100 MHz is obtained for the Co44.3Fe19.1Hf14.5O22.1 film. Needless to say, these films have higher Q values than other metallic films even in a low-frequency range. In the other M systems, similar high-frequency characteristics are observed, so that the Fe—M—O and Co—Fe—Hf—O films can be regarded as low-loss soft magnetic films in a wide frequency range from MHz to around GHz.

11.8 Applications Figure 11.36 summarizes expected application fields for the soft magnetic Fe—M—B(—Cu) type alloys, together with the magnetic characteristics which are required for their applications. Application fields include power transformers, data communication interface components, electromagnetic interference (EMI) prevention components, magnetic heads, sensors, magnetic shielding and reactors. The expectation of applications to power transformers comes from the lower core losses over a wide maximum induction range as compared with oriented Si-steels and amorphous Fe78Si9B13 alloy as shown in Fig. 11.37.[9,10] In addition, the efficiency of the power transformer was examined as a function of output

Figure 11.36 Magnetic characteristics and application fields for the bcc Fe— M—B(—Cu) (M = Zr and/or Nb) type alloys.

11: Magnetic Properties, Inoue et al.

531

Figure 11.37 Core losses at 50 Hz as a function of maximum induction for bcc Fe—Zr—Nb—B—Cu alloy. The data for amorphous Fe78Si9B13 and oriented Si-steel are also shown for comparison.

current for the bcc Fe—Zr—Nb—B—Cu alloys and clarified to be much higher than those for the Fe—Si—B amorphous alloys and oriented Sisteels (Fig. 11.38),[10] in agreement with the tendency for core losses. We have also evaluated noise attenuation characteristics which are important for the common mode choke coil. Figure 11.39 shows that the bcc Fe— Zr—Nb—B—Cu alloys had better noise attenuation values over the whole frequency range when compared with Fe—Si—B amorphous alloys.[10] The better performance as power transformers and common mode choke coils allows us to expect that the newly developed nanocrystalline Fe—M—B alloys are practical soft magnetic materials. Finally, it is important to note that the Fe—Zr—Nb—B—Cu alloys have already been used as pulse transformers in ISDN terminal adapters. Figure 11.35 also summarizes typical application items and characteristics required for their applications for Fe—M—O and Co—Fe—Hf—O films. In the frequency range around 10 MHz, we postulate thin-film inductors or transformers for microswitching converters[46] for portable electrical equipment. Microswitching dc–dc converters using Co-based amorphous alloy film as the core material in thin-film inductors have

532

Properties

Figure 11.38 Change in the efficiency of a power transformer with output current for bcc Fe—Zr—Nb—B—Cu alloy, amorphous Fe—Si—B alloy and Si steels.

already been reported,[47] but they will be operated at higher frequency for further miniaturization and improvement of power supplies. For such high-frequency switching converters operating around 10 MHz, the Fe— M—O films exhibit their ability as a core material. In the high frequency range up to 100 MHz, the Fe—M—O films are useful for magnetic heads for ultrahigh-density recording exceeding 1 Gbit/in2. High Bs soft magnetic films such as Fe—M—N[48] are studied for high-density recording heads, but high-frequency characteristics around 100 MHz are required. In the further high-frequency range around GHz, noise filters, thin-film transformers or other micromagnetic devices dealing with electromagnetic waves are proposed. For other examples of applications, Figs. 11.40 (a) and (b) show a schematic illustration of a planar inductor for microswitching converters and the frequency dependence of Q (= ωL/R) values of a planar Cu coil using each magnetic film, respectively. The inductor is made up of a planar coil sandwiched with two magnetic films facing each other as shown in Fig. 11.40 (a). In the case of the inductor, we use one side of the Cu coil

11: Magnetic Properties, Inoue et al.

533

Figure 11.39 Change in the noise attenuation of a common mode choke coil with frequency for bcc Fe—Zr—Nb—B—Cu alloy and amorphous Fe—Si—B alloy.

as a conductor. As can be seen in Fig. 11.40 (b), an inductor using Fe— Hf—O film shows a maximum. High Q values of 12.3 at 5 MHz and 21.8 at 20 MHz are obtained for an inductor using the Co—Fe—Hf—O film. Therefore, these planar inductors using the Fe—Hf—O and Co—Fe— Hf—O films enable higher frequency operation and higher efficiency than an inductor using Co—Ta—Hf films for microswitching converters because of the loss characteristics of the magnetic films.

11.9 Conclusions The development of new soft magnetic materials with excellent characteristics of high Bs of more than 1.5 T combined with high μe of more than 105 at 1 kHz was achieved by nanocrystallization of Fe—Zr— Nb—B—Cu amorphous alloys. The Cu-free quaternary Fe85.5Zr2Nb4B8.5 alloy which exhibits high Bs of 1.64 T and high μe of 6.0 × 104 at 1 kHz simultaneously, was also developed. In addition, it was found that the simultaneous addition of 1 at% P and 0.1 at% Cu to the Nb-poor Fe85Nb6B9 alloy sesults in good magnetic properties of high Bs of more than 1.6 T

534

Properties

Figure 11.40 Schematic illustration of (a) planar inductor and (b) frequency dependence of quality factor (Q = ωL/R) of the inductor using Fe—Hf—O, Co—Fe—Hf—O and Co—Ta—Hf amorphous films.

and high μe of more than 4 × 104 at 1 kHz as well as good productivity. Furthermore, the dissolution of oxygen in the remaining amorphous phase caused significant improvement of high-frequency permeability of 103 in the frequency range 1 to 100 MHz by a drastic increase in ρRT to 103 μΩm. Considering that these characteristics had not been previously reported, nanostructure control is concluded to be a useful method for the development of new types of high functional materials. The future progress of nanostructure-controlled materials is expected to enable the fabrication of new materials exhibiting useful characteristics which have not been achieved for conventional materials.

References 1. Masumoto, T., Kimura, H., Inoue, A., and Waseda, Y., Mater. Sci. Eng., 23:141–144 (1976).

11: Magnetic Properties, Inoue et al.

535

2. Croat, J.J., Herbst, J.F., Lee, R.W., and Pinkerton, F.E., J. Appl. Phys., 55:2078–2082 (1984). 3. Yoshizawa, Y., Oguma, S., and Yamauchi, K., J. Appl. Phys., 64:6044–6046 (1988). 4. Hasegawa, N., and Saito, M., J. Mag. Soc. Jpn., 14:313–318 (1990); IEEE Trans. J. Magn. Jpn., 6:91–100 (1991). 5. Inoue, A., Kobayashi, K., Nose, M., and Masumoto, T., J. Phys. (Paris), Colloq. C8:831–834 (1980). 6. Suzuki, K., Kataoka, N., Inoue, A., Makino, A., and Masumoto, T., Mater. Trans., JIM, 31:743–746 (1990). 7. Suzuki, K., Makino, A., Inoue, A., and Masumoto, T., J. Appl. Phys., 70:6232–6237 (1991). 8. Makino, A., Inoue, A., and Masumoto, T., Nanostruct. Mater., 6:985–988 (1995). 9. Makino, A., Inoue, A., and Masumoto, T., Mater. Trans., JIM, 36:924–938 (1995). 10. Makino, A., Inoue, A., Hatanai, T., and Bitoh, T., Mater. Sci. Forum, 235–238:723–728 (1997). 11. Makino, A., Hatanai, T., Inoue, A., and Masumoto, T., Mater. Sci. Eng., A226/228:594–602 (1997). 12. Makino, A., Bitoh, T., Kojima, A., Inoue, A., and Masumoto, T., J. Magn. Magn. Mater., 215–216:288–292 (2000). 13. Makino, A., Bitoh, T., Kojima, A., Inoue, A., and Masumoto, T., J. Appl. Phys., 87:7100–7102 (2000). 14. Makino, A., Bitoh, T., Kojima, A., Inoue, A., and Masumoto, T., Mater. Sci. Eng., A304–306:1083–1086 (1997). 15. Makino, A., Novel Nanocrystalline Alloys and Magnetic Materials, (B. Cantor, ed.), pp. 260–277, Institute of Physics Publishing, ristol and Philadelphia (2004). 16. Makino, A., Bitoh, T., Inoue, A., and Masumoto, T., Scr. Mater., 48:869–874 (2003). 17. Makino, A., and Bitoh, T., J. Appl. Phys., 93:6522–6524 (2003). 18. Makino, A., and Hayakawa, Y., J. Jpn. Inst. Metals, 57:1301–1309 (1993). 19. Makino, A., and Hayakawa, Y., J. Magn. Soc. Jpn., 18:411–414 (1994). 20. For example, Naka, M., Masumoto, T., and Chen, H.S., J. Phys., (Paris), Colloq. C8:839–842 (1980). 21. For example, Nose, M., and Masumoto, T., Sci. Rep. RITU, A28:232–241 (1980). 22. For example, Forester, D.W., Vittoria, C., Schelleng, J., and Lubitz, P., J. Appl. Phys., 49:1966–1968 (1978). 23. Suzuki, K., Makino, A., Tsai, A.P., Inoue, A., and Masumoto, T., Mater. Sci. Eng., A179/180:501–505 (1994). 24. Suzuki, K., Makino, A., Inoue, A., and Masumoto, T., J. Jpn. Inst. Metals, 57:964–971 (1993). 25. Zhang, Y., Hono, K., Inoue, A., Makino, A., and Sakurai, T., Acta Metall., 44:1497–1510 (1996). 26. Herzer, G., IEEE Trans. Magn., 26:1397–1402 (1990). 27. Makino, A., Bitoh, T., Inoue, A., and Masumoto, T., J. Appl. Phys., 81:2736–2739 (1997).

536

Properties

28. Hono, K., Zhang, Y., Inoue, A., and Sakurai, T., Mater. Trans., JIM, 36:909–917 (1995). 29. Fujii, Y., Fujita, H., Seki, A., and Tomida, T., J. Appl. Phys., 70:6241–6243 (1991). 30. Wu, Y.Q., Bitoh, T., Hono, K., Makino, A., and Inoue, A., Acta Mater., 49:4069–4077 (2001). 31. Bitoh, T., Makino, A., Inoue, A., and Masumoto, T., Mater. Trans., 44:2011–2019 (2003). 32. Herzer, G., Scr. Metall. Mater., 33:1741–1756 (1995). 33. Suzuki, K., Herzer, G., and Cadogan, J.M., J. Magn. Magn. Mater., 177–181:949–950 (1998). 34. Bozorth, R.M., Ferromagnetism, pp. 811–837, Van Nostrand, New York (1951). 35. Le Guillou, J.C., and Zinn-Justin, J., Phys. Rev. B, 21:3976–3998 (1980). 36. Löffler, J.F., Braun, H.B., and Wagner, W., J. Appl. Phys., 85:5187–5189 (1999). 37. Suzuki, K., and Cadogan, J.M., Phys. Rev. B, 58:2730–2739 (1998). 38. Hernando, A., Vázqauez, M., Kulil, T., and Prados, C., Phys. Rev. B, 51:3581–3586 (1998). 39. Kim, C.S., Kim, S.B., Lee, J.S., and Noh, T.H., J. Appl. Phys., 79:5459–5461 (1996). 40. Smith, C.H., Rapidly Solidified Alloys, (H.H. Liebermann, ed.), pp. 617–663, Marcel Dekker, New York (1993). 41. Hayakawa, Y., Makino, A., Inoue, A., and Masumoto, T., Sci. Rep. RITU, A42:115–119 (1996). 42. Makino, A., and Kojima, A., J. Magn. Soc. Jpn., 17:814–819 (1993). 43. Saito, N., Hiroyoshi, H., Fukamichi, K., and Nakagawa, Y., J. Phys., F16:911–919 (1986). 44. Hayakawa, Y., and Makino, A., Nanostruct. Mater., 6:985–988 (1995). 45. Hasegawa, N., Saito, M., and Makino, A., J. Mgn. Soc. Jpn., 18:750–758 (1994). 46. Hayakawa, Y., Makino, A., Fujimori, H., and Inoue, A., J. Appl. Phys., 81:3747–3752 (1997). 47. Mino, M., Tsukamoto, K., Yanagisawa, K., Tago, A., and Yachi, T., APEC96 Proceedings, pp. 422–426 (1996). 48. Ishiwata, N., Wakabayashi, C., and Urai, H., J. Appl. Phys., 69:5616–5618 (1991).

12 Mechanical Behavior of Nanocrystalline Metals Julia R. Weertman Northwestern University, Evanston, IL, USA

12.1 Introduction The production of metals and alloys with grain size in the range of a few nanometers to about 50–100 nm led to expectations of materials with extremely high strength. The empirical Hall–Petch equation[1,2] predicts that σy = σ0 + k

d

Eq. (12-1)

where σy is the yield strength, σo is a friction stress below which dislocations will not move in the material in the absence of grain boundaries, k is a constant and d is the grain size. However, most measurements seem to indicate that the strength of nanocrystalline metals falls well below that predicted by an extrapolation of Eq. (12-1) to very fine grain sizes. Not only does k, the slope of a Hall–Petch plot, drop below the coarse-grain value with increasing grain refinement, but in some cases it even appears to become negative. Many and varied are the models of deformation in nanocrystalline metals that have been proposed to account for this fall-off in strengthening. Representatives of a number of different types of models are described in the first section of this chapter. Molecular dynamics (MD) computer simulations have become an invaluable source of information on deformation mechanisms and mechanical behavior of nanocrystalline metals. Such information is provided on an atomic level. The second part of the section on modeling and computer simulations is devoted to learning what insight MD simulations can give about deformation processes in nanocrystalline metals. A great spread exists in the strength data reported for the same material by various laboratories, or even by the same laboratory over a period of time. The second section of the chapter is devoted to characterization of the nanocrystalline material used in the mechanical Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 537–564 © 2007 William Andrew, Inc.

537

538

Properties

measurements, particularly characterization of the defects present. Finally, the results of various mechanical measurements are reported. The results presented in this chapter refer to “bulk” material, i.e., the sample dimensions in all directions are large compared to grain size. Further, it is assumed that strain gradient effects are absent.

12.2 Models and Computer Simulations of Mechanical Behavior of Nanocrystalline Materials 12.2.1 Models of Deformation The usual textbook explanation of Eq. (12-1) was given by Cottrell,[3] who calculated the stress ahead of a large array of dislocations piled up at a grain boundary. The length of the pileup is taken to be of the same order of magnitude as the grain size d (Fig. 12.1a). When the stress from the pileup at a Frank–Reed source in the adjacent grain rises to the value needed to activate this source, yielding will spread across the grain boundary and, thence, throughout the material. It is obvious that Eq. (12-1) cannot be extrapolated to arbitrarily small grain sizes. At very small values of d, the pileup containing the large numbers of dislocations assumed in the Cottrell theory requires applied stresses approaching or exceeding the theoretical strength. A number of models have been proposed to estimate the limit of applicability of Eq. (12-1). Nieh and Wadsworth[4] used experimental measurements of the yield strength of several nanocrystalline metals to calculate the smallest

Figure 12.1 Two models proposed to explain Hall–Petch behavior. Models due to (a) Cottrell and (b) Li.

12: Mechanical Behavior, Weertman

539

grain size that could support two dislocations, the minimum number to constitute a “pileup”. Strength measurements in materials with a grain size below this critical value would be likely to remain constant or even drop with increasing grain refinement.[4] More detailed calculations, based on dislocation pileups, of the dependence of yield stress on grain size in finegrained material have been made by Armstrong and colleagues.[5–8] Not all explanations of the strengthening effects of grain refinement are based on dislocation pileups at grain boundaries. Guided by observations that dislocations are seen to emanate from grain boundaries in the very early stages of deformation, whereas dislocation pileups are usually absent except in low stacking fault or ordered alloys, Li[9] and Li and Chou[10] proposed that Hall–Petch behavior can be explained by dislocations generated at grain boundary ledges during deformation (Fig. 12.1b). The flow stress inside a grain can be expressed as proportional to the square root of the dislocation density, ρ. The value of ρ is taken to be proportional to the number of grain boundary ledges per unit volume, assumed to scale with the grain boundary area per unit volume and, thus, with d −1, leading to the familiar Hall–Petch relationship. It is not clear how this model would extrapolate to very small grain sizes. For example, the relationship between flow stress and dislocation density must break down when there are, at most, only a few dislocations per grain. Meyers and Ashworth[11] proposed a model of strengthening by grain refinement that is also based on the generation of dislocations at grain boundaries. In this model, as the material deforms, elastic anisotropy results in stress concentrations in the boundary regions that give rise to the generation of dislocations that effectively form a hardened reinforcing second-phase network. General yielding occurs when the applied stress becomes sufficient to push dislocations through the hardened layers in the grain interior. A judicious assumption on the relationship between the thickness of the hardened layer on d leads to Hall–Petch behavior at large grain sizes and a lowering of the Hall–Petch slope at very small grain sizes. The model predicts that the maximum in σy occurs at a grain size of about 10 nm in Cu and Fe[12] but the model probably is not appropriate for such small grain sizes. Many of the models of the mechanical behavior of fine grain materials are based on a two-component or multi-component concept. In a few[11–13] the components are taken to be the hardened regions adjacent to the grain boundaries and the relatively soft grain interiors. However in most models, the two components are the grain boundaries themselves and the crystalline grains, for example, refs 14–17. Carsley et al.[14] assume Hall–Petch behavior for the crystalline grain interiors of nanocrystalline metals down to the finest grain sizes considered. Grain boundaries are

Properties

540

considered amorphous. Use is made of the relationships that the hardness of an amorphous metal is about 1/6 of its shear modulus, and that this modulus is about one half of the crystalline value. Fu et al.[15] assume that the grain interiors in a nanocrystalline metal are single crystals, while the grain boundary regions have a lower yield stress but higher strain hardening rate. Estrin and colleagues[16,17] model the crystalline grain interiors as deforming by a combination of dislocation glide plasticity, Herring–Nabarro[18] and Coble[19] creep. Mass transport occurs in the grain boundaries through stress-driven vacancy flow. In all of these models softening with decreasing grain size sets in below a critical grain size. A problem in two-component models is the choice of the effective width of the phase associated with the grain boundary, whether it is that of the grain boundary itself as in refs 14–17, or the hardened region of the grains adjacent to the boundaries.[11–13] In the former case, Gleiter and colleagues[20,21] argued that the grain boundaries in nanocrystalline materials are essentially different from those in conventional materials; they are of low density and in a gas-like state of disorder. A number of direct observations in the electron microscope, e.g. refs 22, 23, have indicated that the grain boundaries are similar in nanocrystalline and coarse-grain metals. Values ranging from 0.5–1 nm, or about 2–5 atomic widths, are frequently chosen in two-phase deformation models. Masumura et al.[24], in an important paper, proposed a two-component model that takes into account the dispersion in grain sizes that characterize real nanocrystalline samples (as opposed to model material that conveniently is made up of grains of a single size). These authors assume that the stress, σhp = σ − σo, required to deform those grains in the sample that are larger than a certain critical size, d*, follow Hall–Petch behavior, i.e., σhp = σ − σo = kd −1/2, whereas the strength, σc, of the grains smaller than d* derives from a Coble creep[19] mechanism. The expression for σc is taken to be: σc = A/d + Bd3. The threshold term A/d, which is suggested by experimental results, is presumed to be related to the stress, ≈Gb/d, required by a dislocation loop pinned at the grain boundary nodes to climb. The vacancies created and destroyed in the course of the climbing are needed in the Coble diffusion process. (Here, G is the shear modulus and b is the Burgers vector of the dislocation.) The dispersion in grain size in the nanocrystalline metal is assumed to follow a log-normal distribution, which has been found to give a good description of actual distributions.[25,26] The yield

12: Mechanical Behavior, Weertman

541

stress corresponding to a given average grain size, 〈d〉, is found by a simple volume averaging of the grains, using Hall–Petch behavior for grains greater than d* and the Coble relationship for d < d*. The authors introduce a parameter, p = (A/d*)/B(d*)3, which indicates the relative value at the transition grain size, d*, of the threshold stress compared to the usual grain boundary diffusional creep term. Figure 12.2 shows a plot of normalized stress versus normalized average grain size, 〈d〉, for various values of p (and a fixed value for the standard deviation of the log-normal distribution). It can be seen that, with a suitable adjustment of parameter p, this model predicts that the slope of a Hall–Petch plot at small grain sizes levels off and can become negative. The influence of a dispersion in grain size on the strength of nanocrystalline metals has been considered in several papers.[24–27] A deformation model, with assumptions appropriate to many actual nanocrystalline samples, has been developed by Morita and colleagues.[27] In the case of high purity samples, there usually are few, if any, grains small enough that grain boundary sliding processes are dominant.[26,28] Therefore it was assumed that, for a given applied stress, all grains larger than a critical size (determined by the applied stress) undergo plastic deformation while the smaller grains remain elastic. Morita et al. take into account the internal stresses that arise when some grains deform elastically while others undergo plastic deformation. The micromechanics of inclusions[29,30] were used to solve this problem. An interesting result of the calculations is the strong influence of the width of the grain size dispersion on overall

Figure 12.2 Plot of normalized stress vs ξ−1/2 for various values of the parameter p. ξ is the normalized average grain size 〈d 〉/d*. See text for explanation of symbols. (From Ref. [24].)

542

Properties

strength. Figure 12.3 shows calculated stress–strain curves for a fixed average grain size and varying widths of the spread in grain sizes. It can be seen that the apparent 0.2% offset yield strength is strongly affected by the magnitude of the dispersion. In the preceding models it was tacitly assumed that dislocations, when they enter into the picture, are exactly the same as the dislocations in largegrain material. However it is evident that as the grain size begins to approach the core radius the standard dislocation theory must be modi-

Figure 12.3 The effect of the size of the standard deviation Sinv on (a) the grain size distribution expressed in volume fractions and (b) the stress–strain curves. The mean grain size dm is constant at 20 nm. (From Refs [26], [27].)

12: Mechanical Behavior, Weertman

543

fied. Scattergood and Koch[31] emphasized the drop in dislocation line tension at very small grain sizes in their model to explain negative Hall–Petch slopes. The authors argue that, at large grain sizes, cutting of dislocation forests is the easiest way for glide dislocations to get past these obstacles. But at sufficiently small sizes, the expression for dislocation line tension, T, T = (Gb2/4π)ln R/ro

Eq. (12-2)

becomes sensitive to the value of R, the long-range stress field screening distance. (Here, ro is the dislocation core cutoff radius.) The authors scale R with the grain size d. As a result, the stress required for a glide dislocation to get past the obstacles by Orowan looping, which is proportional to T, becomes comparable to the cutting stress at a critical grain size and is the favored mechanism at still smaller grain sizes. The continued drop of the line tension with decreasing grain size leads to a negative Hall–Petch slope. When the theory is fitted to limited experimental data, values of R, the screening distance, generally are found to be between d/2 and d/4. Attempts have been made by the authors of the various models to validate their calculations by comparing predicted grain size dependence of some measure of strength (often hardness) with experimental data. Experimental results are so diverse that any such comparison is highly suspect.

12.2.2 Molecular Dynamics Computer Simulations Invaluable insight on deformation processes and mechanical behavior in nanocrystalline metals has been gained from molecular dynamics (MD) computer simulations.[28,32,33] The behavior of computer-generated samples with grain sizes approaching those of experimentally available material now can be analyzed and the results compared with observations. Samples also can be made with grain sizes that are smaller than can be produced at present, so that deformation processes at grain sizes <∼10 nm can be studied. Computer-generated samples can be subjected to a wide variety of test parameters to establish their mechanical properties. MD simulations reveal deformation mechanisms as they operate on an atomic scale, information that largely is impossible to obtain from experiments. For example, MD simulations indicate dislocations move across a grain with a speed of about 0.4 nm/psec, about 1/10 the speed of sound.[34] Thus with

544

Properties

current techniques it is impossible to capture dislocations in flight by TEM. MD simulations do have limitations that should be kept in mind when the results are interpreted. Because of the massive number of calculations involved, usually only the first fraction of a nanosecond of deformation can be studied. Very high strain rates/stresses must be used to quickly reach a state that appears to approach steady-state conditions. Evidently ordinary diffusion-based deformation, such as Coble[19] or Herring–Nabarro[18] creep would not show up in an MD computer simulation, nor would stress-induced microstructural changes such as grain growth.[35] Grain-size distributions in computer samples are narrow in contrast to the dispersion invariably seen in real samples, no matter how such samples are synthesized. Grain-size dispersion has a strong influence on deformation processes and strength in nanocrystalline metals.[24–27] Further, the computer samples are free of flaws and impurities, unlike real nanocrystalline samples. Therefore MD computer simulations should be taken as guidelines for experiments and their interpretation, and not as completely accurate predictors of results. Nevertheless the simulations have revealed interesting aspects of mechanical behavior and deformation mechanisms in nanocrytalline metals. As grain size decreases into the nanocrystalline regime, dislocation generation in the grain interiors becomes suppressed. One of the most important revelations of MD simulations is the observation that, as slip ceases to originate in the grain interiors, dislocations are nucleated in the grain boundaries. Grain boundaries emit partial dislocations that travel across grains, finally being absorbed in opposite boundaries. The process is accompanied by atomic shuffles and stress-assisted free volume migration at the grain boundaries.[36] The whole picture is heavily influenced by the structure of the boundaries and their associated stress distributions. Figure 12.4 illustrates this sequence of events in nanocrystalline Al. In Fig. 12.4(a) a partial dislocation is emitted at a triple junction and starts to move across the grain. The black atoms represent two successive layers with HCP packing, i.e., a stacking fault, left behind by the partial. Subsequent views show that a second partial has been emitted and as it also moves across the grain the stacking fault is removed. Eventually both partials (which together constitute a perfect dislocation) are absorbed into the opposite grain boundary. Whether a grain boundary dislocation source emits only a single partial and thus leaves an extended stacking fault behind as it traverses the grain, or whether a trailing partial eventually follows, is an undecided question. An obvious factor in determining emission of a trailing dislocation would seem to be the stacking fault energy (SFE).[37] However for the empirical potentials used, only extended par-

12: Mechanical Behavior, Weertman

(a)

(b)

(c)

(d)

545

Figure 12.4 Nucleation and growth of two partial dislocations emitted from a grain boundary in nanocrystalline Al. In the sequence from (a) to (d), a partial dislocation is emitted at a triple junction and starts to move across the grain. The black atoms represent two successive layers with HCP packing, i.e., a stacking fault, left behind by the first partial. A second partial is subsequently emitted, and as it also moves across the grain the stacking fault is removed. Eventually both partials (constituting a perfect dislocation) are absorbed into the opposite grain boundary. (Courtesy of H. Van Swygenhoven.)

tials were seen in the case of the nanocrystalline material with the highest SFE (nickel). An explanation for this seeming contradiction may be found in the short times and high stresses used in the simulations, which emphasize the role of the unstable stacking fault energy.[38] Taking the unstable SFE into account leads to an explanation for the difference in dislocation character seen in the different materials simulated.[38,39] One of the earliest and most persistent topics of interest in the study of deformation behavior in nanocrystalline metals is the change in strength as the grain size d approaches zero. The limiting case as d → 0 is a metallic glass. Although pure metals such as Cu or Ni have never been prepared

Properties

546

in the amorphous state, it is believed that the strength of such a metallic glass is lower than that of its crystalline counterpart as its grain size approaches zero. Thus it appears that at some point Hall–Petch strengthening must be replaced by softening as the grain size decreases. Experiments have been ambiguous on this subject, as will be seen later in the chapter. As mentioned above, MD computer simulations are well suited to studying deformation behavior in very small grain-size metals and thus shedding light on the cause of grain refinement softening at very small values of d. In an early paper Van Swygenhoven, Spaczer and Caro[34] studied the response to a fixed stress of a series of Cu and Ni samples with grain sizes varying from 3.4 to 12 nm. The strain rate was found to increase with decreasing grain size, indicating that the grain-size dependence of strength is opposite to the usual Hall–Petch behavior. While Hall–Petch softening was observed there is no clear indication of a transition from softening to strengthening with increasing grain size. In a 2003 paper in Science[28] Schiøtz presented the results of MD simulations of the stress–strain behavior of Cu samples with average grain size ranging from 4.7 to 48.6 nm. By considering such a broad distribution in grain sizes he was able to observe a maximum in strength. Figure 12.5(a) shows the stress–strain curves for the various grain sizes. In black and white it is impossible to tell which grain size goes to which curve, but general shapes can be distinguished. The initial peaks in some of the curves are believed

3 A Flow stess (GPa)

Time stress (GPa)

2

1.5 1

0.5 0 0

B

2.4

2.5

2.2 2 1.8 1.6

0.02

0.04 0.06 True strain

0.08

0.1

0

10

20

30

40

50

Grain diameter (nm)

Figure 12.5 The grain-size dependence of the flow stress. (a) Stress–strain curves for 10 simulations with grain sizes ranging from 4.7 to 48.6 nm. (b) Flow stress (defined as the average stress in the strain interval between 7 and 10% deformation) as a function of grain size. The error bars indicate fluctuations in this strain interval (1 standard deviation). A maximum in the flow stress is seen for Cu with grain sizes between 10 and 15 nm. (From Ref. [28].)

12: Mechanical Behavior, Weertman

547

to be a consequence of the high strain rate used (5 × 108/s). It is difficult to determine the yield point from the curves because clearly the elastic modulus E changes with grain size. However all the curves achieve a more or less constant stress with increasing strain after about 6% strain, and it is these “flow stresses” that are used in Fig. 12.5(b). A clear maximum in the flow stress is seen for Cu with grain size between 10 and 15 nm. The extremely high values of the flow stresses are believed to be a consequence of the high strain rate used in the simulations.[28,40] Extensive dislocation activity, including even dislocation pile-ups at grain boundaries, was seen in simulations of the larger grain-size samples after deformation. Van Swygenhoven and colleagues have studied the deformation mechanisms operating in nanocrystalline metals at very small grain sizes. Grain boundary structure has been observed to be independent of grain size for a given misorientation of the grains across the boundary.[41] However it is to be expected that during synthesis of metals with grain sizes in the nanocrystalline range, the increased triple junction density will impose different distributions of the angle of misorientation between adjacent grains, resulting in misorientations with a larger number of grain boundary dislocations (GBDs).[42] Since GBDs assist dislocation emission from the grain boundaries into the grains, their number will influence the relative importance of dislocations in the deformation process.[42] As grain size drops further the GBDs become more delocalized and thus less effective in promoting dislocation emission while climb of the GBDs is encouraged by local atom shuffling and free volume migration.[36] However, again it must be kept in mind that artifacts stemming from the short simulation times may have been introduced. It is to be expected that the structure of the grain boundaries after sample creation could be further relaxed if more simulation time were available. MD simulations show that the elastic modulus E decreases at very small grain sizes.[34,40] Schiøtz et al.[40] also examine the effect of sample porosity on measured values of E, as well as the possible influence of creep on modulus measurements. Both of these effects are discussed later in the chapter.

12.3 Characterization of Nanocrystalline Metals Before valid comparisons can be made between measurements of mechanical properties of nanocrystalline metals and the predictions of

548

Properties

various models, it is necessary to be sure that the data represent the inherent behavior of the material and are not compromised by sample imperfections. Most methods for producing nanocrystalline samples tend to introduce various flaws in the materials such as pores and microcracks, high internal stresses and impurities and trapped gases. While not exactly a flaw, the broad dispersion in grain size of many samples has a strong effect on the overall mechanical properties, as seen in the previous section, and must be taken into account when comparing experimental results with theoretical expectations. It is evident that careful characterization of the samples to be used in measurements of mechanical properties is essential. A second factor compromising mechanical test results is the small size of most nanocrystalline samples. Inadequate measurement techniques, especially in strain measurements, led to erroneous results that were only corrected as the techniques improved.

12.3.1 Density, Pores and Microcracks In the early days of nanocrystalline research, density measurements of samples often gave values ranging from only about 70% to >90% of the single crystal density.[20,43] The density shortfall was interpreted variously as the consequence of nanocrystalline grain boundaries having extremely low densities[20] or as caused primarily by the presence of pores.[44] Since porosity is known to have a strong effect on the elastic moduli and on other mechanical properties as well, it is important to have knowledge of the size and number density of any pores in the mechanical test specimens. Schaefer and colleagues,[45,46] using positron spectroscopy, identified three size classes of voids in nanocrystalline samples. The smallest are about a lattice vacancy in size and are presumed to be located at grain interfaces. The largest are identified with “missing grain” pores and are of the order of the grain size. The middle-sized voids correspond to clusters of about 10 vacancies and have been identified by Schaefer[45] as residing at grain triple junctions. It is likely that the two larger classes make the major contribution to the density shortfall in many samples. Van Petegem and colleagues[47] used electron microscopy and positron lifetime spectroscopy to examine voids in nanocrystalline Ni samples made by three different synthesis methods: electrodeposition, inert gas condensation consolidated at room temperature and high pressure torsion. Positron spectroscopy showed the presence of nanovoids of 12–25 vacancies in all three types of samples.

12: Mechanical Behavior, Weertman

549

The smallest void size detectable by small-angle neutron scattering (SANS) overlaps the useful upper range of positron spectroscopy and continues up to the submicrometer range. Thus SANS can be a useful tool for obtaining void size distributions in nanocrytalline materials. Figure 12.6[48] illustrates the use of this technique in assessing the effectiveness of two improvements to inert-gas condensation (IGC) equipment.[49] SANS becomes insensitive to pores or flaws much above 0.1 μm. Larger flaws, which are especially damaging to tensile strength properties, can result from imperfect bonding during compaction between agglomerations of nanocrystalline powders produced by IGC and compaction. An example of a potential crack nucleation site is shown in Figure 12.7.[50] The “roll-ups” (Fig. 12.7a) shown in this polished and etched transverse section through a nanocrystalline Cu specimen produced by IGC and compaction probably come from the rolls of nanocrystalline powder scraped from the cold finger during the evaporation. The TEM micrograph of part of a roll-up (Fig. 12.7b) shows that the compaction operation was not able to create a perfect bonding between its layers. Trapped gas in nanocrystalline samples can stabilize voids and flaws against sintering under pressure, even at elevated temperatures.[50] Further improvements to the vacuum conditions and helium purity in the IGC equipment made subsequent to the samples of Figures 12.6 and 12.7 have resulted in a significant decrease in the flaw population, but some flaws are still present. Even polished and etched transverse sections of electrodeposited Ni show some defects and large pores, though much less than the IGC Cu.[51]

Figure 12.6 Effect of synthesis conditions and compaction temperature on the pore volume fraction in nanocrystalline Pd. Open symbols are from a sample made before improvements to the vacuum in the synthesis chamber and filled symbols are from samples after the improvements, compacted either at room temperature (circles) or Tm /4 (triangles). (From Ref. [48].)

550

Properties

Figure 12.7 (a) SEM micrograph of a polished and etched transverse section of a nanocrystalline Cu sample made by IGC and compaction showing “roll-ups” (see text). (b) TEM micrograph of an ion beam-etched sample similar to that in (a) showing the spiral morphology of a roll-up. Different nanocrystalline grain sizes are apparent on either side of the roll-up boundary. (From Ref. [50].)

Grain size plays an important role in the theories and MD computer simulations of the mechanical properties of nanocrystalline metals. Obviously an accurate determination of the grain size (and grain size dispersion) of samples used in mechanical tests is required for interpretation of the results. Most measurements of grain size are done by X-ray diffraction or TEM. X-ray diffraction yields a global result, as well as a value for the root mean square (RMS) strain. TEM has the advantage of providing information on the distribution of grain sizes as well as an average

12: Mechanical Behavior, Weertman

551

value, but only for a limited region. Such data are usually given in terms of a grain size distribution and average value based on number density, but a volume density better reflects the influence of large grains, even if they are few in number. A comparison of average grain size and RMS strain as obtained by different methods of analysis of the X-ray data is given by Jiang, Rühle and Lavernia.[52] A heavy concentration of twins can lead to complications in the grain-size measurements, or even to the meaning of “grain size” for the particular application.[26] There have been long-standing concerns about the stability of the internal structure of nanocrystalline metals. The energy associated with the high density of grain boundaries provides a driving force for grain growth. Even in the absence of external stress, an increase in grain size has been observed over time in samples of high purity nanocrystalline metals held at room temperature.[26,53] Stress can greatly accelerate grain growth in a nanocrystalline material, especially if the purity is high. In a series of elegant in-situ TEM measurements of grain growth in ultra-fine grain and nanocrystalline aluminum, Jin et al.[54] examined grain stability under the tip of a nanoindenter. In some cases smaller grains were swallowed up by larger neighbors on either side. Alternatively, two adjacent grains were seen to coalesce, a process involving intense dislocation activity followed by the grain boundary sweeping across the smaller grain. Grain rotation and grain agglomeration are believed to have occurred under the nanoindenter tip in nanocrystalline grains. Each of these events was completed in a span of a few seconds or less. A study of grain growth in high purity nanocrystalline Cu[55] in the stress field of a Vickers microhardness indenter demonstrates the instability of the grain structure under stress. Figure 12.8 shows the evolution of the grain size distribution with increasing dwell time of the indenter. The initial distribution more than doubled in size in less than 10 seconds. This grain growth results in a significant softening of the material.[56] Similar experiments carried out at temperatures close to that of liquid nitrogen show an even faster rate of softening than occurs at room temperature. Analysis of the evolving grain size distributions at ∼80 K reveals that besides the “orderly” rapid grain coarsening such as seen at room temperature, a small number of very large (400–700 nm) grains have formed whose appreciable volume fraction contributes to the rapid softening. This extensive grain growth at cryogenic temperatures indicates that the growth is stress-driven.Introduction of small amounts of impurities/solutes can significantly slow grain growth.[55,57] Evidently, grain size distributions should be examined at the end of a mechanical test as well as at the beginning, to ensure that the properties measured can be identified with a given internal structure, not one that evolved as the test proceeded.

Properties

552 (a)

0.4

grain size distribution away from indents

(b) 0.4

0.2

0.2

0.0 0.2

0.0 Dwell time 10 sec RT

20 µm

0.4

0.0 0.2

Dwell time 60 sec RT

0.1 0.0 0.2 Dwell time 5 mins RT

0.1

Dwell time 10 sec RT

0.2 0.0 0.4

Dwell time 60 sec RT

0.2 0.0

0.4

Dwell time 5 mins RT

0.2 0.0

0.0 0.2

Dwell time 30 mins RT

0.1 0.0

Volume fraction

Number fraction

0.1

grain size distribution away from indents

0.4

Dwell time 30 mins RT

0.2 0

80 160 240 320 400 Grain size (nm)

0.0 0

80 160 240 320 400 Grain size (nm)

Figure 12.8 (a) Number fraction and (b) volume fraction grain-size distributions in the initial condition and after various indenter dwell times for a high purity Cu sample. The optical image in the inset shows that the measurements were taken from the indent tip region. (From Ref. [55].)

Nanocrystalline metals can be synthesized by a variety of methods, leading to some differences in the final product such as dislocation density, impurities, RMS internal strain, flaw content etc. Although not usually considered as a subject for characterization such as the items in the preceding sentence, the history of thermal treatment after synthesis also can impact mechanical behavior. It has been observed that in certain cases heat treatment of as-prepared nanocrystalline metals not only leads to grain growth but also to significantly increased strength.[58–60] A study of several pairs of nanocrystalline Pd samples showed ∼50% increase in hardness of an annealed sample over the as-prepared materials with the same grain size.[59] An MD computer simulation study[61] of the effects of annealing on the internal structure of a nanocrystalline metal gives an explanation of this strengthening effect.

12: Mechanical Behavior, Weertman

553

12.4 Mechanical Behavior Early experimental studies of the mechanical properties of nanocrystalline metals tended to concentrate on determining the dependence of strength on grain size, with the hope of gaining information on deformation processes operating at very small grain sizes. While such experiments are still of great interest, now that the luster of enhanced strength in nanocrystalline metals is clouded by their brittle behavior, many studies have as their goal devising methods to increase ductility and fracture toughness. The elastic behavior of nanocrystalline metals also has been studied, though to a lesser extent than their plastic properties.

12.4.1 Elastic Properties of Nanocrystalline Metals Early studies[20,43] of the elastic behavior of nanocrystalline metals indicated that their elastic moduli are likely to be only a fraction of the coarsegrain values. In general, the first samples were far from fully dense, containing pores and other flaws. Poor measurement techniques on the small samples also contributed to errors in the modulus measurements.[43] Kristic et al.[62] obtained a value for Young’s modulus E in 7 nm grain-size Ni-1.5%P that is similar to the coarse-grain modulus. These authors attributed the measured shortfalls in E described above to the presence in the samples of pores with cracks growing out of them. In both Refs [20] and [43], the samples were made by compacting nanocrystalline powders obtained by IGC to less than 100% density whereas the Ni-1.5%P material was produced by electrodeposition and reported to be fully dense. Young’s modulus was measured in a series of high density IGC Cu and Pd nanocrystalline samples containing various small amounts of porosity.[63] Measurements were done using a pulse echo technique operating at 50 MHz. Figure 12.9 shows that the extrapolation to zero porosity yields modulus values in good agreement with the usual values. Cao et al.[64] found normal bulk values for E for 17 nm Fe and 41 nm Ti using high frequency (10 MHz) acoustic microscopy. Shen et al.[65] employed a nanoindentation technique to investigate the elastic moduli in nanocrystalline Fe, Cu and Cu—Ni alloys made by mechanical milling and mechanical alloying. The nanoindentation was carried out on individual particulates in compacts, away from interpowder porosity. The average grain size in the Cu, Ni and Cu—Ni samples ranged from 17 to 26 nm, and down to 7 nm in the ball-milled iron. Only

554

Properties

Figure 12.9 Young’s modulus as a function of porosity for nanocrystalline Cu and Pd. (From Ref. [63].)

Figure 12.10 Calculated ratios of Young’s (E) and shear (G) moduli of nanocrystalline material to those of polycrystals (Eo, Go) as a function of grain size d. The dashed and solid curves correspond to a grain boundary thickness of 0.5 and 1.0 nm, respectively. The open circles show the E/Eo values of nanocrystalline Fe vs grain size. The horizontal dotted-dashed line represents the E/Eo and G/Go ratios for materials having an infinite grain size. (From Ref. [65].)

at the smallest grain sizes was an appreciable deviation from the coarsegrain value found and even in the case of the 7 nm Fe, this deviation was only 5%. Shen et al.[65] modeled the grain size dependence of E on the basis of a two-component system consisting of crystalline grains and interfacial regions (see Section 12.2.1). A comparison of the model’s prediction and experimental results is shown in Figure 12.10. Relaxation effects

12: Mechanical Behavior, Weertman

555

in nanocrystalline metals at room temperature have been noticed by a number of investigators.[55,56,66–68] The modulus measurements described above that were carried out at high frequencies obviously give unrelaxed values, but other techniques could yield results involving some degree of relaxation, especially in the case of high purity samples. Indeed, in a study of the elastic properties of 100% dense nanocrystalline gold, Sakai et al.[68] found a strain rate effect in the measured values of E. Below about 220 K, values obtained from a vibrating reed technique are similar to results from stress–strain curves, and the temperature dependence of both is similar to that of ordinary gold. Above 220 K, the vibrating reed results continue to track the coarse-grain values, but modulus values from the stress–strain curves, obtained at a very low strain rate, decrease with increasing temperature at a much faster rate. A stress–strain plot taken at 77 K at a strain rate of 10−4/s is linear over a wide strain range, but at 293 K the plot continuously curves away from the initial slope, indicating the presence of creep. The authors conclude that experimental values of E in nanocrystalline material depend on the strain rate of the measurements. Sample impurities and solutes are likely to lessen this strain rate effect.

12.4.2 Hardness, Yield and Ultimate Strengths The great strength and high hardness of nanocrystalline metals are major factors in attracting interest in this class of materials. Nanocrystalline Cu has been synthesized with a yield stress of around 800 MPa and ultimate tensile stress in excess of 1100 MPa[69] whereas for a wellannealed high-purity coarse-grain Cu sample these quantities would be of the order of 100 MPa. Nanocrystalline Ni, with average grain size of 26 nm, reaches a tensile srength greater than 2.2 GPa.[70] In the past, collections of experimental data on hardness or yield stress of a nanocrystalline metal as a function of grain size in a Hall–Petch plot (e.g. Ref. [24]) showed data points all over the figure. With better samples and improved measuring techniques the situation has improved. Figure 12.11 is a Hall–Petch plot of yield stress vs 1 d for Cu. Samples were synthesized by IGC and compaction,[56,71] cryogenic and room temperature ball milling,[69,72] and surface mechanical attrition treatment.[73] Measurements included compressive and tensile yield stresses and hardness tests. It can be seen that the points align quite well on a single curve. Thus yield stress in samples made by IGC (at least with density greater than 97% of theoretical full density, as is the case here), by ball milling, or by surface attrition appears to be largely independent of synthesis method. It is rather

Properties

556 1000

Strength (MPa)

800

1/3 Hardness Compression Tensile

600 400 200 0 0.0

0.1

0.2 -1/2

Grain size

0.3

-1/2

(nm )

Figure 12.11 Hall–Petch plot for nanocrystalline Cu. Yield stresses taken from tension and compression stress–strain curves and hardness tests on samples made by ICG, cryogenic/room temperature ball-milling and surface mechanical attritrition treatment. See text for details.

remarkable that the alignment of the points on a common curve is as good as it is, given that the use of “average” grain size (determined by different measuring techniques) masks the different degrees of grain-size dispersion of the various samples. The data show a bending over away from the straight line starting around a grain size of 25 nm. The slope of the straight line (k in Eq. 12-1) in Fig. 12.11 is the same as that of the Hall–Petch plot for coarse-grain copper. Figure 12.11 should be compared with the results of the MD simulation shown in Fig. 12.5. A deviation from Hall–Petch behavior also was observed in electrodeposited Ni (Fig. 12.12).[74,75] A negative slope is seen for Ni with a grain size less than ∼11 nm. However several of the authors of Refs [74] and [75] later showed that the measured value of the hardness of the 6 nm sample probably was lowered by rapid room temperature creep, leading to an apparent Hall–Petch softening.[76]

12.4.3 Ductility of Nanocrystalline Metals Accompanying the enhanced strength of nanocrystalline metals is a disappointing lack of ductility. Many samples fail in tension after only about 2–3% strain. There frequently is a lack of strain hardening, which leads to strain localization and early failure. Part of the low tensile strainto-failure may be attributed to the presence of flaws and pores, especially

12: Mechanical Behavior, Weertman

557

Figure 12.12 Hall–Petch plot for electrodeposited Ni. (From Ref. [75].)

in IGC material. Artefact-free nanocrystalline copper produced by a combination of cryogenic and room temperature ball milling shows an appreciable ductility.[69] A copper sample with average grain size of 23 nm underwent a 14% uniform elongation and 15.5% elongation to failure. (Samples of IGC Cu and Pd are extremely ductile in compression, while failing quickly in tension, an observation lending support to the theory of flaw-induced brittle behavior.) However fully-dense electrodeposited nanocrystalline Ni was observed to fail at a few percent strain.[70] Evidently flaws are not the only cause of brittle behavior in nanocrystalline metals. Various schemes have been devised to increase the ductility of nanocrystalline material. Wang et al.[77] produced a bimodal grain-size distribution in nanocrystalline/UFG Cu by a combination of cryogenic rolling and careful annealing. The low temperature deformation permits the storage of a high density of dislocations, leading to strain hardening which helps to prevent early failure under tension. The small grains provide strength while the large grains stabilize the deformation of the material. The strength of the material is several hundred MPa below that described in Ref. [69] but it undergoes a 30% uniform elongation and 65% elongation to failure. Lu and colleagues[78,79] used an electrodeposition technique to produce nanocrystalline Cu that has internal strains an order of magnitude smaller than those usually measured in

558

Properties

nanocrystalline metals made by IGC, ball-milling or severe plastic deformation. This electrodeposited Cu was found to undergo extensions of about 30% in tension.[79] The high strain-to-failure probably is related to the internal structure, which consists of large grains or “domains” of a few micrometers in extent that are subdivided into “grains” ranging from a few nanometers in size to about 80 nm. These nanocrystalline grains are separated from one another by boundaries with misorientation angles of 1–10°. In another example,[80] a reasonably high tensile strain-to-failure of ∼8% was achieved in an electroposited Ni which was made up of nanometer-sized grains and clusters of grains in the 150–300 nm size. The grains in the clusters were separated by small angles of misorientation.

12.4.4 Experiments that Shed Light on Deformation Mechanisms It was noted in Section 12.2.2 that MD simulations indicate that in a stressed metal with a grain size <∼20–40 nm, dislocations are nucleated at grain boundaries, propagate across the grains and are absorbed in neighboring boundaries.[36] Upon unloading dislocations may return to the original boundaries. If this picture is correct, no residual dislocation network or debris is expected to remain inside the grains when the external stress is removed. In the case of coarse-grain metals subjected to plastic deformation, dislocation tangles, Lomer locks, etc. form in the grain interiors, and many of these remain upon unloading. To see if this picture of dislocation behavior in nanocrystalline metals is consistent with observation, Budrovic et al.[70] carried out an experiment in which X-ray measurements were made of Bragg peaks in a nanocrystalline electrodeposited nickel sample at the same time it was being loaded and unloaded. The results were compared with similar measurements on a coarse-grain metal. As predicted by the MD simulations the peaks broadened during loading but returned to the original widths when the stress was removed. These results are in accordance with the MD picture of an absence of dislocations in grain interiors after unloading. In contrast, only a small drop occurred in the full-width-half-maximum (FWHM) of the Bragg peaks in a plastically deformed coarse-grain sample upon unloading, indicating the permanent build-up of dislocation tangles inside the grains (Fig. 12.13). A number of researchers have observed that the ratio of the yield stress of a nanocrystalline metal at a cryogenic temperature to the room temperature value is much larger than the same ratio for coarse-grain

12: Mechanical Behavior, Weertman

559

metals.[56,67,81,82] This behavior is an indication of a thermal activation component to nanocrystalline deformation. A loading/unloading experiment similar to that of Ref. [70] but carried out at 180 K[83] showed that the Bragg peaks did not return to the original widths with stress removal.

(a)

(b)

Figure 12.13 (a) Stress–strain curves for nanocrystalline Ni showing loading and unloading sequences. (b) Behavior of FWHM of (400) and (222) Bragg peaks in nanocrystalline Ni and coarse-grain Cu during the loading and unloading sequences of (a). (From Ref. [70].)

560

Properties

Only when the material was warmed to room temperature was the recovery complete, a result in line with the existence of a thermal component to the deformation mechanism of nanocrystalline metals. Activation volume measurements of nanocrystalline Ni and Co by Wang and Ma[81] give values of 10–20 b3, as compared to typical coarse-grain values of ∼103 b3. (Here b is the Burgers vector.) Again, this small activation volume for the nanocrystalline metals is consistent with the picture of dislocations nucleating at grain boundaries and proceeding to a neighboring boundary. A number of workers have noted that the strain rate sensitivity m of nanocrystalline metals is higher than for coarse-grain materials, typically 0.02 vs. 0.001.[80–82] Another consequence of dislocation emission from grain boundaries appears to be the phenomenon of deformation twinning in nanocrystalline material. Deformation twinning in coarse-grain Al, with its high stacking fault energy, is never seen. However plastic deformation such as microindentation, grinding of thin films and ball milling leads to twinning in aluminum samples with grain size below 100 nm.[84,85] The twins are seen more frequently in the smaller grains. The emission of partial dislocations that travel across the grains, leaving behind stacking faults, is seen as a mechanism for twin formation. Deformation twinning was predicted by Yamakov et al.[86]

12.5 Conclusions The high hardness and strength values that have been obtained in nanocrystalline metals indicate the promising potential of these materials. Occasional examples of extensive deformation in tension give hope that elimination of flaws and other defects will lead to nanocrystalline metals with acceptable ductility for many applications. An important goal is to devise an internal structure that produces adequate strain hardening to prevent early plastic instabilities. Clearly synthesis of quality material in useful quantities and at a competitive cost is a top priority. From the scientific point of view, considerable progress has been made in understanding the deformation mechanisms in nanocrystalline metals. Molecular dynamics simulations have been especially helpful in explaining these mechanisms down to the atomic level. The predictions of the simulations have helped to guide the design of a number of exciting experiments. While grain boundary dislocation emission has been extensively treated by both simulations and experiment, deformation by grain boundary processes (sliding) remains somewhat shadowy.

12: Mechanical Behavior, Weertman

561

References 1. 2. 3. 4. 5. 6. 7.

8.

9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

27.

Hall, E.O., Proc. Roy. Soc. London B, 64:474 (1951). Petch, N.J., J. Iron Steel Inst., 174:25 (1953). Cottrell, A.H., Trans. TMS-AIME. 212:192 (1958). Nieh, T.G., and Wadsworth, J., Scripta Met. Mat., 25:955 (1991). Armstrong, R.W., Chou, Y.T., Fisher, R.M., and Louat, N., Phil. Mag., 14:943 (1966). Pande, C.S., Masumura, R.A., and Armstrong, R.W., Nanostruct. Mater., 2:323 (1966). Smith, T.R., Armstrong, R.W., Hazzledine, P.M., Masumura, R.A., and Pande, C.S., in: Grain Size and Mechanical Properties: Fundamentals and Applications (M.A. Otooni, R.W. Armstrong, N.J. Grant, and K. Ishizaki, eds.), 362:31, Materials Research Society, Pittsburgh, PA (1995). Armstrong, R.W., and Smith, T.R., in: Processing and Properties of Nanocrystalline Materials, (C. Suryanarayana, J. Singh, and F.H. Froes, eds.), p. 345, The Minerals, Metals and Materials Society, Warrendale, PA (1996). Li, J.C.M., Trans. TMS-AIME, 227:239 (1963). Li, J.C.M., and Chou, Y.T., Metall. Trans., 1:1145 (1970). Meyers, M.A., and Ashworth, E., Phil. Mag. A, 46:737 (1982). Meyers, M.A., Benson, D.J., and Fu, H-H., in: Advanced Materials for the 21st Century, (Y.-W. Chung, D.C. Dunand, P.K. Liaw, and G.B. Olson, eds.), p. 499, The Minerals, Metals, and Materials Society, Warrendale, PA (1999). Sangal, S., and Tangri, K., Scripta Met., 23:2079 (1989). Carsley, J.E., Ning, J., Milligan, W.W., Hackney, S.A., and Aifantis, E.C., Nanostruct. Mater., 5:441 (1995). Fu, H-H., Benson, D.J., and Meyers, M.A., Acta Mat., 52:4413 (2004). Kim, H.S., Estrin, Y., and Bush, M.B., Acta Mat., 48:493 (2000). Estrin, Y., Hellmig, R.J., and Kim, H.S., J. of Metastable and Nanocryst. Mats., 17:29 (2003). Herring, C., J. Appl. Phys., 21:437 (1950). Coble, R.L., J. Appl. Phys., 34:1679 (1963). Gleiter, H., Progress in Mats. Sci., 33:223 (1989). Gleiter, H., Acta Mat., 48:1 (2000). Thomas, G.J., Siegel, R.W., and Eastman, J.A., Scripta Met. Mat., 24:201 (1990). Kung, H., Sanders, P.G., and Weertman, J.R., in: Advanced Materials for the 21st Century (Y.-W. Chung, D.C. Dunand, P.K. Liaw, and G.B. Olson, eds.) p. 455, The Minerals, Metals, and Materials Society, Warrendale, PA (1999). Masamura, R.A., Hazzledine, P.M., and Pande, C.S., Acta Mat., 46:4527. (1998). Lian, J., Baudelet, B., and Nazarov, A.A., Mater. Sci. Eng. A, 172:23 (1993). Mitra, R., Ungár, T., Morita, T., Sanders, P.G., and Weertman, J.R., in: Advanced Materials for the 21st Century (Y.-W. Chung, D.C. Dunand, P.K. Liaw, and G.B. Olson, eds.), p. 553, The Minerals, Metals, and Materials Society, Warrendale, PA (1999). Morita, T., Mitra, R., and Weertman, J.R., Mats. Trans., JIM, 45:502 (2004).

562

Properties

28. Schiøtz, J., and Jacobsen, K.W., Science, 301:1357 (2003). 29. Mura, T., Micromechanics of Defects in Solids, pp. 79, 204–211, 216, Martinus Nijhoff Publishers, Dordrecht (1987). 30. Mori, T., and Tanaka, K., Acta Mat., 21:571 (1973). 31. Scattergood, R.D., and Koch, C.C., Scripta Met. Mat., 27:1195 (1992). 32. Yamakov, V., Wolf, D., Phillpot, S.R., and Gleiter. H., Acta Mat., 51:4135 (2003). 33. Van Swygenhoven, H., Science, 296:66 (2002). 34. Van Swygenhoven, H., Spaczer, M., and Caro, A., Acta Mat., 47:3117 (1999). 35. Derlet, P.M., Hasnaoui, A., and Van Swygenhoven, H., Scripta Mat., 49:629 (2003). 36. Van Swygenhoven, H., Derlet, P.M., and Hasnaoui, A., Phys. Rev. B, 66:024101 (2002). 37. Yamakov, V., Wolf, D., Phillpot, S.R., Mukherjee, A.K., and Gleiter, H., Nature Mater. 3:43 (2004). 38. Van Swygenhoven, H., Derlet, P.M., and Frøseth, A.G., Nature Mater. 3:399 (2004). 39. Frøseth, A.G., Derlet, P.M., and Van Swygenhoven, H., Appl. Phys. Ltrs., 85:1 (2004). 40. Schiøtz, J., Vegge, T., Di Tolla, F.D., and Jacobsen, K.W., Phys. Rev. B, 60:11971 (1999). 41. Van Swygenhoven, H., Farkas, D., and Caro, A., Phys. Rev. B, 62:821 (2000). 42. Van Swygenhoven, H., Derlet, P.M., and Frøseth. A.G., Acta Mat., 54:1975 (2006). 43. Nieman, G.W., Weertman, J.R., and Siegel, R.W., J. Mats. Res., 6:1012 (1991). 44. Sanders, P.G., Weertman, J.R., Barker, J.G., and Siegel, R.W., Scripta Met. Mat., 29:91 (1993). 45. Schaefer, H.-E., in Mechanical Properties and Deformation Behavior of Materials Having Ultra-Fine Microstructures, (M.A. Nastasi, D.M. Parkin, and H. Gleiter eds.), Kluwer Academic Press, Dordrecht, Netherlands, 1993, p. 81. 46. Würchum, R., Scheytt, M., and Schaefer, H.-E., Phys. Status Solidi. A, 102:119 (1987). 47. Van Petegem, S., Dalla Torre, F., Segers, D., and Van Swygenhoven, H., Scripta Mat., 48:17 (2003). 48. Sanders, P.G., Eastman, J.A., and Weertman, J.R., Acta Mat., 46:4195 (1998). 49. Sanders, P.G., Fougere, G.E., Thompson, L.J., Eastman, J.A., and Weertman, J.R., Nanostruct. Mater., 8:243 (1997). 50. Agnew, S.R., Elliott, B.R., Youngdahl, C.J., Hemker, K.J., and Weertman, J.R., Mater. Sci. Eng. A, 285:391 (2000). 51. Zhang, K., and Weertman, J.R., unpublished results. 52. Jiang, H.G., Rühle, M., and Lavernia, E.J., J. Mats. Res., 14:549 (1999). 53. Gertsman, V.Y., and Birringer, R., Scripta Met. Mat., 30:577 (1994). 54. Jin, M., Minor, A.M., Stach, E.A., and Morris, J.W., Jr., Acta Mat., 52:5381 (2004). 55. Zhang, K., Weertman, J.R., and Eastman, J.A., Appl. Phys. Ltrs., 87:061921 (2005).

12: Mechanical Behavior, Weertman

563

56. Zhang, K., Weertman, J.R., and Eastman, J.A., Appl. Phys. Ltrs., 85:5197 (2004). 57. Soer, W.A., De Hosson, J.Th.M., Minor, A.M., Morris, J.W., Jr., and Stach, E.A., Acta Mat., 52:5783 (2004). 58. Volpp, T., Göring, E., Kuschke, W.-M., and Arzt, E., Nanostruct. Mater., 8:855 (1997). 59. Weertman, J.R., and Sanders, P.G., Solid State Phenomena, 249:35–36 (1994). 60. Wang, Y.M., Cheng., S., Wei, Q.M., Ma, E., Nieh, T.G., and Hamza, A., Scripta Mat., 51:1023 (2004). 61. Hasnaoui, A., Van Swygenhoven, H., and Derlet, P.M., Acta Mat., 50:3927 (2002). 62. Kristic, V., Erb, U., and Palumbo, G., Scripta Met. Mat., 29:1501 (1993). 63. Sanders, P.G., Eastman, J.A., and Weertman, J.R., Acta Mat., 45:4019 (1997). 64. Cao, J.S., Hunsinger, J.J., and Elkedim, O., Scripta Mat., 46:55 (2002). 65. Shen, T.D., Koch, C.C., Tsui, T.Y., and Pharr, G.M., J. Mater. Res., 10:2892 (1995). 66. Bonetti, E., Campari, E.G., Del Bianco, L., Pasquini, L., and Sampaolesi, E., Nanostruct. Mater., 11:709 (1999). 67. Huang, Z., Gu., L.Y., and Weertman, J.R., Scripta Mat., 37:1071 (1997). 68. Sakai, S., Tanimoto, H., and Mizubayashi, H., Acta Mat., 47:211 (1999). 69. Youssef, K.M., Scattergood, R.O., Murty, K.L., Horton, J.A., and Koch, C.C., App. Phys. Ltrs., 87:09104 (2005). 70. Budrovic, A., Van Swygenhoven, H., Derlet, P.M., Van Petegem, S., and Schmitt, Bernd, Science, 304:273 (2004). 71. Sanders, P.G., Youngdahl, C.J., and Weertman, J.R., Mats. Sci. Eng., A234–236:77 (1997). Note that the correction from engineering to true stress/strain was done incorrectly for Figs. 8(a) and (b): after yielding the curves are almost horizontal. 72. Cheng, S., Ma, E., Wang, Y.M., Kesckes, L.J., Youssef, K.M., Koch, C.C., Trociewitz, U.P., and Han, K., Acta Mat., 53:1521 (2005). 73. Wang, Y.M., Wang, K., Pan, D., Lu, K., Hemker, K.J., and Ma, E., Scripta Mat., 48:1581 (2003). 74. El-Sherik, A.M., Erb, U., Palumbo, G., and Aust, K.T., Scripta Met. Mat., 27:1185 (1992). 75. Erb, U., Nanostruct. Mater., 6:533 (1995). 76. Wang, N., Wang, Z., Aust, K.T., and Erb, U., Mats. Sci. Eng., A237:150 (1997). 77. Wang, Y., Chen, M., Zhou, F., and Ma, E., Nature, 419:912 (2002). 78. Lu, L., Sui, M.L., and Lu, K., Science, 287:1463 (2000). 79. Lu, L., Wang., L.B., Ding., B.Z., and Lu, K., J. Mater. Res., 15:270 (2000). 80. Gu, C., Lian, J., Jiang, Z., and Jiang, Q., Scripta Mat., 54:579 (2006). 81. Wang, Y.M., and Ma, E., App. Phys. Ltrs., 85:2750 (2004). 82. Schwaiger, R., Moser, B., Dao, M., Chollacoop, N., and Suresh, S., Acta Mat., 51:5159 (2003). 83. Brandstetter, S., Budrovic, Z., Van Petegem, S., Schmitt, B., Stergar, E., Derlet, P.M., and Van Swygenhoven, H., App. Phys. Ltrs., 87:1 (2005).

564

Properties

84. Chen, M., Ma, E., Hemker, K.J., Sheng, H., Wang, Y., and Cheng., X., Science, 300:1275 (2003). 85. Liao, X.Z., Zhou, F., Lavernia, E.J., He, D.W., and Zhu, Y.T., App. Phys. Ltrs., 83:5062 (2003). 86. Yamakov, V., Wolf, D., Phillpot, S.R., Mukherjee, A.K., and Gleiter, H., Nat. Mat., 1:1 (2002).

13 Structure Formation and Mechanical Behavior of Two-phase Nanostructured Materials Jürgen Eckert IFW Dresden, Institut für Festkörperanalytik und Strukturforschung (IFS), Postfach 27 01 16, D-01171 Dresden, Germany

13.1 Introduction The development of nanostructured materials has led to a new class of materials that are single- or multiphase polycrystals with microstructural features, i.e. particle or grain sizes, layer thicknesses or domain sizes, in the nanometer range (typically less than 100 nm at least in one dimension). Owing to the extremely small dimensions, nanostructured materials have an appreciable fraction of their atoms in defect environments such as grain or interface boundaries. For example, nanocrystalline material with an average grain size of 5 nm has about 50% of its atoms within the first two nearest-neighbor planes of a grain boundary, in which distinct atomic displacements from the normal lattice sites are exhibited. These unique features have important impacts on their physical and chemical properties, which may be significantly different compared to conventional coarse-grained polycrystals of the same chemical composition.[1–3] For example, nanostructured materials may exhibit enhanced diffusivity, superior soft or hard magnetic properties, enhanced catalytic activity, ultrahigh strength or hardness or improved ductility and toughness in comparison with conventional polycrystals. This not only offers new scope for applications,[4] but also opens up fundamental questions regarding the basic understanding of the structure–property relations of materials as the microstructural scale is reduced to nanometer dimensions. While considerable progress has been achieved in the basic understanding of the structures of nanophase materials and their interplay with different physical and chemical properties, the shift from basic science to the design of nanostructured materials with optimum mechanical Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 565–676 © 2007 William Andrew, Inc.

565

Properties

566

properties has been rather limited. One of the reasons for this is the fact that it is relatively easy to produce powders, thin films or ribbons with nanoscale structure, but it is rather difficult to obtain nanostructured materials in bulk form, which is what is required for structural applications utilizing mechanical properties. It has been known for many years that the mechanical strength of crystalline metals or alloys is largely controlled by the grain size d. The wellknown empirical Hall–Petch equation[5,6] relates the yield strength σy to the average grain size d according to σy = σ0 + k d −1/2

Eq. (13-1)

where σ0 is the friction stress and k is a constant. A similar relation exists between the hardness and the grain size. Consequently, reducing d to the nanometer regime increases the strength considerably. However, the limits of the conventional description of yielding and of new mechanisms that may come into play at these small dimensions need to be explored and studied in much more detail. Besides, the intrinsically high interface-tovolume ratio of nanostructured materials may enhance interfacecontrolled processes to extend the strain to failure or plasticity. However, other structural features such as pores or larger flaws, grain boundary junctions and other crystal lattice defects that can depend on the preparation technique and the processing conditions also play an important role. It has become increasingly clear during recent years that all these structural aspects have to be carefully considered in order to fully understand the mechanical properties of nanostructured materials.[7–10] This chapter will refer to some of the material presented in previous reviews but will emphasize the more recent work on the mechanical properties of bulk nanostructured materials consisting of at least two different phases, in particular, nanostructured materials derived from bulk glass-forming alloys which contain amorphous and nanocrystalline phases. In view of the importance of processing methods on materials structure and properties, this overview starts with a description of the methods and the process variables typically employed for producing bulk nanostructured materials. Experimental data for the resulting nanostructures determining their mechanical behavior at room and elevated temperatures will be described for selected materials, and the toughness and the ductility increase due to the nanoscale structure will be critically assessed. Consideration of further developments as well as a discussion of the outstanding questions and challenges will also be addressed at the end of this chapter.

13: Structure Formation and Mechanical Behavior, Eckert 567

13.2 Methods of Preparation Detailed overviews on the different processing techniques employed for nanostructure formation can be found in several review articles or books, such as in [1, 3, 11]. Here, only processes are considered that are relevant for nanostructured two-phase materials, i.e. nanostructured metallic materials whose common characteristic is to consist of at least two different phases and to have a microstructural scale which is in the nanometer regime. A brief presentation of the most common synthesis routes and the typical microstructures obtained will first be given, as such basic information is necessary for an understanding of the microstructures and for the mechanical behavior discussed later on.

13.2.1 Rapid Solidification Techniques The advent of the ‘gun’ technique developed by Duwez and coworkers in 1960[12] to rapidly solidify metallic melts at a cooling rate of about 106 Ks−1 has most strongly influenced materials science by vastly expanding the processing window for a variety of existing materials and enabling the synthesis of entirely new compositions and alloys. Since then, rapid solidification processing has come to be recognized as resulting in characteristic constitutional and microstructural changes. The constitutional changes include extension of solubility limits, formation of new nonequilibrium crystalline or quasicrystalline phases and preparation of metallic glasses. The microstructural effects include changes in the morphology and the refinement of the dimensions of microstructural features such as size, shape and location of grains and phases. In general, the change is towards a more uniform and finer microstructure with a large reduction of solute segregation effects. Based on these effects, rapidly solidified materials have been extensively investigated over the last three decades. Several books, reviews and conference proceedings give an overview on the state-of-the-art situation on structure, properties and applications of rapidly solidified materials, e.g. [13–15]. High enough cooling rates during solidification can be accomplished when some important requirements are satisfied. The solidification rate, T˙ , during cooling is related to the section thickness, z, through the relation T˙ = 104 z−2

Eq. (13-2)

568

Properties

suggesting that the solidification rate increases by two orders of magnitude when the section thickness decreases by one order of magnitude. Hence, the molten metal must be delivered in a thin enough stream, at least in one dimension, and must have a large surface area-to-volume ratio to allow rapid heat removal. This can be achieved by maximizing the contact area between the melt and the cooling medium by rapidly increasing the surface area of the liquid alloy, which can be realized by spreading the melt as a thin layer on a substrate. Alternatively, an increase of the surface area can be effected by disintegrating the melt into small droplets, as it is the case for atomization techniques.

Chill Methods The most commonly used method for rapid solidification is the meltspinning technique. This process involves directing a molten metal jet onto a cold, moving heat sink where the jet is reshaped and solidified.[13] Thin ribbons or foils of up to 40–60 μm thickness are produced by this technique. The cooling rates can reach up to 106 Ks−1, thus making it possible to produce a variety of non-equilibrium phases and a large number of alloy systems, not only on a laboratory scale, but also for industrial production. The die method involves forcing a liquid metal into copper or steel chill-mold cavities of small cross-section through the use of vacuum[16], gravity[17] or pressure plus vacuum.[18] The advantage of this method is that specimens with predetermined cross-section (mostly circular) can be obtained. The cooling rates achieved strongly depend on the thickness of the specimen, ranging from about 106 to 104 Ks−1 for thicknesses of 0.2 or 0.7 mm, respectively.[19] Hence, the cooling rates achieved and the critical thickness of the sample are strongly coupled or, in other words, the usually high critical cooling rate necessary for metastable phase formation usually limits the thickness of the specimen to thin strips or foils significantly thinner than 1 mm. In principle, die casting into chill-mold cavities also enables direct preparation of three-dimensional bulk specimens and can be easily scaled up to industrial scale for appropriate materials and processing conditions. For example, copper mold casting, high pressure die casting or suction casting methods can be used for production of multicomponent bulk metallic glasses with dimensions in the centimeter range.[20,21] In this case, the excellent glass-forming ability of these alloys permits glass formation at cooling rates as low as 1–100 Ks−1, thus enabling the preparation of bulk specimens by die casting methods. Combined with devitrification such materials can be easily transformed into fully dense bulk nanostructured materials.

13: Structure Formation and Mechanical Behavior, Eckert 569

Spray Methods In these methods, a continuous stream of liquid metal is atomized, i.e. broken down into fine droplets by means of a gas (e.g. argon) or a liquid (e.g. water).[13,15] The resultant product after solidification is a powder with typical particle size in the micrometer range. The mechanism of achieving atomization and the means of cooling can be different in different techniques. In particular, atomization techniques are applied for large-scale production of rapidly solidified metal and alloy powders for commercial application. The typical solidification rates achieved are on the order of 106–102 Ks−1.

13.2.2 Mechanical Attrition Besides rapid quenching techniques, the formation of metastable phases and, in particular, also nanostructured materials can be achieved within the solid state. Starting in the 1970s, mechanical attrition (MA) of powders as a method for materials synthesis has been developed as an industrial process to successfully produce new alloys, phases and phase mixtures. Powder metallurgy allows the preparation of alloys and composites which cannot be synthesized via conventional casting or rapid solidification routes. Examples are uniform dispersions of ceramic particles in a metallic matrix or alloys of metals with rather different melting points with the aim of improved strength and corrosion resistance.[22,23] Over the years, mechanical attrition has gained a lot of interest as a nonequilibrium processing technique resulting in solid-state alloying beyond the equilibrium solubility limit and the formation of amorphous, quasicrystalline or nanostructured materials for a broad range of alloys, intermetallic compounds, ceramics and composites. The details of the process of MA, the microstructure development, alloy formation and the properties of the different classes of materials have been described in a variety of overviews, book chapters, special issues in journals etc.[24–31]

Powder Synthesis Mechanical attrition is usually carried out in high energy mills. For this purpose, a variety of different types of mills with different characteristics has been developed, including attrition mills, shaker mills, planetary ball mills, vibratory mills etc.[25] Powders with typical particle diameters of 50–150 μm are placed together with a number of balls in a sealed container

570

Properties

which is violently agitated at high frequencies or rotational velocities. The energy of the milling media (typically steel or tungsten carbide) depends on the internal mechanics of the specific mill, the power supplied for agitation, and the mass, the size and the number of the balls. Mostly the milling is carried out under dry conditions in an inert gas atmosphere (e.g. argon) but also milling under air, reactive gases with or without liquid or solid process control agents is possible, depending on the nature of the milled powders, in order to prevent sticking of the materials to the milling tools.

The Process of Mechanical Attrition The central event of the milling process is the ball–powder collision. Powder particles are trapped between the colliding balls during milling. They undergo repeated severe plastic deformation and fracture processes, leading to incorporation of lattice defects and to a continuous refinement of the initial structure of the powder particles to the nanometer regime. The nature of these processes strongly depends on the mechanical behavior of the powder components, their phase equilibria, and the stress state during milling, thus being different for pure metals or combinations of ductile/ductile, ductile/brittle or brittle/brittle components.[25] Hence, the behavior of the different classes of materials during MA depends on the intrinsic properties of the materials and, for alloys, is governed either by the stable or metastable phase equilibria. The temperature that the powders experience during milling also depends on the type of mill used and the actual milling parameters. However, experimental observations and modelling of the mechanics, kinetics and the energy transfer during collision suggest that the temperature rise during MA is about ≤100–200 K.[24,25] Hence, (metastable) alloy formation by MA is driven by intermixing, solid state interdiffusion and supersaturation beyond the equilibrium solubility limit and chemical reaction, whereby the individual processes are governed by the high defect density and the severe plastic deformation during MA, resulting in a highly non-equilibrium state of the powder particles.[24,25,32,33]

Influence of Milling Parameters: Milling Induced Phase Transitions The actual milling parameters, i.e. the kinetic energy transferred and the local temperature during a collision event, may control phase forma-

13: Structure Formation and Mechanical Behavior, Eckert 571 tion and the final structure of the material for a given composition.[24,34] Adjusting or changing the milling conditions can affect the phase formation or can lead to phase transitions between different phases. Depending on the chosen experimental conditions, an amorphous phase as well as a quasicrystalline or a crystalline phase can be synthesized directly from the composite of the starting elemental powder mixture. Which phase forms depends on the actual milling intensity used for the experiment (i.e., energy input, ambient milling temperature). In addition, the different phases can be transformed into each other by additional milling at higher or lower milling intensity. In particular, this has been demonstrated for the crystal-to-quasicrystal transition, the crystal-to-amorphous transition, and the quasicrystal-to-amorphous transition[24,35–37]

Nanostructure Formation upon Mechanical Attrition Besides the variability with respect to metastable phase formation, one other important feature inherent to MA is the development of nanoscale microstructures. As described above, during milling the powder particles are subjected to severe mechanical deformation from the collisions with the milling tools. Consequently, plastic deformation at high strain rates (∼103–104 s−1) occurs in the particles and a high level of internal strain is created due to the large number of dislocations and other deformation faults introduced. The dislocations rearrange into cell structures and deformation shear bands and finally evolve into high-angle grain boundaries by recovery and recrystallization processes. At this stage, no further grain refinement is possible but only grain boundary sliding can occur, which does not refine the microstructure any further.[34,38] Such results were first observed for pure metals and intermetallics[38–40] and have since then been reported for a variety of materials including supersaturated solid solutions[41,42] and multicomponent alloys.[43,44] Detailed overviews on nanostructure formation by MA for a variety of systems were recently given by Koch[25] or Fecht.[32]

Nanoscale Phase Mixtures Besides preparation of homogeneous nanocrystalline materials, mechanical attrition can also be applied to produce nanoscale phase mixtures of materials with distinctly different ductility. For example, 10-nm Ge particles can be embedded in a ductile Pb or Sn matrix.[45] Similarly, ultrafine dispersions have been produced for different systems such as

572

Properties

Fe—W, Cu—Ta, TiNi—C and Ag—Fe.[46,47] These results link the field of nanocrystalline materials to the well-known oxide-dispersion strengthened materials, where nanometric dispersions of oxides are embedded in a metallic matrix.[22,23] Hence, MA allows not only the production of different metastable phases and the utilization of mechanically induced phase transitions in order to either fully transform one phase into another, or to achieve phase mixtures of different metastable phases by properly adjusting the milling conditions, but also simultaneously offers the opportunity to achieve nanoscale microstructures by the mechanical deformation of the material.

Consolidation of Powders Consolidation of nanocrystalline powders into fully dense bulk specimens is of primary interest for the development of near-net shape parts for technological applications. The essence of all compaction techniques is to apply high pressure for densification, and rather high temperature to soften the material so that plastic deformation allows better filling and material flow by diffusion to remove the remaining porosity.[48] In addition, sliding or shearing of powder particles over each other will help to remove oxide and other surface layers to assist bonding. The fact that the nanocrystalline powders of interest are typically hard renders difficult the consolidation into essentially fully dense specimens with complete bonding between the initial particles. Moreover, diffusional processes will not only assist densification, but will also allow grain coarsening to occur. As such, the temperature and the time span of the consolidation process have to be adjusted carefully in order to achieve a balance between densification and grain growth. With respect to the pressure requirements, as a rule of thumb, a pressure of about three times the material flow stress, that is of the order of the hardness of the material (p ∼ H), will be required for continued densification by deformation.[48] However, using high pressure for consolidation bears the problem of cracking as the pressure is released.[49–53] At the other extreme of the temperature–pressure requirements are processes of pressureless sintering, hot isostatic pressing or uniaxial hot pressing, where rather high temperatures are required to compensate for low pressures.[50] However, high temperatures bear the risk of significant structural coarsening or even phase transitions, such as crystallization of metastable nano-quasicrystalline phases. Therefore, there is strong interest in using as low temperatures as possible in order to avoid structural changes.

13: Structure Formation and Mechanical Behavior, Eckert 573 The difficulties associated with the use of high pressures and the excessive structural coarsening associated with using high temperatures[50] render commercial processes such as Hot Isostatic Pressing (HIP) at reasonably high pressures (about 200–300 Mpa) and moderate temperatures (of the order of 0.5–0.65 the melting temperature) the most frequently used for consolidation of nanostructured powders. This gives highly dense (above 97%) specimens with final grain sizes after compaction of the order of 50–150 nm.[54–57] While methods using such pressure and temperature conditions are most common and areeasy to use, there are considerable advantages to using consolidation techniques that incorporate large amounts of shear deformation which assists cracking of surface layers and particle deformation to enhance interparticle bonding and densification. Particularly interesting are forging and extrusion methods, where high pressures and shear strains are imposed at moderate temperatures to achieve good densification and interparticle bonding.[58,59] In many cases, improvements in density and particle bonding not only require optimized consolidation techniques and parameters, but also great care has to be taken with respect to clean powder handling and compaction conditions. This often requires the use of closed-loop processing or ultrahigh vaccuum systems to ensure a high purity and good dynamic degassing.[58] In particular, restricting oxygen contamination and reducing the water and hydrogen content of powders seems to play a major role in achieving high densities, preventing residual porosity and promoting good bonding of particle surfaces. The most significant problems of powder consolidation stem from the risks of possible phase transformations and structural coarsening, i.e. grain growth or second phase or particle growth, during high temperature exposure. Grain growth can usually be described by an equation of the type dn − d0n = K · t

Eq. (13-3)

which relates the growth from an initial grain size d0 to a larger grain size d during the annealing time t. Simple theory suggests a value of n = 2, but larger values are frequently found. The kinetic term K is sensitive to the annealing temperature, but also to such factors as the chemistry (cleanliness) and structure of the grain and particle boundaries. For nanocrystalline materials the driving force for coarsening is high, because of the high surface area, and growth can occur even at temperatures as low as room temperature.[58,60,61] Grain growth can be hindered by a narrow size distribution,[61] and by impurities.[58]

574

Properties

Almost any deviation from pure single phase material will reduce the tendency for grain growth. This includes residual pores,[62] impurities and solutes[63] and second phase particles,[64] all slowing down grain growth. Pinning by fine particles is, in fact, a very efficient way of slowing or stopping grain coarsening, with the Zener drag process giving a relationship between the grain size d and the volume fraction f and the size of particles D, as approximately d ∝ D/f.[64] Hence, incorporating a significant volume fraction of fine second phase particles enables the retention of a fine grain size both during and after processing.[59,65–67]

13.2.3 Devitrification of Metallic Glasses An alternative to directly synthesizing metastable phases with nanocrystalline microstructure by rapid quenching or mechanical attrition techniques is controlled devitrification (crystallization) of amorphous solids to produce large quantities of nanocrystalline material. The basic principle for the devitrification method is to control the crystallization kinetics by optimizing the heat treatment conditions (annealing temperature and time, heating rate etc.) so that the amorphous phase partially or completely transforms into nanocrystalline material.

Nucleation and Crystal Growth Crystallization of metallic glasses is known to generally proceed by nucleation and growth processes. The phases and morphologies of crystallization products are dominated by the transformation mechanism, which is closely related to the chemical composition of the amorphous state and the thermodynamic properties of the corresponding phases. Both crystal nucleation and the growth rate depend strongly on temperature. Since the driving force for crystallization increases with decreasing temperature, but the diffusivity decreases with decreasing temperature, the highest growth rate is observed close to the melting temperature. The maximum nucleation rate is observed at significantly lower temperatures. Prerequisites for nanocrystallization, i.e. for the formation of material with a fine-grained microstructure upon crystallization, is a very high nucleation rate I combined with a low growth rate u. To obtain a very fine microstructure one has to choose an annealing temperature for which I/u is maximum,[68–71] i.e. crystallization has to proceed at temperatures just below the temperature for the highest nucleation rate. The nucleation rate is given by[68,69]

13: Structure Formation and Mechanical Behavior, Eckert 575 Q ⎞ ΔGc ⎞ I = I0 ⋅ exp⎛ − ⋅ exp⎛ − ⎝ R⋅T ⎠ ⎝ R⋅T ⎠

Eq. (13-4)

where Q is the activation energy for diffusion and ΔGc the nucleation barrier. For homogeneous nucleation the nucleation barrier is given by ΔGc =

γ3

( ΔG)

2

Eq. (13-5)

In general, the nucleation rate can be influenced by the interfacial energy γ, the driving force ΔG and the atomic mobility (diffusivity or viscosity) of the system. Most efficient for increasing the nucleation rate is a reduction of the interfacial energy γ.[70] It is known that γ is smaller for metastable phases or can be reduced by small additions of other elements. For example, in Co50Zr50 metallic glasses the shape of the stable crystalline CoZr phase changes significantly when the oxygen content of the glass increases to about 4 at%. This observation has been explained by a reduction of the interfacial energy by oxygen;[70] a significant increase of the driving force ΔG due to the small oxygen content is less likely. Whereas in Co50Zr50 metallic glass no metastable phases have been observed during crystallization, amorphous Co67Zr33 crystallizes first into a metastable cubic CoZr2 phase before transforming into another metastable hexagonal phase and finally into the stable tetragonal CoZr2 phase.[72] The lower driving force ΔG for the formation of metastable phases might be responsible for a reduced growth rate. The lower interfacial energy γ, which is typical for metastable phases, however, will reduce the nucleation barrier ΔGc significantly, thus increasing the nucleation rate. Due to the reduced nucleation barrier, crystallization of a metastable phase should always lead to a much finer microstructure than direct crystallization into the equilibrium phase.

Types of Transformation During conventional crystallization of amorphous solids, three different types of transformation are identified depending on the composition of the respective alloy:[68–71] polymorphous, eutectic and primary crystallization. All these types of transformations have been observed in nanocrystallization of amorphous alloys in different alloy systems. Figure 13.1 shows a hypothetical free energy diagram to illustrate the different reactions occuring upon crystallization of amorphous phases, displaying the

576

Properties change of free energy with composition of the glass and different nanocrystalline phases (e.g. two nanophases, a solid solution α and an intermetallic compound β) at a given temperature. Possible metastable equilibria between the different phases are marked by dashed tangents, the equilibrium coexistence tangent is represented by the solid line. Depending on composition, the transition of the metastable amorphous phase into the crystalline phases can proceed by one of the following reactions:

Figure 13.1 Hypothetical diagram of the free energy of the amorphous phase (am), a terminal solid solution (α) and an intermetallic phase (β), illustrating the nanocrystallization of amorphous solids. The numbers (1) to (4) correspond to the crystallization reactions mentioned in the text.[68]

Polymorphous Crystallization

An amorphous phase crystallizes into a single crystalline phase without any change in composition. The crystallization product can be a supersaturated solid solution or a metastable or stable crystalline compound. This type of reaction can occur only in composition ranges near compounds or pure elements (C1 or C2 in Fig. 13.1). As far as supersaturated phases (solid solutions) form during this reaction, they decompose by subsequent reactions. A metastable intermetallic compound will undergo a phase transition into the stable thermodynamic equilibrium phases.

Eutectic Crystallization An amorphous phase crystallizes simultaneously into two nanocrystalline phases (e.g. reaction 3: am → α + β) by a discontinuous reaction. By this, two nanophases grow in a coupled fashion analoguously to the eutectic crystallization of liquids. This type of reaction has the largest driving force and can occur in the concentration range between the two stable phases. The overall composition of the two phases remains the same as that of the inital amorphous phase. There is no concentration difference

13: Structure Formation and Mechanical Behavior, Eckert 577 accross the reaction front. However, in the reaction front the components have to separate into the different phases. Therefore, this type of reaction is initially slower compared to a polymorphous crystallization without any separation of the components.

Primary Crystallization An amorphous phase with a composition deviating from that for either polymorphic or eutectic reaction crystallizes in a first step into a primary nanocrystalline phase (either a supersaturated solid solution or an intermetallic compound) embedded in the amorphous matrix (reaction 4: am(C4) → α + am′(C3)). During this reaction the residual amorphous phase will change its composition until further crystallization is stopped by reaching the metastable equilibrium given by the dashed line. The residual amorphous phase (concentration C3) can transform in a second step (i.e. later or at higher temperatures) into nanophases in the mechanism of either eutectic or polymorphous nanocrystallization. The dispersed primary crystallized phase may act as preferred nucleation site for the subsequent crystallization of the residual amorphous matrix. In any case, the finest nanocrystalline structures should be obtained during primary crystallization. Primary crystallization is governed by long-range diffusion of the constituents and proceeds by a time-dependent growth rate. As the crystals grow their growth rate is further decreased until the diffusion fields of the individual crystals overlap. This long-range diffusion-controlled growth rate is significantly smaller than growth during polymorphous or eutectic reaction processes, which are controlled rather by interfacial diffusion processes. However, in a lot of cases a clear distinction between the different types of crystallization reactions is not unambiguously possible. This is due to the possibility of formation of intermediate metastable crystallization products. For example, oxygeninduced metastable phase formation makes it very difficult to clearly distinguish between primary or polymorphous crystallization for (CoZr2)1–xOx metallic glasses.[70]

Amorphous Phase Separation A large glass-forming ability is generally related to deep eutectics, indicating a strongly negative enthalpy of mixing of the constituents.[73] However, there is also evidence for the existence of systems which show

578

Properties

at least in some concentration range a different behavior, i.e. a positive enthalpy of mixing leading to a miscibility gap which may be due to singularities in the structure of the amorphous state. Therefore, a concentration range exists where the free energy of a mixture of two coexisting amorphous phases is lower than that of a single amorphous phase. Amorphous materials in this concentration range not only have the possibility of transforming by one of the crystallization reactions mentioned above, but may also decompose into two amorphous phases, i.e. may undergo amorphous phase separation. Amorphous phase separation occuring by nucleation and growth processes or even by spinodal decomposition without any nucleation process are well-known for oxide glasses.[74,75] There, extremely finegrained glass ceramics (partially crystallized glasses) can be produced, which are generally assumed to be a result of amorphous phase separation followed by subsequent crystallization. As illustrated in Figure 13.2, in the case of a miscibility gap there exists a concentration range between a and b where phase separation into two amorphous phases can occur by nucleation and growth or in a narrower concentration range between c and d by spinodal decomposition. The spinodal is the boundary between the metastable and the unstable state of the phase. In the metastable state a finite fluctuation, i.e. a nucleus of a new phase is required for phase transformation. In contrast, beyond the spinodal a single phase is unstable with respect to infinitesimal fluctuations in composition and begins to separate into two related phases with different compositions without nucleation. In

Figure 13.2 Spinodal decomposition:[70] (a) schematic phase diagram showing the phase boundary and the spinodal of a two-liquid immiscibility region, (b) free energy vs. composition diagrams for the temperatures given in (a).

13: Structure Formation and Mechanical Behavior, Eckert 579 a binary system, the condition for metastability is given by ∂2G/∂c2 < 0, where G is the molar free enthalpy and c is the mole fraction.[70] The driving force for increasing the composition fluctuations is the gain in free energy upon decomposition. Outside this concentration range decomposition into two amorphous phases will proceed by nucleation and growth processes. The kinetics of spinodal deomposition and the characteristic wavelength for a given annealing temperature have been determined by Huston et al.[76] Typical wavelengths at temperatures not too far below the critical temperature Tc are in the range of 5 to 10 nm. The decomposition rate is zero at the temperature of the spinodal and passes through a maximum with decreasing temperature. Amorphous phase separation is expected to have a strong potential for increasing the nucleation rate, and to decrease the growth rate of crystals. Hence, it promotes nanocrystal formation. For example, an increase in the nucleation rate may be due to an increase in the diffusivity of the respective element(s), a decrease in the crystal/glass interfacial energy, a local increase in the thermodynamic driving force or heterogeneous nucleation of crystals at the amorphous/amorphous interface. Nucleation can start only after amorphous phase separation, thus leading to significant incubation times for crystallization. On the other hand, phase separation is expected to reduce the crystal growth rates, due to composition shifts induced by the phase separation and/or by interference between the growing crystals and the amorphous regions.

Phase Separation in Bulk Glass-Forming Alloys Evidence for amorphous phase separation in metallic glasses was reported more than 20 years ago in Pd74Au8Si18,[77] (Pd0.5Ni0.5)81P19[78] or Be40Ti24Zr36[79,80] metallic glasses, either prior to crystallization or even in the as-quenched state. Very recently, strong evidence for amorphous phase separation was found for new bulk glass-forming alloys. In Zr41.2Ti13.8Cu12.5Ni10Be22.5 metallic glass Busch et al.[81] observed significant composition fluctuations in Be and Zr concentration in as-quenched bulk specimens by field ion microscopy. Schneider et al.[82,83] concluded from small-angle neutron scattering (SANS) data that a second miscibility gap opens in the undercooled liquid near the glass transition temperature, primarily involving Ti and Cu. They revealed the formation of Ti-rich regions which in later stages promote nanocrystallization. Depending on the annealing temperature, the wavelength of phase separation has been determined to be about 13–35 nm, following a relation λ−2∝T, as suggested by Cahn’s theory of spinodal decomposition. Furthermore,

580

Properties

decomposition triggers nanocrystal nucleation within the decomposed zones. The growth of these nanocrystals is limited by the size of the domains in which they nucleate. Similar observations were made for Mg—Y—Cu—Li bulk metallic glasses,[84] also showing evidence for phase-separated domains in highresolution TEM images. Additional small-angle X-ray scattering (SAXS) and anomalous small-angle X-ray scattering (ASAXS) studies of such alloys also gave evidence for phase separation into Cu-rich and Cu-poor regions.[85] Again, these phase-separated domains act as preferred nucleation sites for nanocrystals.[85,86] Hence, the precipitation of primary nanocrystalline phases in these bulk glass-forming systems is governed by the initial chemical composition of the undercooled liquid, and chemical decomposition controls their crystallization behavior. Choosing alloys with different composition, e.g. preparing an alloy having the same composition as the Ti- and Cu-poor decomposed regions in the Zr—Ti—Cu—Ni—Be system, allows one to obtain bulk metallic glasses without tendency to decomposition. For example, a Zr46.75Ti8.25Cu7.5Ni10Be27.5 glassy alloy shows no evidence of phase separation in SANS.[87] Hence, the occurence of phase separation and nanocrystallization strongly depends on the actual composition of the alloy, giving the unique opportunity to specially design bulk nanostructured materials by appropriate choice of alloy composition and optimization of annealing conditions.

13.3 Phenomenology of Nanostructure Formation and Typical Microstructures The nanostructured materials being discussed here may contain crystalline, quasicrystalline or amorphous phases of the metallic constituents. In addition, nanostructured composites based on metallic glasses will be considered. The materials can be produced by fabrication routes based on rapid quenching or slow cooling from the melt, or powder metallurgy, respectively. These synthesis routes may directly lead to nanostructured materials or, in other cases, have to be combined with an appropriate heat treatment to obtain the desired nanoscale microstructure, such as in the case of (partial) devitrification of metallic glasses. Whereas crystallization of bulk metallic glasses can directly yield bulk nanostructured materials, subsequent consolidation of rapidly quenched powders and ribbons, or compaction of mechanically attrited powders, has to be performed in order to obtain bulk samples.

13: Structure Formation and Mechanical Behavior, Eckert 581

13.3.1 Rapidly Solidified Materials Substantial increases in strength along with good ductility have been observed in a number of alloys with multiphase nanoscale microstrutures. Such ‘nanophase composites’ can be made directly by liquid quenching or by annealing of initially fully amorphous material.[88,89] In addition, devitrification can also be induced by mechanical working.[90] Most of the work on multiphase nanoscale microstructures has concentrated on Al-, Mg- and Zr-based multicomponent systems with the goal of developing new materials with high specific strength (and thus low density) and high ductility. By the control of alloy composition and quenching conditions several different types of alloys containing non-equilibrium phases and microstructures have been fabricated. In particular, great attention has been paid to non-peridoc structure alloys consisting of amorphous and quasiperiodic phases.

Al-Based Nanostructured Alloys The first success in synthesizing Al-based high specific strength material with good ductility was achieved for amorphous Al—Ni—Si alloys by Inoue et al. in 1987.[91] Since then, a large number of Al-based amorphous alloys has been found.[92,93] These alloys can be classified into Almetal and Al-metalloid systems.[94] Among these alloy systems the Al—Ln—LTM and Al—ETM—LTM (Ln = lanthanide metal, ETM = early transition metal (groups IV–VI), LTM = late transition metal (groups VII and VIII)) amorphous alloys are particularly important because of the simultaneous achievement of high strength and good bending ductility. Subsequently, it was found[90] that a homogeneous dispersion of nanoscale Al particles in an amorphous matrix causes a strong increase in mechanical strength. This fine-mixed structure consisting of nanoscale fcc Al particles 3 5 nm in size embedded in an amorphous phase is typically found in Al—Ln—LTM alloys, and can be produced either by annealing the amorphous single phase[95] or by decreasing the cooling rate upon quenching from the liquid.[88] More recently, similar microstructures have also been observed in alloys not containing any lanthanides.[96] In the historical progress of the development of high-strength Al-based alloys, a drastic increase in tensile fracture strength was found for the more recent nonequilibrium base alloys including amorphous or quasicrystalline phases. Figure 13.3 summarizes the features of the microstructure and the mechanical strengths of the different types of non-equilibrium alloys

582

Properties

Figure 13.3 Microstructures and mechanical properties of non-equilibrium Al-based alloys.[99]

developed so far.[94,97–99] The non-equilibrium structures for Al-based alloys can be classified into the follwing six types: (1) an amorphous single phase, (2) a nanostructure of Al and intermetallic compounds obtained by devitrification of an initial amorphous phase, (3) a partially crystallized structure of nanoscale Al particles embedded in an amorphous matrix, (4) a nano-quasicrystalline structure consisting of nanoscale icosahedral particles which are surrounded by a nanoscale fcc Al phase without grain boundaries, (5) a phase mixture of coexisting nanogranular amorphous and Al phases, and (6) a nanogranular Al phase surrounded by an amorphous network phase.

13: Structure Formation and Mechanical Behavior, Eckert 583

Amorphous Single-Phase Alloys The amorphous alloys obtained by melt spinning can be divided into metal–metal and metal–metalloid systems. Among these systems, both Al—Ln—TM[93,100] and Al—ETM—LTM[101] systems are more important because of the achivement of higher fracture strength. In addition to meltspun thin ribbons, Al-based amorphous wires with diameters in the range 40–120 μm, which exhibit high strength and good ductility, have been produced by a melt-extraction method.[102] No bulk amorphous single-phase alloys in Al-based systems have been obtained by any kind of solidification method up to date. This is due to the relatively low glass forming ability of these systems. However, amorphous alloys with Al concentrations above about 85 at% exhibit a glass transition phenomenon, followed by a rather narrow supercooled liquid region of about 20 K prior to crystallization.[103] This promises that by utilizing the viscous flow of the supercooled liquid, bulk amorphous samples can be produced by extrusion of pieces of melt-spun ribbons or from gas atomized powders with particle sizes below about 25 μm.[104] For example, first results for the extrusion of Al85Ni10Ce5 amorphous powders at temperatures between 443 and 493 K yielded mostly single-phase bulk amorphous samples with diameters of up to 20 mm.[105]

Partially Crystallized Alloys The phase formation from the highly undercooled liquid (or the amorphous phase) has the following features:[106] (1) homogeneous nucleation with a high nucleation rate and a low growth rate, (2) a large concentration gradient of solute elements resulting from the low atomic diffusivity of the solute atoms and a highly dense-packed structure at the liquid/solid interface, (3) formation of metastable phases with new compositions, (4) formation of a residual amorphous phase with high solute concentration, and (5) precipitation of almost ideally spherical defect-free nanosized particles with nanoscale interparticle spacings. These features enable the synthesis of new nanoscale mixed phase alloys upon rapid solidification or during subsequent annealing. In general, Al-rich amorphous alloys with high Al contents above about 85 at% crystallize through two stages upon heating to elevated temperatures. The first exothermic reaction is due to the precipitation of fcc Al and the second stage results from the decomposition of the residual amorphous phase into intermetallic compounds.[107] Appropriate control of the

584

Properties

cooling rate during quenching from the melt gives a nanostructure consisting of 3–5 nm Al particles embedded in an amorphous matrix in meltspun ribbons.[107] The volume fraction of precipitates can be adjusted by changing the rotation speed of the wheel. The hardness, the yield strength and the Young’s modulus of the material increases for volume fractions of up to about 25%. For larger volume fractions the yield strength decreases again, which is attributed to the embrittlement of the residual amorphous phase due to structural relaxation and enrichment of solute elements above a critical level. Similar microstructures form upon devitrification of initially amorphous ribbons, i.e. for annealing in the first-stage exothermic reaction range.

Bulk Nanocrystalline Alloys Raising the annealing or extrusion temperature to temperatures corresponding to the second-stage exothermic reaction allows the production of fully dense nanoscale multiphase alloys consisting of fine intermetallic compounds of about 50 nm in size, which are embedded in an fcc Al matrix with typical grain sizes of about 100–200 nm.[106] This yields a unique microstructure in which intermetallic phases are dispersed homogeneously within the Al matrix and on the grain boundaries. In particular, the density of the dispersed compound particles is not high enough to suppress all dislocation activity and a significant fraction of dislocations can interact with the grain boundaries of the Al phase. Such nanocrystalline Al-based allloys have been commercialized under the name of “GIGAS”[99] and have already reached some technological importance.

Nanoscale Quasicrystalline Alloys A finely mixed structure consisting of nanoscale icosahedral (I) quasicrystalline particles with sizes in the range 20–50 nm surrounded by an fcc Al phase with a thickness of about 5–10 nm was found in melt-spun Al—(Mn, Cr)—Ce,[108–110] Al—(Mn, Cr)—Ce—TM,[111,112] Al—V—Ce— TM[113] and Al—(V, Cr, Mn)—TM[114] systems (TM = Fe, Co, Ni or Cu). These alloys were chosen because Al—Mn and Al—Cr alloys are wellknown I-phase forming systems and Al—Ln-based alloys possess a good glass-forming ability (see above). Figure 13.4 illustrates the compositional dependence of structure formation for melt-spun Al—Mn—Ce alloys as a typical example for this type of material. Depending on the actual alloy

13: Structure Formation and Mechanical Behavior, Eckert 585

Figure 13.4 Compositional dependence of structure, tensile strength, σf, and ductility of melt-spun Al—Mn—Ce alloys.[99]

a

b

Figure 13.5 X-ray diffraction pattern of a melt-spun Al92Mn6Ce2 alloy. The selected-area electron diffraction patterns confirm the formation of an icosahedral phase. The DSC trace (inset) reveals a transition of the metastable icosahedral phase starting at about 720 K.[116]

composition, different phase mixtures of amorphous, quasicrystalline and fcc Al phases form directly upon quenching.[115] X-ray diffraction (XRD) analysis and transmission electron microscopy (TEM) investigations proved the formation of a mixed structure consisting of nanoscale icosahedral and fcc Al phases in the Al-rich composition range above 92 at% Al, as displayed in Figure 13.5 for an Al92Mn6Ce2 alloy.[116,117] The inset in Figure 13.5 shows a typical DSC trace for the metastable phase mixture, revealing that the I-phase starts to decompose into Al6Mn at about 720 K. The nanoscale I-phase particles with diameters of about 50–100 nm have

586

Properties

Figure 13.6 Bright-field TEM micrograph of a melt-spun Al92Mn6Ce2 alloy and selected-area electron diffraction pattern taken from one of the particles.[117]

an almost spherical morphology and random orientation, and are homogeneously dispersed in the Al phase without high-angle grain boundaries (Figure 13.6). The volume fraction of the I-phase is as high as 60–70%. A similar microstructure forms for other rapidly solidified Al—(Mn, Cr, V)—Ce—TM alloys and also in Al—Fe—Cr—Ti alloys without any lanthanide metal.[118] In addition to melt-spun-ribbons, a mixed microstructure of nanoscale I-phase surrounded by an Al phase is also obtained for atomized Al— Mn—Ln and Al—Cr—Ln—TM powders, as well as in Al—Mn—TM and Al—Cr—TM systems.[98] The microstructural scale of the different phases is slightly larger, i.e. 50–200 nm, than in the case of melt-spun ribbons. In addition, some intermetallic phases, such as Θ′-Al13Cr2 and similar compounds, may also be present. Both the slightly coarser microstructure as well as the appearance of additional compounds is due to the lower quenching rate achievable by atomization compared to melt spinning. Extruding the atomized powders at temperatures below 673 K, i.e. below the decomposition temperature of the icosahedral phase, renders fully dense bulk specimens which maintain the mixed microstructure of the initial powder and show no extensive growth of the quasicrystalline domains or the fcc Al phase.

Nanogranular Two-Phase Alloys In addition to the Al—Mn/Cr- and Al—Fe-based alloys, amorphous and quasicrystalline phases also form in melt-spun Al—V- and Al—Ti-based

13: Structure Formation and Mechanical Behavior, Eckert 587 systems containing solute elements from groups IV and V of the periodic table. For example, melt-spun Al94V4Fe2 consists of nanogranular amorphous grains with a size of about 10 nm in diameter coexisting with fcc Al grains with a size of about 7 nm.[119] The nanoscale amorphous phase crystallizes through two stages upon heating to elevated temperatures, i.e. first transforms into an icosahedral phase, which in a second step crystallizes into an Al11V compound. A similar nanogranular amorphous microstructure is observed for Al95Ti3Fe2.[120] In contrast, melt-spun Al93Ti5Fe2 with a slightly different composition consists of fcc Al grains with a size of about 30–40 nm, which are surrounded by an amorphous phase network with a thickness of about 7–10 nm.[120] This indicates that the Al phase precipitates as primary phase and the amorphous network phase subsequently forms from the remaining liquid. This change in solidification mode causes the disappearance of the nanogranular amorphous phase observed for larger Al contents, revealing that small changes in composition may have a rather drastic effect on the solidification behavior and the resulting microstructure of the alloys.

Mg-Based Alloys From the point of view of developing high strength materials with light weight, Mg-based alloys have attracted a lot of attention because of the low density of Mg. Since amorphization of metallic alloys by rapid quenching causes a significant increase in mechanical strength compared with the corresponding crystalline alloys, the formation of amorphous Mg-based alloys as new structural materials with higher specific strength has been investigated extensively in recent years. This has led to the development of new Mg-based alloys with high tensile strength in a number of alloy systems, such as Mg—Ln—TM (TM = Ni, Cu, or Zn),[121,122] Mg— Y—Ln,[123] Mg—Y—Al,[123] Mg—Ca—Al,[124] Mg—Zn—Al[125] and Mg—Al—Ga[126] ternary systems. These alloy systems can be devided into two types: metal–metal systems and metal–metalloid systems. Table 13.1 summarizes the various alloy systems in which an amorphous phase containing more than 50 at% Mg can be obtained by single-roller melt spinning. In particular the alloys of the Mg—Ln—TM group are of importance because of their high tensile fracture strength (see 13.4.2). In addition, the Mg—Ca—Al, Mg—Y—Al and Mg—Y—Ln systems are also attractive for achieving high corrosion resistance.[125] In general, glass formation in these ternary Mg-based alloy systems is observed over a rather broad range of compositions. This is exemplified in Figure 13.7 for the Mg—Ni—Y and Mg—Cu—Y systems.[127] An amorphous phase forms in

Properties

588

Table 13.1 Mg-Based Amorphous Alloys Produced by Melt Spinning[125]

Metal–Metal Systems Mg—Ca—Al, Mg—Ca—Li, Mg—Ca—M, Mg—Sr—M, Mg—Al—Ln, Mg—Al—Zn, Mg—Ni—Ln, Mg—Cu—Ln, Mg—Zn—Ln

Metal–Metalloid Systems Mg—Ca—Si, Mg—Ca—Ge, Mg—Ni—Si, Mg—Ni—Ge, Mg—Cu—Si, Mg—Cu—Ge, Mg—Zn—Si, Mg—Zn—Ge,

Ln = Lanthanide Metal; M = Transition Metal (Ni, Cu, Zn).

almost the whole composition range when the solute concentration is above 12 at% in both alloy systems. Similar data are observed for other ternary Mg-based alloys.[127]

Glass Transition and Crystallization in Mg-Based Alloy Systems The crystallization temperatures of the Mg-based amorphous alloys are in the range of 440–600 K. There is a clear tendency for the crystallization temperature, Tx, to increase with increasing solute content,[125] and Tx is rather insensitive to the actual solute element introduced into the amorphous phase. However, only the Mg—Ln—TM amorphous alloys exhibit a distinct glass transition and a rather wide supercooled liquid region before crystallization as well as a large glass-forming ability.[128] Figure 13.8 shows differential scanning calorimetry (DSC) traces for Mg—Ni— Y and Mg—Cu—Y amorphous alloys as typical examples for the thermal stability behavior of such alloys.[128] The temperature interval ΔTx between the glass transition temperature, Tg, and the onset of crystallization at Tx (ΔTx = Tx − Tg), i.e. the extension of the supercooled liquid region, is as large as 69 K for Mg65Cu25Y10 (at a heating rate of 40 K/min). The glass transition phenomenon is observed for a variety of compositions, e.g. for 3–5 at% Y and 5–50 at% Cu.[127] Similar data were found for the Mg— Ni—Y system.[127] The supercooled liquid crystallizes through one exothermic crystallization event, leading to the simultaneous formation of nanoscale Mg2Cu and Mg24Y5 precipitates.[84,123,129,130] Mg-rich alloys with more than 80 at% Mg crystallize through several stages upon heating to elevated temperatures.[123,131,132] The first

13: Structure Formation and Mechanical Behavior, Eckert 589

Figure 13.7 Composition ranges in which an amorphous phase is formed by melt spinning for (a) Mg—Ni—Y and (b) Mg—Cu—Y systems.[127]

590

Properties

Figure 13.8 Typical DSC traces of Mg—Ni—Y and Mg—Cu—Y amorphous alloys.[125]

exothermic reaction is due to precipitation of hcp Mg nanocrystals, coexisting with a residual amorphous matrix/supercooled liquid with changed composition. Subsequent crystallization proceeds by decomposition of the residual amorphous phase into intermetallic compounds. The volume fraction of the precipitates can be controlled by adjusting the annealing temperature and time properly. Hence, similarly to the Al-based alloys, fine-scale mixed structures consisting of homogeneously dispersed nanoscale hcp Mg precipitates in an amorphous matrix can be obtained by partial devitrification of the amorphous phase.

Bulk Nanostructured Samples by Extrusion Melt-spinning or gas-atomization combined with extrusion at elevated temperatures in the supercooled liquid state or near the onset of crystallization can also be applied to Mg-based powders, similarly to the way already shown for Al-based amorphous alloys (see above). Since Mgbased powders are extremely reactive, this requires very stringent control of the oxygen and moisture contents in the atmosphere during atomization, and also during consolidation of the resulting fine powders. Hence, the production of bulk Mg-based alloys from atomized powders is typically carried out in a closed-loop processing system, where melting of the master alloy, atomization, powder collection, sieving as well as precompaction and sealing into Al or Cu cans for subsequent extrusion are con-

13: Structure Formation and Mechanical Behavior, Eckert 591 ducted under high-purity controlled argon with less than 1 ppm oxygen in a single chamber.[133] Using this type of equipment, Mg-rich atomized powders (e.g., Mg85Y10Cu5) were successfully produced for different alloy systems, such as Mg—Cu—Y, Mg—Ni—Y, Mg—Ca—Al, Mg—Al—Zn etc.[125] The atomized powders exhibit a spherical morphology and the average particle size is about 20 μm. After atomization, the powders consist mainly of an amorphous phase. Besides, they contain a small amount of a fine hcp Mg-phase,[133] which can not be suppressed due to the lower cooling rate upon atomization compared to melt-spinning. The typical grain size of the hcp Mg particles is on the order of 10 nm and their interparticle spacing is also about the same scale.[132] The thermal stability data and the features of the DSC scan for the atomized powders are almost the same as for single-phase amorphous melt-spun ribbons. For Mg—Cu—Y alloys, extruding at temperatures between about 460 and 630 K[134] yields dense bulk compacts without significant structural transformation,[133] i.e. the bulk alloy also consists of an amorphous matrix phase with embedded hcp Mg particles. Using higher extrusion temperatures leads to the formation of nanostructured phase mixtures of hcp Mg and intermetallic compounds, e.g. Mg24Y5 and Mg2Cu phases, as found for extruded bulk Mg85Y10Cu5.[133] TEM investigations revealed a microstructure consisting of Mg grains with diameters of about 150–200 nm, in which Mg24Y5 particles are homogeneously distributed. The Mg2Cu precipitates mainly reside at the grain boundaries. A similar fine mixed structure was obtained in a bulk Mg70Ca10Al20 alloy produced by extrusion at 673 K.[124] The extruded alloy exhibits an hcp matrix with a grain size of about 100 nm, and the particle size and the interparticle spacing of the Al2Ca compound are about 80 nm and 50 nm, respectively. Even larger grain and particle sizes were found for a Mg–8.3 wt% Al–8.1 wt% Ga alloy, where the average grain size of the hcp Mg matrix is about 600 nm and that of the intermetallic Mg17Al12 and Mg5Ga2 compounds ranges from 100 to 300 nm.[126,135] The Mg grains are almost equiaxed and the intermetallic compounds are evenly distributed throughout the grains and at grain boundaries, yielding a rather homogeneous microstructure after extrusion without any significant texture.

Zr-Based Alloys Multicomponent Zr-based alloys (e.g. Zr—Al—Cu—Ni or Zr—Ti— Al—Cu—Ni)[136–139] exhibit exceptional glass-forming ability and belong

592

Properties

to the best glass-forming systems known so far. They form a glass at cooling rates of under 100 K/s with critical casting thicknesses ranging upward toward several centimeters thickness.[20] This is illustrated in Figure 13.9, showing the critical cooling rate for glass formation, Rc, and the maximum sample thickness for glass formation, tmax, as a function of the reduced glass transition temperature, Trg = Tg/Tm (Tm is the melting temperature of the alloy). However, the glass-forming ability and the maximum attainable sample thickness strongly depend on a suitable alloy composition and the requirement of Figure 13.9 Relation between the high purity starting materials and procritical cooling rate for glass formation, Rc, the maximum cessing conditions. In particular, the sample thickness, tmax, and the overall oxygen content of the alloy is reduced glass transition a key parameter for bulk glass formatemperature, Tg/Tm, for amorphous tion. For example Lin et al.[138] studied [127] alloys. the effect of oxygen on the crystal nucleation and growth for a Zr—Ti— Cu—Ni—Al alloy upon cooling from the melt. Over the range of oxygen contents studied (300 to 5000 atomic ppm), the time–temperature–transformation curves vary roughly by two orders of magnitude along the time axis. In other words, oxygen contamination ranging up to 0.5 at% can increase the necessary critical cooling rate for glass formation by two orders of magnitude. Similar behavior has been found for other Zr-based alloys, such as quaternary Zr—Al—Cu—Ni alloys.[140] This effect is related to oxygen-induced formation of metastable phases which, in turn, can act as heterogeneous nucleation sites for further crystallization upon solidification from the melt. This, of course, strongly affects bulk glass formation and limits the production of bulk specimens to high-purity material and optimum compositions with a large resistance against oxygen-induced crystal nucleation. To circumvent the limitations of casting processes at low cooling rates, there have also been efforts to produce bulk samples through consolidation of rapidly quenched ribbons or atomized powders,[141–143] since the nucleation of the competing metastable crystalline phases can be sup-

13: Structure Formation and Mechanical Behavior, Eckert 593 pressed by fast enough cooling from the melt. Choosing appropriate consolidation parameters yields dense bulk samples without crystallization and almost identical thermal stability as that of melt-spun ribbons. In particular, the material retains an extended supercooled liquid region before crystallization, which can be used for further viscous flow deformation for shaping of three-dimensional parts.

Oxygen-Induced Precipitation of Nanoscale Compound Particles upon Crystallization Oxygen impurities not only affect the solidification of multicomponent Zr-based alloys upon quenching from the melt, but also strongly influence the crystallization behavior during heating to elevated temperatures or upon isothermal annealing. This has been studied in detail for (Zr0.65Al0.075Cu0.175Ni0.10)100–xOx rapidly quenched amorphous ribbons (x = 0.2, 0.4 and 0.8).[144] X-ray diffraction and TEM proved that all the samples are fully amorphous after melt-spinning. The DSC scans of the ribbons show an endothermic heat event characteristic of the glass transition, followed by exothermic heat release events due to crystallization (Figure 13.10). For example, amorphous (Zr0.65Al0.075Cu0.175Ni0.10)99.8O0.2 exhibits a wide supercooled liquid region of more than 100 K and crystallizes via one sharp exothermic peak. Most prominent is the change of the DSC signal from one sharp exothermic crystallization peak for x = 0.2 to two or three more or less well resolved crystallization peaks for samples with higher oxygen content. This indicates a successive stepwise transformation from the supercooled liquid to the equilibrium crystalline intermetallic compounds at different temperatures. Tg slightly increases and the onset of crystallization, Tx, decreases with increasing oxygen content, thus leading to a reduction of the extension of the supercooled liquid region, ΔTx, for oxygen-rich samples. The phases formed upon devitrification of the (Zr0.65Al0.075Cu0.175Ni0.10)100–xOx amorphous alloys were studied by time- and temperature-resolved XRD and TEM measurements.[144] Figure 13.11 shows schematic phase formation diagrams for the (Zr0.65Al0.075Cu0.175Ni0.10)100–xOx ribbons with different oxygen content. The details of phase formation strongly depend on the annealing conditions. For x = 0.2, crystallization is governed by simultaneous precipitation of quasicrystalline, tetragonal CuZr2, and hexagonal Zr6NiAl2 phases which are embedded in the residual amorphous matrix (Fig. 13.11(a)). A metastable fcc NiZr2-type phase forms as an intermediate crystallization product. At elevated temperatures the quasicrystals transform into CuZr2,

594

Properties

Figure 13.10 DSC traces of amorphous (Zr0.65Al0.075Cu0.175Ni0.10)100−xOx melt-spun ribbons (x = 0.2, 0.4 and 0.8).[144]

and the fcc phase transforms into the stable Zr6NiAl2 compound. The grain sizes of the different phases vary between about 10 and 70 nm, depending on the actual annealing conditions (further details are given in ref. [144]). Sequential transformations from the supercooled liquid to the intermetallic compounds are also observed for (Zr0.65Al0.075Cu0.175Ni0.10)99.6O0.4 and (Zr0.65Al0.075Cu0.175Ni0.10)99.2O0.8 ((Figs. 13.11(b) and (c)). With increasing oxygen content the existence regions of the metastable fcc and quasicrystalline phases become more extended. This reveals that the presence of oxygen promotes the formation of these intermediate phases and stabilizes the metastable structures against transformation into the equilibrium compounds. Hence, the phase formation and stability regions of the oxygen-stabilized metastable phases depend systematically on the oxygen content of the material and the actual annealing conditions. Therefore, the alloy has in fact to be viewed as a five-component system in which the dilute component oxygen plays a decisive role for the characteristics of crystallization. Similar results were obtained by different groups for multicomponent Zr-based alloys of different composition[138,145–147] On one hand, the effect

13: Structure Formation and Mechanical Behavior, Eckert 595

Figure 13.11 Schematic phase formation diagram for (Zr0.65Al0.075Cu0.175Ni0.10)100−x Ox ribbons with different oxygen content: (a) x = 0.2, (b) x = 0.4 and (c) x = 0.8. The dashed lines are only guidelines for the eye illustrating the temperature— time regimes where the different phases were detected.[144]

of oxygen lowers the thermal stability of the material against crystallization and is, therefore, detrimental. On the other hand, this leads to phases and microstructures which cannot be achieved in the pure metallic glass, and allows the synthesis of amorphous alloys with nanoscale compound particles in an amorphous matrix by proper annealing. Therefore, oxygen or other trace elements are useful for obtaining new nanostructured composites from amorphous Zr-based alloys.

596

Properties

Nanoscale Compound Formation in Zr-Based Alloys with Different Composition The effect of successive stepwise transformation from the supercooled liquid to the crystalline equilibrium phases is not limited to the influence of trace elements such as oxygen. One example is the multicomponent alloy system Zr60Al10Cu30–xPdx.[148,149] The crystallization of the alloy with x = 10 occurs also through two stages: amorphous → Zr2(Cu, Pd) + amorphous’ → Zr2(Cu, Pd) + Zr2Al. For this alloy, annealing for 1 h at 726 K between Tg and Tx induces precipitation of a large volume fraction (about 70%) of Zr2(Cu, Pd) nanocrystals with a diameter of about 5–10 nm, which are embedded in a remaining amorphous matrix with changed composition, occupying a volume fraction of about 30%. Similar results were found for a variety of Zr-based alloys with different alloying elements, e.g. Zr—Al—Cu—Ni—Fe,[150,151] Zr—Al—Cu—Pd—Fe,[149] Zr—Al— Ti—Cu—Ni,[139,152,153] Zr—Al—Cu—Ni—Pd,[154] Zr—Al—Ni—Cu—M (M = Ag, Au, Pt).[155] In all cases, a sequential crystallization process was observed, involving precipitation of a nanoscale intermetallic or quasicrystalline phase in the first step of transformation, which coexists with an amorphous phase with changed composition. At higher temperatures, the residual amorphous phase also crystallizes and the material becomes fully nanocrystalline. The details of precipitation and transformation strongly depend on the actual composition and the annealing conditions.[151] As an example, Figure 13.12 shows the average crystallite size, D, of the primary nanocrystals with metastable fcc NiTi2 structure precipitated during the first crystallization step of (Zr0.65Al0.075Cu0.175Ni0.10)100–xFex alloys. The average crystallite size fits well to 3 x functional behavior, indicating that the number of fcc crystallites, N, is proportional to the iron content in the alloy (N ∼ x). Figure 13.13 shows the behavior of the average crystallite size versus annealing time for a (Zr0.65Al0.075Cu0.175Ni0.10)80Fe20 alloy for the annealing temperature TA = 748 K. The crystallites grow to a saturation value of only about 4 nm upon annealing for up to 16 h. For longer annealing times, the fcc phase starts to transform into the stable equilibrium phases.[151] These results prove that the composition of the alloy, which is directly linked to the nucleation and growth characteristics of nanocrystal formation, determines the nature and the size of the primary crystals precipitating from the amorphous phase. Hence, optimizing the stoichiometry and the annealing conditions enables the synthesis of nanostructured two-phase Zr-based materials with extremely fine nanometerscale precipitates in a residual amorphous matrix.

13: Structure Formation and Mechanical Behavior, Eckert 597

Figure 13.12 Average crystallite size, D, of the primary fcc nanocrystals formed upon annealing of amorphous (Zr0.65Al0.075 Cu0.175Ni0.10)100−xFex alloys at 773 K for 30 min vs. Fe content x.

Figure 13.13 Average crystallite size, D, of the primary fcc nanocrystals formed upon annealing of amorphous (Zr0.65Al0.075 Cu0.175Ni0.10)80Fe20 at 748 K vs. annealing time.

598

Properties

13.3.2 Conventional Solidification and Devitrification of Bulk Samples As described above, rapid solidification techniques, in some cases combined with proper annealing conditions, enable the preparation of nanoscale multiphase Al-, Mg- and Zr-based alloys. Combined with consolidation at elevated temperatures, this can yield bulk specimens with dimensions in the millimeter or centimeter range. However, in most cases phase transitions and microstructural coarsening during consolidation cannot be completely suppressed, which may deteriorate the desired mechanical properties of the material compared to the as-quenched state. Moreover, the consolidation of rapidly quenched ribbons or powders involves several processing steps, which have to be carefully optimized and carried out under high purity conditions. Hence, alternative processing techniques based on conventional casting techniques, such as die casting, copper mold casting, suction or squeeze casting, are highly desirable for large scale production of such materials. Due to the lower cooling rates of these casting techniques, they may not be applicable for multiphase nanostructure formation in all cases, but have to be combined with additional annealing treatments to produce a nanoscale microstructure, e.g. through crystallization of a bulk amorphous precursor obtained during slow cooling from the melt. In the following, selected examples for bulk multiphase microstructures in different alloy systems will be reviewed which were prepared through slow cooling methods.

Al-Based Alloys Containing Quasicrystalline Particles Bulk Al-based alloys containing a dual-phase microstructure of quasicrystalline particles embedded in an fcc Al matrix can be produced by squeeze casting the melt into a water-cooled copper mold.[156] This allows bulk rod-shape samples of several millimeters diameter and about 50 mm in length (this geometry is determined by the copper mold used) to be obtained. Although the attainable quenching rate depends on the size of the sample and is limited to some 100 Ks−1 and, therefore, is more than three orders of magnitude smaller than the typical cooling rates achievable upon melt spinning or atomization, this quenching rate proved to be sufficient for quasicrystal phase formation in Al—Mn—Ce and Al— Mn—Fe alloys.[156] Figure 13.14 shows an X-ray diffraction pattern and the corresponding DSC scan for an Al92Mn6Ce2 sample as a typical example, revealing the coexistance of an icosahedral phase, fcc Al and a small volume fraction of the competing Al6Mn intermetallic equilibrium

13: Structure Formation and Mechanical Behavior, Eckert 599

Figure 13.14 X-ray diffraction pattern of a squeeze-cast Al92Mn6Ce2 bulk sample, revealing the coexistence of an icosahedral phase, fcc Al and a small volume fraction of the competing Al6Mn equilibrium compound. The corresponding DSC trace shows the decomposition of the metastable icosahedral phase around 750 K.

phase. The DSC scan recorded at a heating rate of 40 Kmin-1 reveals the decomposition of the metastable I-phase at around 750 K. Both, the XRD pattern as well as the thermal stability data are very similar to the results for melt-spun ribbons of same composition (Fig. 13.5). Figure 13.15 shows a scanning electron microscopy (SEM) image taken from the cross-section of the squeeze cast rod, revealing a homogeneous distribution of globular-shaped quasicrystals with sizes between 0.5 and 5 μm embedded in the Al matrix. This two-phase structure is coarser than the nanoscale microstructure observed in melt-spun ribbon samples. This is due to the slower cooling rate achieved upon squeeze casting, which is even lower than that achievable upon atomization which, in turn, changes the solidification behavior with respect to nucleation and growth of the different phases. This also leads to a gradient of the microstructural scale originating from slightly different cooling conditions in the center of the rod and at the outer surface layer. Although this type of slowly cast bulk multiphase material does not exhibit a truly nanoscale microstructure, the dimensions of the different phases and their morphology are quite similar to the microstructure obtained for extruded bulk specimens prepared from atomized powders,[98,157–159] also being characterized mainly by a rather inhomogeneous microstructure consisting of submicrometer- to micrometer-sized quasicrystalline and intermetallic particles embedded in an Al matrix.

600

Properties

Figure 13.15 SEM micrograph of the cross section of a squeeze-cast Al92Mn6Ce2 bulk sample, showing homogeneously distributed quasicrystalline particles in the fcc Al matrix.

Mg-Based Nanostructured Alloys from Bulk Metallic Glasses It was shown in Section 13.3.1 that the family of Mg-based ternary alloys exhibits a high glass-forming ability upon quenching from the liquid state. Moreover, the Mg—Ln—TM amorphous alloys show a distinct supercooled liquid region before crystallization, indicating that the phase transition into a crystalline phase or phase mixture is rather difficult. This result also implies that the supercooled liquid obtained by quenching the melt from temperatures above the melting point, Tm, also has a high resistance against nucleation and growth of crystalline phases upon cooling to room temperature. While systematically studying rapid solidification in Mg-based alloys, Inoue et al.[20,131,160] developed Mg-based alloys, which can be cast into fully glassy rods or bars with cast thicknesses of several millimeters. Figure 13.16 shows the compositional dependence of the maximum diameter for formation of an amorphous phase in Mg90−xCuxY10 cylinders prepared by a high-pressure die casting method.[131] The maximum diameter, d, increases with increasing Cu content from about 3 mm for Mg80Cu10Y10 to 7 mm for Mg65Cu25Y10. A further increase in Cu content causes a decrease in the maximum achievable diameter. The significant compositional dependence of d reflects the stability of the supercooled liquid against crystallization and reveals, that the ability of bulk glass formation upon casting is closely correlated with

13: Structure Formation and Mechanical Behavior, Eckert 601 the optimum composition of the glass close to the deep eutectics of the phase diagram.[21] The thermal stability data of the cast bulk glasses reveal no appreciable difference between the bulk samples and rapidly quenched meltspun ribbons, as long as the bulk samples are fully amorphous.[20,160] Hence, partial or complete devitrification of the cast specimens by annealing at temperatures in the supercooled liquid region can be used to produce nanostructured bulk specimens. There is evidence, that the ascast bulk specimens undergo phase separation upon quenching from the melt, at least for certain alloy compoFigure 13.16 Maximum diameter sitions.[84,130] Figure 13.17 shows a for the formation of an amorphous high-resolution TEM micrograph of a phase for cast Mg90−xCuxY10 Mg—Y—Cu—Li alloy obtained by [131] cylinders. Liu et al.[130] The micrograph gives evidence of phase-separation into Curich and Cu-poor domains, which was further corroborated by small-angle X-ray scattering (SAXS) and anomalous small-angle X-ray scattering (ASAXS) studies.[85] These phase-separated domains appear to be preferred sites for nucleation of bcc Mg6Li nanocrystals.[85,86] Hence, both preceding phase separation in the supercooled liquid, perhaps already during cooling from the melt or after annealing, as well as the nucleation and growth characteristics of the competing crystalline phases seem to contribute to the microstructure evolution of bulk nanostructured materials prepared from bulk glassy specimens.

Zr-Based Bulk Alloys with Nanoscale Precipitates The high glass-forming ability of multicomponent Zr-based alloys cannot only be used for the production of completely glassy bulk specimens with dimensions in the millimeter to centimeter range, but can also be exploited for the formation of bulk nanostructured materials. For this, the as-cast glassy specimens are annealed at temperatures within the

602

Properties

Figure 13.17 High-resolution bright-field TEM micrograph of a cast Mg62Li3Cu25Y10 alloy.[131]

supercooled liquid region or close to the onset of crystallization. This procedure is based on the results first obtained for rapidly quenched thin ribbons, where sequential crystallization was observed for a variety of Zrbased bulk glass-forming alloys such as Zr—Cu—Al—Ni—Ti,[139,152,153] Zr—Al—Cu—Ni—Pd,[154] Zr—Al—Ni—Cu—M (M = Ag, Au, Pt)[155] etc. This stepwise devitrification behavior leads to primary precipitation of nanoscale intermetallic or quasicrystalline phases from the supercooled liquid, which are embedded in an amorphous phase with changed composition. Figure 13.18 shows the DSC traces for Zr55Cu30Al10Ni5 and Zr57Cu20Al10Ni8Ti5 fully amorphous rods with 5 mm diameter[161] as typical examples for bulk glass-forming Zr-based alloys. Both alloys exhibit an extended supercooled liquid region before crystallization. Whereas Zr55Cu30Al10Ni5 appears to crystallize through a single exothermic event, Zr57Cu20Al10Ni8Ti5 exhibits a stepwise crystallization behavior from the supercooled into crystalline phases, indicating a primary-type of precipitation upon devitrification. This change in crystallization mode is attrib-

13: Structure Formation and Mechanical Behavior, Eckert 603

Figure 13.18 DSC traces of (a) amorphous Zr55Cu30Al10Ni5 and (b) amorphous Zr57Cu20Al10Ni8Ti5.[161]

uted to the Ti addition. Figure 13.19 displays the DSC traces for Zr62−xCu20Al10Ni8Tix bulk amorphous cylinders with different Ti content. The alloy without Ti shows just one exothermic peak. When Ti is added, the DSC scans exhibit two exothermic peaks. With increasing Ti content, the first peak shifts to lower temperatures and the enthalpy of the second exothermic peak decreases.[162] Isothermal annealing experiments together with systematic X-ray and TEM investigations revealed[162] that for alloys with low Ti content (x < 5) the first exothermic event is related to primary precipitation of a nanoscale icosahedral quaFigure 13.19 DSC traces of sicrystalline particles (Figure 13.20). Zr62−xTixCu20Al10Ni8 amorphous cylinders with different Ti content x. Hence, similarly to Al-based alloys, so in Zr-based multicomponent alloys bulk nanostructured quasicrystal-based two-phase materials can be produced. Similar results were found recently for other Zr-based

604

Properties

Figure 13.20 X-ray diffraction patterns of Zr62−xTixCu20Al10Ni8 amorphous cylinders with different Ti content x after primary crystallization.

5-component alloy systems.[154,163,164] With increasing Ti content a metastable fcc-type phase forms besides the quasicrystals (Figure 13.20). Moreover, the size of the nanoscale grains precipitated during the first crystallization step decreases with increasing Ti content, yielding extremely fine-grained material with grain sizes of about 2 nm for x = 10. At higher temperatures the metastable icosahedral and fcc-type phases successively transform into equilibrium compounds. Figure 13.21 shows isothermal DSC traces for the Zr57Cu20Al10Ni8Ti5 bulk alloy annealed at 673 K (curve (a)) and 703 K (curve (b)), respectively. Three significantly different regions can be distinguished. Initially, there is a short period of incubation without exothermic heat flow, followed by a sharp drop of the curve indicating pronounced exothermic heat flow and rapid crystallization. Finally, a third part with small exothermic heat flow is observed, which reveals the continuing crystallization of the residual amorphous phase with lower crystallization rate. As shown in curve (a), annealing at 673 K yields an incubation time of about 4 min and

13: Structure Formation and Mechanical Behavior, Eckert 605

Figure 13.21 Isothermal DSC curves of amorphous Zr57Cu20Al10Ni8Ti5 upon annealing at (a) 673 K and (b) 703 K.[161]

Figure 13.22 High-resolution bright-field TEM micrograph of amorphous Zr57Cu20Al10Ni8Ti5 annealed at 673 K for 30 min.[152]

a duration of rapid crystallization of about 40 min. Increasing the annealing temperature to 703 K (curve (b) in Fig. 13.21), enhances diffusion and chemical redistribution, leading to reduced times for incubation (1 min) and rapid crystallization (7 min). This demonstrates that a desired volume fraction of crystallites can be obtained by adjusting the annealing temperature and time for an amorphous alloy of given composition.[152,161,162] Figure 13.22 displays a high-resolution bright-field TEM image of a Zr57Cu20Al10Ni8Ti5 bulk specimen after annealing at 673 K for 40 min as a

606

Properties

Figure 13.23 (a) X-ray diffraction patterns of (I) amorphous and (II) partially nanocrystallized Zr55Cu30Al10Ni5 (volume fraction of 25% crystallites); (b) brightfield TEM micrograph of the annealed Zr55Cu30Al10Ni5 sample showing elliptical nanosize precipitates (cf. curve II).

typical example for the microstructure obtained upon annealing. The annealed sample consists of nanosized grains with grain sizes between 2 and 10 nm which are embedded in the residual amorphous matrix. Due to the small grain size, the structure of the phases cannot be determined unambiguosly. Presumably, both nanoscale quasicrystalline and metastable fcc-type phases coexist for the alloy with x = 5 (162). The volume fraction of nano(quasi)crystals is about 40%, as estimated from the remaining crystallization enthalpy of the partially crystallized sample compared with the total crystallization enthalpy of the fully amorphous as-cast material. The Zr55Cu30Al10Ni5 alloy, which exhibits just one single exothermic crystallization event, can also be partially crystallized by proper annealing, resulting in nanocrystalline precipitates with particle sizes between 50 and 100 nm which are embedded in the residual amorphous matrix.[161] Curve (I) in Figure 13.23(a) reveals the common broad amorphous X-ray diffraction maxima for the as-cast specimen. In contrast, curve (II) exhibits additional superimposed diffraction peaks due to the crystalline particles embedded in the amorphous matrix. The bright-field TEM image in Figure 13.23(b) shows the corresponding microstructure of the annealed partially crystallized sample (Fig. 23(a), curve (II)). Elliptical crystallites were found with a length of up to 200 nm. Where possible, their aspect ratio (AR) was evaluated to yield AR = 2. From images such as Figure 13.23(b) the average size of the nanocrystalline precipitates was determined to be ≈50 nm. Hence, formation of nanocrystals embedded in

13: Structure Formation and Mechanical Behavior, Eckert 607 a residual amorphous matrix is also observed for this alloy, which appears to crystallize through one single exothermic event and not through a successive stepwise transformation into the equilibrium compounds as in the case of the Zr57Cu20Al10Ni8Ti5 glassy alloy. However, the size of the precipitating nanocrystals (2–10 nm) is much smaller for the Zr57Cu20Al10Ni8Ti5 alloy with primary-type crystallization than for the Zr55Cu30Al10Ni5 alloy (50–100 nm), indicating that the Ti-containing 5component alloy exhibits a higher nucleation rate and a smaller growth rate than the quaternary Zr—Cu—Al—Ni metallic glass. Again, this reveals that the alloy composition and the details of nucleation and growth play a decisive role for the formation of ultrafine-scaled microstructure formation of two-phase nanostructured materials.

Slowly Cooled Zr-Based Bulk Composites Besides crystallization of glassy precursors, composite materials can be produced from the melt directly during solidification. Such in-situ composites generally consist either of nano/micrometer-size intermetallic particles[161,165,166] or ductile dendritic precipitates in a residual glassy matrix.[167–171] In particular composites containing intrinsically ductile micrometer-size phases are suitable to enhance the plastic deformability of monolithic bulk metallic glasses.[167,169] For example, Hays et al.[167] reported the formation of a composite microstructure consisting of bcc β-Ti-type dendrites embedded in a glassy matrix in a Zr—Ti—Nb—Cu— Ni—Be alloy, which exhibits 5–6 % plastic strain. A similar microstructure has also been reported[169] for a Zr—Nb—Cu—Ni—Al alloy with 1.3% plastic strain. It has been suggested[167,168,170,171] that in such composite microstructures, the β phase acts as seed for the initiation of organized shear bands, and confines the propagation of individual shear bands. As typical examples for bulk Zr-based composites containing ductile phases, the microstructure of two selected alloys with compositions (A) Zr66.4Cu10.5Ni8.7Al8Nb6.4 and (B) Zr73.5Cu7Ni1Al9.5Nb9 are presented in Figure 13.24. The alloys were designed by modification of well-known bulk metallic glass-forming alloy compositions.[169] Cylindrical rods were prepared by several techniques, i.e. (i) suction, (ii) centrifugal and (iii) cold crucible casting, in order to test the effect of the processing route on the as-cast microstructure. The scanning electron microscopy (SEM) secondary electron images of such as-cast alloys (Figure 13.24) reveal a composite microstructure

Properties

608

(a)

(d)

5 µm

(b)

5 µm

(e)

5 µm

5 µm

(c)

(f)

3 µm

5 µm

Figure 13.24 SEM (SEI) micrographs of the two as-cast Zr-base alloys: (a) suction, (b) centrifugal and (c) cold crucible cast of alloy A; (d) suction, (e) centrifugal and (f) cold crucible cast of alloy B.

containing a dendritic phase embedded in a matrix. The volume fraction of the dendritic phase in alloy A, illustrated in Figures 13.24(a)–(c) is estimated to be less than 25–40 vol% while that in alloy B (Figs. 13.24(d)–(f)) is around 70–90 vol%. The amount of dendritic phase in the microstructure increases as the Nb content in the alloy increases from 6.4 to 9 at%. The microstructure of the differently cast samples of alloy A shows a distinct variation not only in the morphology of the dendrites but also in their volume fractions. In the case of suction cast rods (Fig. 13.24(a)), a homogeneous distribution of flowery shaped dendrites is observed at the center of the casting up to a depth of 300 μm from the surface. The volume

13: Structure Formation and Mechanical Behavior, Eckert 609 fraction of the dendrites was estimated to be 40 ± 5 vol% and 30 ± 5 vol% for the top and bottom regions of the rod, respectively. The average dendrite cell size is 10 ± 5 μm at the center. Primary dendrites grow along with the secondary dendrites, and their shape appears to be equiaxed (Fig. 13.24(a)). In contrast, centrifugally cast rods not only exhibit a lower volume fraction of dendrites, but they also have a more spherical shape with limited growth and less secondary dendritic arms (Fig. 13.24(b)). The volume fraction of the dendrites in such samples is around 25 ± 5 vol% with a cell size of 5 ± 2 μm. Finally, cold crucible cast rods (Fig. 13.24(c)) exhibit a mixture of fine and coarse dendrites not only at the surface but also at a depth of 300 μm beyond which the differences in the dendrite cell size decrease towards the center. At the center, the arm spacing was estimated to be 5.5 ± 0.25 μm. The interdendritic region consists of 300 to 450 nm size fine particles having a regular arrangement. Energy-dispersive (EDS) analysis of the differently cast samples of alloy A reveals that the dendrites are enriched in Zr and Nb and depleted in Cu, Ni and Al compared to the nominal alloy composition. The matrix is in general enriched in Cu and Ni. The overall microstructures of the differently solidified rods with composition B (Figs 13.24(d)–(f)) contain dendrite fractions of 75–85 vol%. The dendrite cell size is about 35 ± 5 μm and very similar for all the differently as-cast specimens. The microstructure of the suction cast rod is homogeneous at the surface as well as to a depth of 25 μm, beyond which dendrites are elongated in the radial direction of the mold with a tip diameter of less than 2 μm (Fig. 13.24(d)). At a distance of 200 μm from the surface, primary dendrites appear along with secondary and the secondary dendritic arm spacing increases gradually towards the center. Such alloys contain about 70 ± 5 vol% dendrites with a cell size of 25 ± 5 μm. The overall microstructure of the centrifugally cast rods consists of coarser dendrites with less pronounced growth of secondary dendrite arms. However, distinguishable secondary dendrite arms appear at a distance of 200 μm from the surface of the rod towards the center. These secondary dendrite arms are rather coarse and blunt. Moreover, Figure 13.24(e) demonstrates that the primary dendrites are not elongated but their cells are mostly globular with a size of 30 ± 5 μm. On the contrary, the microstructure of the cold crucible-cast rods exhibits a mixture of coarse and fine dendrites at the surface of the casting. However, the cell size of the dendrites becomes almost equal towards the center of the casting (Fig. 13.24(f)). The secondary dendrites are blunt and the estimated average diameter of the primary dendritic trunk is 25 μm, associated with a slightly higher interdendritic volume fraction, as observed in

610

Properties

the central region of the casting. At the center and the surface of the cast rod, the secondary dendrite arm spacings are estimated to be 5.5 and 3.52 μm, respectively. EDS analysis of the differently cast samples of alloy B also confirms that the dendrites are enriched in Zr and Nb and depleted in Cu, Ni and Al, corresponding to a composition of Zr80Cu3Ni0.25Al7Nb9.75. On the other hand, the matrix is enriched in Cu, Ni and Al yielding an average composition of to Zr65.5Cu19Ni3.5Al9.5Nb2.5.

Slowly Cooled Ti-Based Bulk Nanostructured Composites Among the different candidates that are promising for use as advanced high-strength materials for structural applications, particular interest has been recently devoted to Ti-based composites. The microstructure of Tibased nanostructured/glassy alloys can be schematically illustrated with the help of a pseudo-ternary phase diagram (Fig. 13.25) considering three different groups of constituent elements: (i) Ti (base element), (ii) Nb/Ta/Zr/Mo (bcc β-phase isomorphous stabilizers) + Sn, and (iii) Cu/Ni/Fe (eutectic matrix formers). The different possible phase fields are plotted on the basis of the microstructural evolution at a slow cooling rate of 10–102 Ks−1 and the metastable phase formation on different lengthscales. There are three (partly) overlapping phase fields that can be dis-

Figure 13.25 Pseudo-ternary phase formation diagram for the Ti—(Nb,Ta,Zr,Mo,Sn)—(Cu,Ni/Fe) system.

13: Structure Formation and Mechanical Behavior, Eckert 611 tinguished: a glassy region, a eutectic region (nanometer length-scale) and finally a region where a body centered cubic (bcc) β-Ti-type solid solution forms, which sometimes coexists with other crystalline phases rather than bcc or nano-eutectic phases (marked as multiphase intermetallics). The overlapping of these regions produces composite microstructures with different length-scale. The glass-forming alloy composition is marked by a circle (G) in Figure 13.25. In recent years bulk glass formation was observed first in Ti50Cu25Ni20Sn5,[172] and subsequently quite a large number of bulk metallic glass-forming Ti-containing alloys has been found, e.g. Ti—Cu— Ni—Si—B,[173] Ti—Cu—Ni—Sn—Be,[174] Ti—Cu—Ni—Sn—Be— Zr[175] or Cu-rich Ti—Cu—Ni—Zr[176] alloys. These compositions are close to the circle marked as G. However, the glass-forming ability of the multicomponent Ti-based alloys[172–175] is rather poor in comparison to Zr-based multicomponent alloys[177] and it abruptly decreases for higher (Ti + Zr) contents in the alloys.[176] Moving to the area with higher (Ti + Sn) content located to the upper left to the glassy region in the phase formation diagram yields nanostructured/glassy composites with submicronsize hcp Ti particles, e.g. for a Ti50Cu23Ni20Sn7 alloy cast to a thickness of 3 mm diameter.[178] A typical microstructure of this type of alloy is presented in Figure 13.26. The selected area diffraction pattern clearly confirms the existence of nanocrystalline particles in a glassy matrix. Modification of the glass-forming alloy composition and higher (Ti + Nb)

Figure 13.26 Bright field TEM image of the matrix microstructure of Ti50Cu23Ni20Sn7. The inset shows a selected area electron diffraction (SAED) pattern taken from the matrix confirming the formation of nanocrystalline phase(s).[178]

612

Properties

contents in a Ti50Cu20Ni20Nb10 alloy[179] lead to the formation of a nanoeutectic microstructure in 2–3 mm diameter cast rods. On the other hand, micrometer-size bcc β-Ti phase primary precipitates have been observed to form along with a nano-eutectic microstructure in a rather wide range of cooling rates from 10 to 103 Ks−1.[180] The composition of these alloys lies in the ‘composite’ region marked in Figure 13.25. Moreover, a broad region of fully bcc β-Ti phase alloys exists near to the left axis of Ti—(Nb/Ta/Mo/Zr), which can be easily understood when considering the respective binary alloy phase diagrams.[181]

Effect of Ta The variation in the (Ti + Ta) content of the alloys like, for example, (T1) Ti52.0Cu19.6Ni16.8Sn5.6Ta6.0, (T2) Ti56.0Cu16.8Ni14.4Sn4.8Ta8.0, (T3) Ti60.0Cu14.0Ni12.0Sn4.0Ta10, (T4) Ti64.0Cu11.2Ni9.6Sn3.2Ta12.0 and (T5) Ti58.8Cu1.1Ni3.2Sn5.0Ta31.9 has been investigated systematically.[182,183] These compositions are marked as (T1–T5) in Figure 13.25. They lie near to the glassy region and are shifted towards the left axis of the phase formation diagram. The microstructure of alloy T1 (2 mm  rods solidified at 102–103 Ks−1) exhibits mixed amorphous and nano/ultrafine-grained regions in the matrix. This composition lies in the ‘nanoeutectic region’ in Figure 13.25. The presence of a submicron size β-Ti phase was revealed from scanning electron microscopy (SEM) and X-ray diffraction (XRD) investigations.[182] With increasing Ti + Ta content in alloy T3 (3 mm rods) the existence of a micrometer-size β-Ti phase with dendritic morphology can be observed[182] that is homogeneously distributed in a nanostructured matrix (Fig. 13.27(a)). The X-ray diffraction (XRD) pattern (inset in Fig. 13.27(a)) clearly shows the presence of strong reflections from the β-Ti phase and broad diffuse peaks from the matrix. The details of the nanostructured matrix revealed by SEM are presented in Figure 13.27(b), showing the eutectic morphology of the phases. Slowly cooled (10–100 Ks−1) arc-melted ingots with different alloy compositions (T2–T5) show a wide variation in the volume fraction of the β-Ti dendrites, between about 20 vol%, 35 vol%, 65 vol% and 95 vol% for alloys (T2–T5), respectively.[182] A representative microstructure of alloy T3 as observed by SEM and a bright field transmission electron microscopy (TEM) image from the matrix are presented in Figures 13.27(c) and (d), respectively. The size of the dendritic phase in the arcmelted ingot (Fig. 13.27(c)) is larger than for the 2 mm rods (Fig. 13.27(a)) due to the lower cooling rate realized upon arc-melting. Also

13: Structure Formation and Mechanical Behavior, Eckert 613

Figure 13.27 SEM backscattered electron (BSE) image of an as-cast Ti60.0Cu14.0Ni12.0Sn4.0Ta10 3 mm rod showing (a) in-situ formed dendrites (β-Ti phase) in a dark matrix. (Inset: XRD pattern showing the dendrites to be a bcc phase and weak intensity from the matrix.) (b) Magnified image of the dark matrix showing the fine nano-eutectic microstructure. (c) SEM backscattered electron (BSE) image of an arc-melted Ti60.0Cu14.0Ni12.0Sn4.0Ta10 alloy showing a coarse dendritic microstructure. (d) TEM bright field image of the nano-eutectic matrix.[180,182]

the eutectic phases in the matrix are coarser than those in the faster cooled rods. The eutectic matrix mainly consists of β-Ti and g-CuTi-type phases (Fig. 13.27(d)) in the alloy T3, as reported earlier.[180]

Effect of Nb A large number of alloys with varying (Ti + Nb) content have been investigated[184] by mixing (Ti80Nb20) and (Ti40Cu28Ni24Sn8) with different portions in order to obtain (N1) Ti56.0Cu16.8Ni14.4Sn4.8Nb8.0, (N2) Ti60.0Cu14.0Ni12.0Sn4.0Nb10.0, (N3) Ti64.0Cu11.2Ni9.6Sn3.2Nb12.0and (N4) Ti66.1Cu8.0Ni4.8Sn7.2Nb13.9 alloys. The alloys have been prepared by arcmelting and were solidified in the shape of 60 mm × 15 mm × 12 mm bars.

614

Properties

Figure 13.28 SEM backscattered electron (BSE) image of arc-melted Ti60.0Cu14.0Ni12.0Sn4.0Nb10 showing (a) in-situ formed dendrites (β-Ti phase) in a dark matrix and (b) TEM bright field image showing the micrometer-size dendrites dispersed in a nanoeutectic matrix. BSE images showing an increase in the volume fraction of dendritic phase with increase in (Ti + Ta + Sn) content in (c) Ti64.0Cu11.2Ni9.6Sn3.2Nb12.0 and (d) Ti66Cu8.0Ni4.8Sn7.2Nb14 arc-melted ingots.[184]

The compositions of these alloys are indicated as (N1–N4) in the phase formation diagram (Fig. 13.25) and lie to left in the diagram according to their chemical composition. The changes in the microstructure with composition are clearly visible in Figure 13.28. Alloy N1 contains a relatively fine dendritic phase with distinct epitaxial features, indicating a strong tendency for preferential growth of the dendritic bcc β-Ti phase.[184] The volume fraction of the dendrites is about 20 vol%. TEM reveals that the matrix contains a complex eutectic structure with grain sizes of about 30–200 nm.[184] In contrast, alloy N2 (Fig. 13.28(a)) contains “lopstick”like dendrites, which are isolated by the nanostructured matrix. The volume fraction of the dendrites in alloy N2 is about 40 vol%.[184] The matrix exhibits a typical eutectic structure with an interlamellar spacing of about 50–150 nm, as shown in Figure 13.28(b). The high resolution

13: Structure Formation and Mechanical Behavior, Eckert 615 TEM image from such a fine eutectic structure mainly consists of Moiré fringes.[184] In the case of alloys (N3) and (N4), both the dendritic phase and the eutectic matrix exhibit a coarser morphology (Figs. 13.28(c) and (d)) than that observed for the former alloys (N1 and N2). The volume fraction of the β-Ti dendrites is about 60 vol% and 95 vol% for alloys N3 and N4, respectively.[184] Except for alloy N1 (dendrite composition: Ti55.8Cu1.4Ni1.0Sn20.1Ta21.7), there is only an insignificant change in the composition of the dendrites for the alloys (N2–N4), which varies within 65–67 at% Ti, 3.5–4.5 at% Cu, 2–3 at% Ni, 5.5–9.5 at% Sn, and 19.5– 20 at% Nb.

Effect of Sn The evolution of the microstructure has also been studied for Ti57−x Cu15Ni14Sn4+xNb10 (0 ≤ x ≤ 10) alloys.[185] These alloys (S1: x = 0, S2: x = 5, S3: x = 10) are located in the pseudo-ternary phase formation diagram (Fig. 13.25) from the bottom axis upwards in the direction of increasing Sn content. All the alloys were cast as 5 mm diameter cylindrical rods. The microstructures of these alloys reveal that the phase field moves with increasing Sn content from a dendrite + nanoeutectic (hypo-eutectic) morphology towards multiphase intermetallics in the hyper-eutectic region. Figure 13.29(a) shows XRD patterns for the alloys (S1–S3) with different Sn content. It can be clearly seen that more unknown intermetallics appear with increasing amounts of Sn in the alloy composition, which is also confirmed by microstructural observations [Figs 13.29(b)–(d)].

Effect of Fe, Cr, Co In order to prepare “biocompatible” nanostructure–dendrite composites, He et al.[186,187] substituted Ni in the alloy composition by Fe, Co and Cr. Since Ti-base alloys have been paid a lot of attention in recent years due to their good biocompatibility, low density, excellent corrosion resistance and good balance of mechanical properties,[188,189] this alloy design approach is highly interesting. The Young’s modulus of the so-far available commercial biocompatible Ti-base alloys is very high (110 GPa) compared with that of bone (15–25 GPa).[190] This rather high value is detrimental, since low rigidity of the implants is considered to be effective for promoting bone healing and remodeling.[189] Accordingly, it is highly desirable to develop advanced Ti-base alloys which could imitate the elastic modulus of cortical bone.[191]

616

Properties

Figure 13.29 (a) XRD patterns of as-cast 5 mm, Ti57−xCu15Ni14Sn4+xNb10 (x = 0,5,10 at%) alloys showing the formation of multiphase intermetallics with increase in Sn content. The SEM images of Ti57−xCu15Ni14Sn4+xNb10 (b) x = 0, (c) x = 5 and (d) x = 10.[185]

Along this line the microstructure and the mechanical properties of Ti60Cu14Ni(Fe,Cr,Co)12Sn4Ta (Nb)10 alloys have been investigated.[186,187] All these alloys were prepared by arc-melting under slow cooling conditions (10–100 Ks−1) and reshaped into 60 mm × 15 mm × 12 mm bars. In general there was no change in the microstructural constituents of these alloys observed compared to the earlier discussed nanoeutectic-dendrite composites. Figure 13.30 exemplifies the characteristic micrometer-scale bright β-Ti phase in an ultrafine eutectic matrix for Ti60Cu14Fe12Sn4Nb10.[186] The composition of the dendrites was estimated to be Ti64.36Cu9.34Fe7.61Sn6.99Nb11.7,[186] indicating an enrichment in Ti, Nb and Sn compared to the nominal composition of the alloy. In contrast, the matrix is enriched in Cu and Fe.[186] XRD analysis reveals the presence of the bcc β-Ti phase and an hcp α-Ti phase.[186] An increase in the (Ti + Sn) content without altering the Nb content in the alloy composition was found to increase the volume fraction of the dendritic phase.[186]

13: Structure Formation and Mechanical Behavior, Eckert 617

Figure 13.30 SEM (BSE) image of a Ti60.0Cu14.0Fe12.0Sn4.0Nb10 arc-melted ingot.[186]

13.3.3 Mechanically Attrited Powders Metastable phase formation and the development of nanostructures in multicomponent alloy systems can also be achieved by solid state processing, e.g. through mechanical alloying of elemental powder mixtures.[35,43,44,116,117,192] Similarly to liquid quenched specimens, the mechanically alloyed powders can be subsequently annealed and partially nano-crystallized in order to achieve nanoscale two- or multiphase microstructures,[44,193] which are almost identical to that of Al-, Mg- or Zrbased melt-spun ribbons or slowly cooled bulk samples.[117,144,193,194] Moreover, mechanical alloying also allows the direct synthesis of nanostructured composites by blending a metallic glass with insoluble metallic or ceramic particles which usually induce heterogeneous crystallization upon cooling from the melt.[195] This is due to the different reaction temperatures and timescales for the phase formation by solid state processing compared to quenching from the melt.[24,34,193]

Al-Based Amorphous or Quasicrystalline Alloys The work on mechanical alloying of Al-based alloys is triggered by the attempt to explore the possibilities of metastable phase formation in the solid state compared to what is already known from rapid solidification of such alloys, with special focus on the similarities and difference of phase formation under completely different thermodynamic and kinetic conditions. Moreover, mechanical alloying as an alternative method of

618

Properties

metastable phase formation combined with appropriate consolidation techniques should circumvent the limitations of melt-spinning with respect to the restrictions in geometry and size of the resulting nanostructured multiphase ribbons and, simultaneously, may also enable the production of bulk material without microstuctural coarsening as observed for consolidated atomized powders or squeeze-cast bulk samples (see Sections 13.3.1 and 13.3.2).

Al-Rich Amorphous Alloys The first attempts to synthesize Al-rich high-strength ternary and quaternary amorphous alloys with more than 80 at% Al for compositions close to that which were known for good glass-forming ability upon rapid solidifcation, were reported by Dougherty et al.[43] and Benameur et al.[196,197] in 1994. For Al—Ni—Fe—Gd alloys, a fully amorphous phase could only be achieved for a maximum Al content of 80 at%, i.e. for powder corresponding to an overall composition of Al80Ni8Fe4Gd8.[43] No complete amorphization could be achieved for larger Al contents, but the powders consist of an amorphous phase–fcc Al phase mixture, most likely also coexisting with some intermetallic compound(s). Nevertheless, the thermal stability of the amorphous phase is comparable to that of meltspun ribbons. Similar results were found for mechanically alloyed Al— Ni—Co—(Zr) powders[196,197] as well as for Al—Y—Ni—Co/Fe alloys.[198] These data indicated that the composition range for optimum amorphization is different for rapid quenching and mechanical alloying techniques. However, more recent experiments[116] employing a systematic variation and optimization of the milling parameters, i.e. using proper milling conditions such as interval-milling at low intensity corresponding to a rather low kinetic energy and resulting ambient milling temperature during milling, revealed that for Al85Y8Ni5Co2 complete amorphization can be achieved and no indications for remaining crystalline phases were found by X-ray diffraction and TEM (Fig. 13.31). From energy-dispersive X-ray analysis, the composition of the amorphous phase was proved to agree with the nominal composition within the accuracy of the measurement. This, together with the thermal stability data obtained from DSC measurements (Fig. 13.32), which are almost identical to that of melt-spun ribbons,[196,199] reveal a good homogeneity of the amorphous phase. The first sharp exothermic DSC peak at about 620 K corresponds to precipitation of fcc Al nanocrystals, whereas the second transition step is related

13: Structure Formation and Mechanical Behavior, Eckert 619

Figure 13.31 (a) X-ray diffraction patterns of mechanically alloyed Al85Y8Ni5Co2 after different milling times; (b) bright-field TEM micrograph and corresponding electron diffraction pattern of the powder after 280 h of milling.[116]

Figure 13.32 DSC traces of mechanically alloyed amorphous Al85Y8Ni5Co2 powder after 280 h of milling.[116]

to the formation of intermetallic phases from the remaining amorphous phase. Again, this is almost identical to what is known for the crystallization of melt-spun ribbons.[196,199] This seems to be the first example for complete amorphization of Alrich multicomponent alloys by mechanical alloying and, combined with proper annealing, promises a further extension of the mechanical alloying technique for the preparation of Al-rich multiphase nanostructured materials.

Nanostructured Quasicrystalline Alloys

Based on previous investigations on quasicrystal formation by mechanical attrition,[35–37] Al-rich Al—Mn—Ce and Al—Mn—Fe ternary alloys with more than 85 at% Al and a nanoscale microstructure consisting of nanoscale quasicrystalline particles embedded in an fcc Al matrix were succesfully prepared by mechanical alloying of elemental powder

620

Properties

mixtures, as well as by ball milling of pre-alloyed ingots.[116,117,156] For example, Al92Mn6Ce2 or Al91Mn7Fe2 powders with grain sizes of the Iphase of about 10–30 nm and a large volume fraction of I-phase particles of up to about 70% form upon milling. This yields powders with similar microstructural features and thermal stability as reported for melt-spun samples.[108,114] Compared to amorphization of multicomponent Al-based alloys, higher milling intensities, i.e. a higher kinetic energy during milling, has to be used for quasicrystal formation. This is consistent with previous results for metastable quasicrystal formation by mechanical alloying,[37] and can be understood in terms of the thermodynamic relations between the metastable amorphous and quasicrystalline phases. Besides formation of metastable icosahedral phases in Al—Mn-based systems, mechanical alloying can also be applied to alloy systems known to form thermodynamically stable quasicrystals.[200–202] Figure 13.33 illustrates the progress of phase formation for a mechanically alloyed elemental powder mixture with overall composition of Al70.5Mn8.5Pd21.[156] After about 20 h of milling a quasicrystalline phase with face-centered (fci) structure shows up in the X-ray diffraction pattern. The quasicrystal grain size at this stage is about 20 nm. The quasicrystal formation and the

Figure 13.33 X-ray diffraction patterns of mechanically alloyed Al81Mn14Fe5 powder after different milling times. After comparatively short milling time the stable quasicrystalline fci phase forms. The quasicrystalline powder was then blended with pure Al powder and further milled for up to 200 h.

13: Structure Formation and Mechanical Behavior, Eckert 621 resulting microstructure have been corroborated by TEM. Milling for up to 100 h does not lead to any structural changes. In order to artificially create a dual-phase microstructure of coexisting nanoscale fci particles and fcc Al phase, the initially formed quasicrystalline Al70.5Mn8.5Pd21 powder was blended with an excess of pure aluminum powder and subsequently milled for extended time (Fig. 13.33). The desired phase mixture of fci and Al phases can be obtained after blending with 20 vol% of Al and milling for additonal 5 h. However, additional SEM and TEM investigations revealed a rather poor distribution of quasicrystalline particles in the Al matrix after short milling times. The homogeneity of the fci phase distribution improves with increasing milling, but care has to be taken to properly adjust the milling parameters in order not to lead to significant alloying of Al with the quasicrystalline phase upon extended milling. This would shift the composition of the quasicrystalline phase towards a more Al-rich stoichiometry, followed by the formation of the competing Al3Pd2 phase which was reported as a milling product in an earlier study.[203] Despite the fact that further more detailed investigations are necessary to fully elaborate the potential of mechanical alloying for artifically creating nanoscale microstructures based on stable quasicrystalline phases, this synthesis route is believed to be an interesting alternative method for the preparation of such novel microstructures.

Mg-Based Metallic Glasses and Metallic Glass Composites Upon quenching Mg-based alloys from the melt, special care has to be taken to account for the high reactivity of these alloys in order to prevent contamination-induced heterogeneous crystal formation, which would reduce the castability and the thermal stability of the material. In contrast, phase formation by mechanical alloying occurs on a completely different timescale than for quenching from the melt.[24,193] Thus, mechanical alloying is expected to allow glass formation at low temperatures, which may help to circumvent the limitations of quenching techniques with respect to unwanted crystallization effects. Based on this idea a systematic study of glass formation by mechanical alloying of elemental powder mixtures of Mg-based alloys has been carried out.[129,193,204,205] Figure 13.34 shows a typical X-ray diffraction pattern for as-milled Mg—Y—Cu powders, revealing the formation of an amorphous phase after 60 h of milling. Besides, some traces of unreacted Mg, Y and Cu together with Y2O3 are visible. No complete amorphization

622

Figure 13.34 Typical X-ray diffraction patterns of mechanically alloyed Mg–Y–Cu powders, showing the formation of an amorphous phase after 60 h of milling. Besides, some traces of unreacted Mg, Y or Cu, and of Y2O3 impurities are visible.[204]

Figure 13.35 X-ray diffraction patterns of mechanically alloyed Mg55Y15Cu30 powders without dispersoids and with 5 vol% MgO, CeO2, Cr2O3 or Y2O3 after 60 h of milling.[209]

Properties was observed even after long milling times, but XRD and TEM investigations proved the formation of an amorphous phase coexisting with nanoscale crystals of about 10–20 nm in size.[204] This inhomogeneity of the material is related to the high reactivity of the constituents. Especially Mg and Y tend to readily react with oxygen traces stemming from the starting materials or from the argon inert gas atmosphere used for milling. Moreover, thin oxide layers covering the grain boundaries and particle interfaces of the starting powders are present, preventing complete alloying upon milling.[193,204] Hence, mechanically alloyed Mg—Y—Cu alloys have to be considered as nanostructured metallic glass matrix composites which form in situ during processing. These powders exhibit a supercooled liquid region of about 40 to 50 K, being almost identical to that of high-purity melt-spun ribbons of comparable composition,[204–206] which can be used to easily extrude the powders into bulk specimens. Combined with partial or complete devitrification, this yields bulk nanostructured composites consisting of two or more nanoscale phases.[207] Since the high reactivity of the powders leads to nanostructured metallic glass matrix composites already during milling, it is tempting to further intentionally introduce oxide particles into the Mg—Y—Cu metallic glass matrix in order to synthesize oxide-dispersion strengthened Mg-based metallic glasses.[208] Figure 13.35 shows X-ray diffraction

13: Structure Formation and Mechanical Behavior, Eckert 623

Figure 13.36 DSC traces of mechanically alloyed Mg55Y15Cu30 powders without dispersoids and with up to 30 vol% Y2O3 dispersoids after 60 h of milling.

patterns for an alloy with nominal composition of Mg55Y15Cu30 in comparison with the results obtained for composites blended with 5 vol% MgO, CeO2, Cr2O3 or Y2O3 oxide particles after 60 h of mechanical alloying. The XRD patterns of the composite powders reveal peaks of the oxides superimposed on the broad diffraction maxima of the amorphous phase. Besides, traces of residual unreacted material and Y2O3 are present (see above). No other crystalline phases were detected by XRD or TEM. The average grain size of the oxide particles as estimated from X-ray line broadening and TEM is about 10 to 20 nm after milling. The addition of oxide particles is not limited to a small volume fraction.[208] Figure 13.36 shows the DSC traces for Mg55Y15Cu30 powders blended with up to 30 vol% Y2O3 particles. All samples exhibit a glass transition before crystallization, revealing that mechanical alloying can yield metallic glass matrix composites with an extended supercooled liquid region for up to rather large volume fractions of particles. Detailed investigations proved that the thermal stability of the glassy matrix alloy

624

Properties

Figure 13.37 DSC scans for mechanically alloyed amorphous Zr65Al10Cu25−xNix powders with different Ni content x.[193]

is not markedly affected by the presence of the nanoscale Y2O3 dispersoids.[209]

Zr-Based Multicomponent Alloys Glass formation by mechanical alloying has been observed for several multicomponent Zr-based systems, such as Zr—Al—Ni,[209,210] Zr—Al— Cu—Ni,[44,193] Zr—Ti—Cu—Ni,[193,211] Zr—Al—Ti—Cu—Ni,[209] Zr—Al—Cu—Ni—Co,[192] etc. Complete amorphization over a wide range of compositions is achieved for such alloy systems. The mechanically alloyed powders show a glass transition and a rather wide supercooled liquid region before crystallization, as exemplified in Figure 13.37 showing the DSC traces for Zr65Al10Cu25−xNix alloys. In particular, Zr— Ti—Cu—Ni metallic glasses can be formed by mechanical alloying also for compositions in the central part of the pseudoternary phase diagram (Fig. 13.38) where glass formation by slow cooling from the melt is prevented due to the existence of a very stable low-melting Laves phase.[212] These results reveal that the glass-forming ranges accessible by mechanical alloying or by liquid quenching can differ significantly.

13: Structure Formation and Mechanical Behavior, Eckert 625

Figure 13.38 Formation ranges of amorphous Zr–Ti–Cu–Ni: the lines surround the regions where bulk samples can be prepared by slow cooling from the melt (after ref. [212]). The dots represent mechanically alloyed glassy powders with extended supercooled liquid region.[193]

Influence of Oxygen on Thermal Stability and Nanostructure Formation Contamination effects are always a concern in powder metallurgy due to the high reactivity of the fine powders produced by milling. The contamination level due to the wear debris of the milling tools (WC or steel vials and balls) is typically smaller than 0.5 at%.[193] Detailed investigations proved that the influence of such low values on the properties of the material is negligible.[44,213] The oxygen content after milling mostly stems from the contamination of the starting powders, and can reach values above 1 at%, even for careful processing under inert gas atmosphere, because of the large surface area and the high reactivity of the fine powder particles. Although some oxides are present in as-milled material with large oxygen content, e.g. for material with up to about 3 at% oxygen,[214] most of the oxygen is alloyed in the glassy phase, thus changing the actual composition and the short-range order of the alloy compared to the pure material.[144] The rather high oxygen content in general changes Tg, Tx and ΔTx compared to high purity melt-quenched alloys, as exemplified in Figure 13.39 for Zr65Al7.5Cu17.5Ni10 alloys with different oxygen content. This was investigated in detail for mechanically alloyed Zr65Al7.5Cu17.5Ni10 and Zr55Al10Cu30Ni5 alloys.[144,193,214,215] Similar to melt-spun ribbons with rather large oxygen content (see Section 13.3.1), the crystallization of mechanically alloyed powders is determined by oxygen-triggered

626

Properties

crystallization of a metastable fcc NiZr2-type phase.[209] Nevertheless, despite the rather high oxygen content the milled powders are in a truly glassy state without indications for the formation of intermetallic phases. This is in striking contrast to slowly cooled bulk specimens where already a much smaller oxygen content prevents the formation of fully glassy material.[140] Hence, mechanical alloying allows a much larger oxygen level for glass formation than slow cooling from the melt. This is due to the completely differFigure 13.39 Tg, Tx and ΔTx ent mechanisms of glass formation (heating rate 40 Kmin−1) for for the different techniques. Whereas Zr65Al7.5Cu17.5Ni10 slowly cooled bulk samples (SC), rapidly quenched cooling from the melt is governed by ribbons (RQ) and mechanically the thermodynamics and kinetics alloyed powders (MA) vs. oxygen of nucleation and growth at rather content.[214] high temperatures,[21,216] mechanical alloying is a low-temperature process involving different constraints for phase formation, thus enabling metallic glass formation also for rather large impurity contents. Considering that oxygen-triggered formation of metastable phases upon crystallization not only changes the thermal stability and crystallization sequence, but also allows sequential precipitation and transformation of nanoscale crystalline phases upon devitrification, this offers interesting perspectives for nanostructure design in such materials, yielding unique nanostructured two- or multiphase materials which cannot be obtained by any other processing technique. When the annealing treatment is combined with powder consolidation, i.e. by carefully adjusting the temperature–time regime applied for extrusion or hot pressing, this easily enables the preparation of bulk nanostructured composites from mechanically alloyed powders.

Zr-Based Metallic Glass Matrix Composites Nanostructured Zr-based metallic glass composites can also be prepared directly by mechanical alloying by blending the metallic constituents of the glass with insoluble oxide,[195,217] carbide[218] or nitride[219]

13: Structure Formation and Mechanical Behavior, Eckert 627

Figure 13.40 Bright-field TEM micrograph and corresponding electron diffraction pattern of mechanically alloyed Zr65Al7.5Cu17.5Ni10 powder containing 5 vol% CaO particles after 100 h of milling, revealing a uniform distribution of the dispersoids in the amorphous matrix.[195]

particles. For Zr-based alloys with various compositions the phase formation and the development of the microstructure upon milling was characterized by XRD, TEM and DSC measurements. The XRD patterns of composite powders show the diffraction peaks of the particles superimposed on the broad amorphous maxima.[195,213] Figure 13.40 shows a bright-field TEM image and the corresponding diffraction pattern for Zr65Al7.5Cu17.5Ni10 with 5 vol% CaO particles as a typical example. The diffraction pattern shows a diffuse halo besides some diffraction spots of the crystalline oxide particles, proving the formation of an amorphous phase coexisting with the CaO dispersoids. The bright-field image reveals a uniform distribution of 5–20 nm dispersoids embedded in the glassy matrix. Similar results were obtained for other composite samples with different volume fraction of particles, Vf, and also for Zr—Al—Cu—Ni glassy alloys containing insoluble W particles.[195,213] Figure 13.41 shows DSC traces for glassy Zr55Al10Cu30Ni5 powders blended with different volume fractions of W particles as a typical example for the thermal stability data of mechanically alloyed composite powders. The scans exhibit a glass transition at low temperatures, followed by an exothermic crystallization peak. The addition of W particles into the glassy Zr55Al10Cu30Ni5 matrix leads to a shift of the crystallization peak to slightly higher temperatures. This indicates that a small amount of W might be alloyed into the matrix and, therefore, slightly changes the overall composition of the glass. A careful examination of the thermal stability data reveals that Tg and Tx as well as the incubation time

628

Properties

Figure 13.41 DSC traces for mechanically alloyed Zr55Al10Cu30Ni5 with different volume fractions of W particles, Vf. The inset displays an enlarged part of the DSC scan for the sample with Vf = 8.5% as a typical example for the glass transition behavior of the composite powders.[213]

for crystallization under isothermal conditions remain almost unchanged upon particle addition.[213] However, crystallization of the composites proceeds faster than for the particle-free material. Hence, nanosized W particles do not significantly reduce the stability of the supercooled liquid against crystallization, but promote formation of a nanoscale microstructure upon crystallization due to interface controlled nucleation at the particle/supercooled liquid interfaces.[213] Similar data were obtained for composites containing nanoscale oxide particles with volume fractions of up to 30%.[195,217] Again, the overall thermal stability of the metallic glass remains almost unchanged compared to the particle-free material (Fig. 13.42), but the presence of nanoscale oxides promotes nanocrystal formation upon crystallization.

13.4 Mechanical Properties at Room and Elevated Temperatures On one hand, the Hall–Petch relationship predicts large strength increases as the grain size approaches the nanometer regime.[5,6] On the other hand, theory indicates that a high volume fraction of interfacial

13: Structure Formation and Mechanical Behavior, Eckert 629

Figure 13.42 Thermal stability data (heating rate 40 Kmin−1) for mechanically alloyed amorphous Zr55Al10Cu30Ni5 and for composites with different volume fraction of Y2O3 particles.[217]

regions leads to enhanced deformation due to grain boundary sliding in the low end of the nanometer range.[220] Despite a significant strength increase at room temperature and the observation of superplastic flow at elevated temperatures,[9] the room temperature ductilty of single-phase nanocrystalline metals, alloys and compounds is typically rather poor.[10] In this sense, their deformation and fracture behavior is quite similar to that of amorphous alloys, i.e. there seems to be an analogy of the deformation behavior of nanocrystalline and non-crystalline structures. Detailed descriptions of the mechanical behavior of ‘conventional’ nanostructured materials can be found in refs [7 and 8]. Here, the deformation behavior of recently developed two- or multiphase alloys with nanoscale microstructure will be described for selected Al-, Mg-, Zr- and Ti-based multicomponent alloys.

13.4.1 Al-Based Two-Phase Nanostructured Alloys Conventional high-strength aluminum alloys make use of different strengthening mechanisms, such as solid solution strengthening, precipitation hardening, grain size refinement, dispersion strengthening, work

630

Properties

hardening and fiber reinforcement. Utilizing these different mechanisms leads to an upper limit for the tensile strength of about 600 MPa at room temperature for conventional crystalline Al alloys. To develop new types of high-strength Al-based alloys with improved strength levels, the use of different strengthening mechanisms is required. This can be realized by fabricating alloys with non-periodic structures, i.e. amorphous alloys without long-range periodicity or quasicrystalline alloys. In such alloys, there are no specific slip planes and, therefore, there is no plastic deformation via dislocation motion at low stress levels. However, this, in turn, may limit the room temperature ductility of the material. To overcome this problem, different types of two-phase nanostructured materials heave been developed.

Two-Phase Nanostructured Alloys with Amorphous Matrix The starting point for the development of new high-strength Al alloys was the discovery of high tensile strengths, σf, exceeding 1200 MPa for melt-spun Al—Ln—TM[93,100] and Al—ETM—LTM[101] amorphous ribbons or amorphous wires prepared by melt-extraction.[102] Such strength levels are about twice as high as for conventional crystalline Al alloys. The achievable σf and hardness values depend, of course, on the actual composition of the respective alloy. Similar high hardness data were found for mechanically alloyed powders and consolidated bulk specimens. For example, Vickers hardness (Hv) values of about 430 were found for mechanically alloyed and consolidated Al85Y8Ni5Co2.[116] Neglecting residual porosity in the bulk samples, the fracture strength is estimated to be about 1400 MPa. Dougherty et al.[43] report even higher Hv values (625–650) for mechanically alloyed Al80Ni8Fe4Gd8, corresponding to an estimated fracture strength of about 2000 MPa. This very high value indicates the presence of additional intermetallic precipitates and/or oxides in the material. Even higher strength levels are observed for nanostructured two-phase alloys with nanoscale fcc Al particles in an amorphous matrix. Such nanoscale mixed-phase materials can be made directly upon rapid quenching or by partial devitrification of the melt-spun amorphous ribbon. They exhibit tensile fracture strengths reaching up to 1560 MPa for an Al88Ni9Ce2Fe1 alloy,[107] which is considerably larger than that of the corresponding single-phase amorphous alloy (1100 MPa). The tensile strength increases almost linearly with the volume fraction of particles, Vf, reaching its maximum value for Vf = 25%. Simultaneously, the Vickers

13: Structure Formation and Mechanical Behavior, Eckert 631 hardness increases from 280 to 400, and the Young’s modulus increases from 63 to 71 GPa with increasing Vf.[107] Similar data were found for a variety of alloys with different composition,[89,221,222] including two-phase nanogranular amorphous alloys, such as Al94V4Fe2,[119] or Al–Ti–Fe alloys with a amorphous network phase.[120] The increase in σf by the dispersion of nanoscale fcc Al particles in the amorphous matrix is due to a combination of different effects.[99] The very small fcc Al nano-grains of about 3–5nm in size are too small to contain dislocations[7,8] and have a perfect crystalline structure. In addition, the particles precipitate from a supercooled liquid/amorphous phase. Thus, the interfacial energy of the amorphous/Al particle interface is lower by about an order of magnitude[223,224] than the solid/solid interfacial energy in the case of the precipitation of a crystalline phase from a crystalline fcc Al alloy. Consequently, the amorphous/Al particle interface has a highly dense-packed structure without excess vacancies or voids.[223] This interfacial structure suppresses crack generation at the the interface, favoring the transmittance of the applied load between the different phases. In addition, the nanoscale of the fcc Al particles and the interparticle spacing are smaller than the width of the shear bands which carry deformation in the amorphous phase. Hence, the Al particles can act as an effective barrier against the shear deformation of the material. Finally, chemical redistribution and solute enrichment of the remaining amorphous matrix after precipitation might also contribute to the observed strength and hardness increase.[90,224–226] However, the tensile strength of these alloys is too high to produce bulk alloys by extrusion at low temperatures, which would be required to maintain the extremely fine nanoscale phase mixture obtained after solidification. Moreover, some of the alloys tend to significantly embrittle upon annealing at elevated temperatures, yielding fracture toughness values on the order of 1–5 MPam1/2.[43,92,103] This is believed to be due to relaxation of the amorphous phase coupled with solute redistribution.[90,224] This limits the practical use of such two-phase nanostructured alloys containing an amorphous matrix phase.

Fully Nanocrystalline Bulk Alloys Extruding atomized amorphous powders or pieces of melt-spun ribbon at temperatures near the onset of crystallization results in bulk nanostructured alloys consisting of intermetallic compounds (about 50 nm in size) and fcc Al with a grain size of up to 200 nm.[106] Alloys with this type of microstructure exhibit a tensile strength above 900 MPa, a Young’s

632

Properties

modulus of about 100 GPa, and a rotating-beam fatigue strength of about 300–350 MPa at room temperature.[227] The high strength values result from the combination of dispersion strengthening and grain refinement. Such alloys show elongations of up to 10%, indicative of a good room temperature ductility. This is attributed to the presence of dislocations, in particular in the submicrometer-sized Al phase. On the other hand, the dispersion of intermetallics together with the grain boundaries of the Al phase effectively hinders dislocation motion, thus yielding the observed strength increase. Besides, these alloys exhibit a high strain rate sensitivity exponent m = 0.7 as well as large elongations of 600–700% at a strain rate of 1 s−1 at 885 K, revealing that they can be deformed by high strain rate superplasticity. This makes them also attractive for shaping operations at moderate to elevated temperatures.

Alloys with Quasicrystalline Phases Micometer-sized quasicrystalline poly- or single crystals exhibit high strength and Young’s modulus, but are brittle at room temperature.[228–230] They also show increased ductility at elevated temperatures.[231,232] Due to their room temperature brittleness, their practical relevance has been limited to low-friction or thermal barrier coatings so far.[233,234] Hence, it is of major importance to increase the ductility of quasicrystal-based Al alloys for exploiting their potential as high-strength materials. This goal can be realized in various Al—(Mn, Cr)—Ce,[108–110] Al—(Mn, Cr)—Ce— TM,[111,112] Al—V—Ce—TM,[113] and Al—(V, Cr, Mn)—TM[114] alloys (TM = Fe, Co, Ni or Cu), forming a nanoscale phase mixture of quasicrystalline icosahedral particles coexisting with a ductile fcc Al phase upon quenching from the melt or through solid state reaction. This type of mixed structure typically forms for more than about 90 at% Al and exhibits good ductility and high room temperature tensile strength of about 1000–1340 MPa[99,112,115] (see Fig. 13.4). These properties originate from the fine distribution of globular quasicrystalline particles in an Al matrix on a nanometer scale, where the quasicrystals act as the strengthbearing component, while the Al matrix supplies ductility. The existence of a crystalline approximant phase at the interface between the I-phase particles and the fcc Al matrix improves interfacial bonding between the different phases and is, thus, important for the combination of high strength and good ductility without failure at the interface.[91] Although melt-spun ribbons exhibit ultimate strengths exceeding those of conventional crystalline Al alloys by a factor of 2–3, their small size

13: Structure Formation and Mechanical Behavior, Eckert 633

Figure 13.43 Relation between tensile strength and plastic elongation for extruded bulk I-phase-containing alloys. Data for conventional Al-based alloys are also shown for comparison.[99]

prevents engineering applications. For that reason, powder metallurgical methods such as gas atomization or mechanical attrition have been employed to create powder particles with the desired microstructure, which in a second step can be consolidated into nearly full density bulk specimens by extrusion at temperatures below the decomposition temperature of the nanoscale I-phase.[93,116,235] Figure 13.43 illustrates the relation between the tensile strength and the plastic elongation at room temperature for consolidated bulk I-phase-containing alloys[99] in comparison with data for conventional crystalline Al alloys obtained from the Japanese Institute of Steels (JIS) handbook. The tensile strength and the elongation of the bulk I-phase-based alloys with different composition are in the range of 500–850 MPa and 5–25%, respectively. Hence, both the strength and the ductility values of these alloys are superior to those of conventional Al alloys. The bulk alloys also have a high Young’s modulus of about 100 GPa and exhibit a rather high elevated temperature strength, i.e. about 350 MPa at 473 K and 200 MPa at 573 K.[105] This elevated temperature strength can even be maintained after quite long temperature exposure, e.g 1000 h at 573 K, indicating a high stability of the fine-scale microstructure against phase transition or coarsening. Similar data were also reported for Al—Mn—LTM (LTM = Fe, Co, Ni or Cu) alloys.[227] Their impact fracture energy reaches values of about 180 kJ/m2, which is

634

Properties

Figure 13.44 Compressive true stress—true strain curves of squeeze-cast Al92Mn6Ce2 and Al91Mn7Fe2 bulk samples in comparison with a gas-atomized and extruded Al–Cr–Cu–Mn sample.

significantly larger than the values for typical age-hardened crystalline Al alloys after T6 annealing treatment. Bulk I-phase containing Al-based alloys can also be obtained directly by squeeze casting without any additional consolidation step as is necessary in the case of atomized or mechanically alloyed powders (see Section 13.3.2). Although the microstructural scale of such specimens is larger than for extruded material and, therefore, this material is better regarded as a ‘submicron-to-micron-sized’ composite rather than a ‘nanosized’ twophase mixed alloy, this type of microstructure exhibits similar mechanical properties as extruded bulks. This is exemplified in Figure 13.44, comparing the room temperature deformation behavior under constant true strain rate compression conditions for squeeze-cast Al92Mn6Ce2 and Al91Mn7Fe2 alloys with that of gas-atomized and extruded Al—Cr—Cu— Mn.[156] The combination of a rather high strength of 565 MPa and good ductility of about 13% for the Al92Mn6Ce2 cast rod is comparable to the data for the extruded bulk specimens, despite the coarser, but also rather homogeneous globular microstructure of the cast material. The Al91Mn7Fe2 sample shows a lower strength, and failure already occurs after less than 6% plastic deformation. This is attributed to a dendritic microstructure formed upon casting, leading to inhomogeneous deformation and extensive slip band formation.[156] Hence, care has to be taken to optimize alloy composition and microstructure of the cast alloys. Nevertheless, the further optimization of the squeeze casting technique as a rather simple single-step process is believed to be promising for the production of Al-based bulk alloys containing quasicrystalline particles for strengthening, at least for specimens with not too large dimensions.

13: Structure Formation and Mechanical Behavior, Eckert 635

13.4.2 Mg-Based Amorphous and Nanostructured Alloys Rapidly Quenched and Cast Materials The development of high-strength two-phase nanostructured Mg-based alloys makes use of a similar concept to that already described for Albased alloys: nanoscale microstructures, consisting of amorphous and nanocrystalline phases or of nanocrystalline phases with different strength and ductility, are combined to achieve a high room temperature strength together with good ductility, and good deformability at elevated temperatures. This will be described in the following for different types of materials and samples, i.e. rapidly quenched ribbons, slowly cooled bulk samples, as well as bulk samples prepared from gas-atomized powders.

Deformation Behavior of Amorphous Alloys Most Mg-based amorphous alloys containing more than about 80 at% Mg have good bending ductility when prepared as rapidly quenched thin ribbons and can be bent through 180° without fracture. The tensile fracture strength, σf, the Young’s modulus, E, the Vickers hardness, Hv, the specific strength, σf /ρ, and the fracture elongation, εf, for different Mgbased alloys with good bending ductility are summarized in Table 13.2. Alloys belonging to the Mg–Ln–TM group exhibit higher σf values than the other amorphous systems. The highest σf value for the Mg—Ln—TM

Table 13.2 Mechanical Properties of Mg-Based Amorphous Alloys[125]

Alloy (at%) Mg70Ca10Al10 Mg90Ca2.5Ni7.5 Mg87.5Ca5Ni7.5 Mg80Y5Ni7.5 Mg85Y10Cu5 Mg80Y10Cu10 Mg91Y5Cu4

sf (MPa)

E (GPa)

Hv

sf /r (105 Nm kg−1)

ef (%)

670 670 720 830 800 820 550

35 40 47 46 44 46 30

199 182 176 224 205 218 140

3.7 3.4 3.7 2.5 2.3 2.5 2.3

1.9 1.7 1.5 1.8 1.8 1.8 1.8

636

Properties

alloys reaches 830 MPa, which is about twice as high as for conventional crystalline Mg-based alloys.[125] The highest E and Hv values found for this group of alloys are 46 GPa and 224, respectively. Although the σf values of Mg—Ca—Al amorphous alloys are considerably lower than those of Mg–Ln–TM alloys, there is no distinct reduction in the specific strength of the Mg–Ca–Al alloys, because of their very low densities. The fracture surface under tensile loading conditions typically shows a smooth region due to shear band sliding and a vein pattern due to the final fracture after sliding. This feature is the same as for other conventional amorphous alloys with good bending ductility.[125] Most of the ternary Mg-based amorphous alloys known so far exhibit a good ductility in the as-quenched state even for high concentrations of alloying elements, such as Y or Cu. However, the amorphous alloys with less than 80 at% Mg embrittle during aging and the degree of embrittlement increases with decreasing Mg content.[125] This effect is attributed to the development of short-range ordering and rearrangement of Mg and the other alloying elements upon aging.[121] Besides melt-spinning, the high glass forming ability of Mg-based alloys, in particular of Mg—Cu—Y or Mg—Ni—Y alloys, enables the production of bulk amorphous specimens by high pressure die casting or suction casting.[130,131,160] Figure 13.45 shows nominal tensile stress–elongation curves for bulk amorphous Mg80Cu10Y10 obtained at different temperatures between 273 and 473 K.[131] The deformation behavior can be devided into three different types. At room temperature (293 K) the material has a high fracture stress and shows no appreciable elongation under tensile conditions. In the temperature range between 353 and 393 K a high yield stress combined with a rather small elongation is observed. In contrast, the material exhibits a distinct yield phenomenon, followed by a significant decrease in nominal flow stress accompanying large elongations at temperatures above 423 K. The fracture stress reaches 630 MPa at 293 K and remains at this level for temperatures up to 353 K. For higher temperatures the fracture stress decreases rapidly with increasing temperature. Hence, the material keeps a mechanical strength exceeding 400 MPa for temperatures up to about 373 K. On the other hand, the elongation is below 7% for temperatures lower than 393 K, and the material reveals inhomogeneous deformation and fracture behavior without distinct work hardening in this temperature range. Above 400 K, the elongation increases rapidly to about 35%. This is due to the viscous flow of the material in this temperature range. Similar results were found for other cast bulk Mg-based amorphous specimens,[125,131] as well as for extruded amorphous samples prepared from atomized powders.[133,134]

13: Structure Formation and Mechanical Behavior, Eckert 637

Figure 13.45 Temperature dependence of the nominal tensile stress— elongation curve of amorphous Mg80Cu10Y10 bulk samples.[131]

Strength of Amorphous Alloys Containing Nanoscale Particles Partially crystalline Mg-rich alloys with more than 80 at% Mg consisting of amorphous and nanoscale hcp Mg phases can be prepared either directly by quenching from the melt or through partial crystallization.[123,131,132] About 5–10 nm sized hcp particles are dispersed homogenously in the amorphous matrix with a typical interparticle spacing of about 3–10 nm.[131,132] Compared to the fully amorphous alloy, the ultimate tensile strength, the Vickers hardness and the Young’s modulus increase for the mixed phase alloys. They also exhibit good room temperature ductility. For example, for Mg—Zn—La alloys the ultimate room temperature tensile strength increases from about 600 MPa for the amorphous phase to about 700 MPa for the mixed structure containing nanoscale hcp Mg particles.[132] In particular, the nanostructured two-phase Mg85Zn12La3 alloy exhibits a pronounced yielding phenomenon with a maximum at

638

Properties

about 3% plastic strain. This indicates that the homogeneous dispersion of ultrafine hcp particles embedded in the amorphous matrix suppresses the generation of local shear deformation and assists homogeneous plastic deformation as long as the amorphous matrix phase exhibits some ductility. The ductilization effect has been attributed to the small size of the Mg precipitates (5–10 nm). They are smaller than the width of the inhomogeneously deformed regions (10–20 nm), which allows the second phase to act as effective barrier for shear deformation by increasing the flow stress on the shear plane by the presence of the nanoscale particles. As a result, subsequent deformation on the same shear plane is suppressed, leading to an increase in plastic elongation at room temperature. With increasing temperature, the mixed phase nanostructured alloys exhibit a decrease in yield strength and an increase in the elongation, similar to that observed for fully amorphous alloys.

Deformation Behavior of Nanostructured Alloys Crystallization of the amorphous phase or extrusion at higher temperatures produces nanostructured phase mixtures consisting of hcp Mg grains and intermetallic compounds with a size of about 100–200 nm.[124,133] Figure 13.39 shows the temperature dependence of the tensile yield strength, σ0.2, for bulk nanostructured Mg87.5Cu5Y7.5[133] and Mg70Ca10Al10[124] alloys. For comparison, the data for the heat-resistant commercial crystalline WE54-T6 alloy are also shown in Figure 13.46.[236] The tensile yield strength at room temperature is 740 MPa for the Mg87.5Cu5Y7.5 alloy and 600 MPa for Mg70Ca10Al10. Although the σ0.2 values decrease with increasing temperature, both alloys keep a high tensile yield strength exceeding that of the commercial crystalline alloy up to rather high temperatures. Simultaneously with the decrease in yield strength the elongation increases with increasing temperature, in a similar way as has been shown above for fully amorphous or partially crystalline alloys of comparable composition. The specific tensile yield strength, σ0.2/ρ, of the Mg—Cu—Y and Mg—Ca—Al alloys is as high as 3.3 × 105 Nmkg−1, which is much higher than the values of typical commercial Mg-based alloys.[237] The significant increase in elongation with increasing temperature points to the possibility of superplastic deformation characteristics of the nanostructured Mg-based alloys. In fact, this has been observed for some of the alloys. For example, evidence for superplastic behavior was found

13: Structure Formation and Mechanical Behavior, Eckert 639

Figure 13.46 Temperature dependence of the tensile yield stress, σ0.2, for extruded Mg87.5Cu5Y7.5 and Mg70Ca10Al10 alloys annealed for 2 h at each testing temperature.[125] Data for the commercial WE54-T6 (Mg–Y–Nd) alloy taken from ref. [209] are also shown for comparison.

for a Mg-8.3 wt% Al-8.1 wt% Ga alloy, which was produced by rapid quenching and extrusion.[135] A maximum elongation to failure of 1080% was obtained at 573 K at a relatively high strain rate of 10−2 s−1. This has been attributed to the fine-scale equiaxed microstruture, consisting of a hcp Mg matrix with a grain size of about 2 μm and intermetallic compounds with sizes of 100–300 nm at 573 K, the optimum temperature for superplastic deformation. The superplastic strain rate is higher than for conventional superplastic aluminum and magnesium alloys. This opens perspectives for investigations of possible high strain-rate superplasticity in other Mg-based two-phase materials.

Mechanically Alloyed Powders Mechanically alloyed Mg-based alloy powders typically contain nanoscale unreacted elemental material and traces of Y2O3 homogeneously embedded in the amorphous materix even after extended

640

Properties

milling.[193,204] Nevertheless, the thermal stability of the powders is comparable with rapidly quenched or cast fully amorphous specimens, and reveals a distinct supercooled liquid region. Heating through the glass transition is accompanied by a softening of the material, leading to viscous flow behavior for temperatures above Tg. The typical behavior of mainly amorphous powders under static compression is shown in Figure 13.47 for Mg55Y15Cu30 as a typical example.[205] Figure 13.47 compares Figure 13.47 DSC trace (a) and the DSC scan for the mechanically corresponding TMA scan (b) of alloyed powder with the deformation mechanically alloyed Mg55Y15Cu30 data obtained from thermomechanipowder.[205] cal analysis (TMA). Above the glass transition temperature a distinct contraction of the sample, i.e. a penetration of the probe, occurs due to viscous flow. This allows an easy consolidation by hot pressing or extrusion in the viscous state above Tg. Figure 13.48 shows an amorphous Mg55Y15Cu30 plate prepared by hot compaction. The spherical indent in the middle of the plate is 1.5 mm deep and was produced by additional dieupset deformation of the initially flat plate above Tg, showing that the as-consolidated dense bulk sample can be easily deformed subsequently without crack formation or Figure 13.48 Example for a bulk sample of an amorphous crystallization.[129] Mg55Y15Cu30 alloy prepared by The Vickers hardness of such conconsolidation in the supercooled solidated samples is about 345– liquid region.[205] 385 Hv for samples with different composition.[129,205] Using the relation [186] this points to a fracture strength of about 1150—1280 MPa Hv = 3σy, for the mechanically attrited material. These data are significantly larger

13: Structure Formation and Mechanical Behavior, Eckert 641 than the values for as-quenched amorphous or partially crystallized ribbons or cast bulk specimens.[125,130] Thus, the nanocrystalline metallic particles or oxides present after mechanical attrition and also after consolidation effectively increase the mechanical strength of the powder metallurgically prepared specimens.

13.4.3 Zr-Based Alloys Glassy and Partially Crystallized Cast Material High strengths and good ductility have been observed in a number of multicomponent Zr-based alloys belonging to the group of bulk metallic glass-forming systems, e.g. Zr—Cu—Al—Ni—Ti[152,153,162] or Zr—Al— Cu—Ni—Pd[154] etc. Already the bulk metallic glasses are known to reveal outstanding mechanical properties at room temperature, namely a beneficial combination of very high strength, relatively low Young’s modulus, some microplasticity and high wear resistance.[238—242] In contrast, crystalline specimens of the same nominal composition are brittle and show significantly lower values of fracture strength.[238] Primary crystallization of such metallic glasses further increases the strength of the material due to the precipitation of nanoscale particles while maintaining a good ductility as long as the volume fraction of precipitates does not exceed a critical limit.[161,162,243] In contrast, fully nanocrystallized material fails in a completely brittle manner.[161] Since good ductility is observed when the alloy contains an amorphous phase, the good ductility of the material is attributed to the ductile nature of the remaining amorphous phase, which seems to retain a significant amount of free volume even after partial crystallization. The following sections focus on some details regarding the effect of nanoscale precipitates on the room temperature and elevated temperature mechanical behavior of bulk glass forming Zr-based alloys, comparing data for fully amorphous and partially crystallized Zr57Al10Cu20Ni8Ti5 and Zr55Al10Cu30Ni5 alloys as examples for this class of materials.

Room Temperature Mechanical Properties For the compression tests at room temperature a strain rate of ˙ε = 1 × 10−3 s−1 was employed. Figure 13.49 compares typical compressive stress—strain curves for as-cast fully amorphous Zr57Al10Cu20Ni8Ti5 and

642

Properties

Figure 13.49 Compressive stress—strain curve of the amorphous and partially crystallized Zr57Al10Cu20Ni8Ti5 alloy: (a) as-cast, (b) 40 vol% nanocrystals, (c) 45 vol% nanocrystals and (d) 68 vol% nanocrystals.[161]

for partially crystallized samples with different volume fractions of nanosized particles prepared by isothermal annealing at 673 K for different times.[162,244] Fully amorphous samples exhibit a combination of high strength of up to 1800 MPa, a relatively low Young’s modulus of about 70 GPa, which is comparable to Al-based alloys, and some microplasticity of up to 1%. Their fracture toughness amounts to about 55 MPam1/2, which is sufficient for engineering applications.[238] While the elastic behavior of the partially crystallized two-phase samples is almost identical to that of the fully amorphous alloy regardless of the amount of nanoparticles, only samples with a crystalline volume fraction of less than 45% show microplasticity of up to 1%. The deformation proceeds by inhomogeneous deformation localized in thin shear bands.[161,243,245] The precipitation of the fine and homogeneously dispersed nanocrystals increases hardness and the flow stress significantly. For example, the Zr57Al10Cu20Ni8Ti5 sample containing a volume fraction of 40% nanoparticles [Fig. 13.49, curve (b)] appears to provide the best compromise between high strength and good ductility. This shows that a homogeneous dispersion of nanosized particles in the amorphous matrix can lead to significant strengthening of the material. Detailed TEM studies proved that the nanoscale precipitates are free of dislocations, which may give rise to the observed high mechanical strength.[244] A further increase in the volume fraction of nanocrystals to 45% (Fig. 13.49, curve (c) ) and to 68% (Fig. 13.49, curve (d) ) mainly favors brittleness. Hence, despite a high hardness the fracture stress decreases considerably for samples containing larger volume fractions of nanocrystals than about 45%. These results agree well with data reported by several authors.[187,246—248]

13: Structure Formation and Mechanical Behavior, Eckert 643

Figure 13.50 SEM micrograph showing the fracture surface of a Zr57Al10Cu20Ni8Ti5 sample containing 25 vol% of nanocrystals embedded in an amorphous matrix.

The fracture morphology of partially crystallized samples in comparison with fully amorphous samples remains unchanged. Figure 13.50 shows a typical fracture surface for a partially crystallized Zr57Al10Cu20Ni8Ti5 specimen containing about 25 vol% of nanocrystals as a typical example. The fracture surface reveals a well-developed vein pattern, indicating that the deformation mechanism is governed by the amorphous phase and not by the nanocrystals. Similar findings were obtained for samples containing up to about 50 vol.% of nanocrystals. Assuming the viscous shear band deformation mechanism discussed in ref. [245] to be the dominant deformation mechanism at room temperature, precipitated crystallites interacting with the shear bands do not initiate fracture provided their grain size is smaller than the thickness of the shear band. When the volume fraction of nanosized precipitates becomes larger than about 50%, the brittle nature of the precipitated intermetallic compounds is likely to dominate the overall mechanical behavior, leading to the observed decrease in ductility since the deformation is no longer governed by the deformation mechanism of the amorphous phase. This is supported by the brittle deformation behavior of the sample containing 68 vol% nanocrystals (Fig. 13.49, curve (d) ).

Deformation Behavior at Elevated Temperatures In order to investigate the deformation characteristics of the supercooled liquid, compression tests at the onset of the glass transition, Tg, as

644

Properties

Figure 13.51 Compressive true stress—true strain curves of (I) amorphous, (II) and (III) partially nanocrystallized samples with 25 and 15 vol% of nanocrystals at different strain rate, ˙ε.

well as at the point of inflection of the glass transition were carried out at constant true strain rates ranging from ˙ε = 3 × 10−3 to 1 × 10−5 s−1. Because of the well-known dependence of the glass transition temperature on the heating rate,[249] a heating rate of 20 Kmin−1 for all creep specimens was applied prior to testing and the test temperature was set to the corresponding Tg value to ensure that the deformation occurs in the supercooled liquid state. Figures 13.51 (a) and (b) show typical examples for true stress—true strain curves of fully amorphous and partially crystallized Zr55Al10Cu30Ni5 samples at the onset of Tg at strain rates of ˙ε = 1 × 10−3 and 3 × 10−5 s−1, respectively. In contrast to the room temperature measurements, the overall stress level is much lower. Moreover, predominantly homogeneous flow is observed, resulting in extended plasticity without failure. Obviously, the plastic deformation starts with a stress maximum (stress overshoot), which is also observed in other metallic glasses at temperatures below and around Tg,[250,251] and then decreases to an almost constant (steady-state) flow level at plastic strains of up to 30%. The underlying reason for the presence of the stress overshoot in these bulk

13: Structure Formation and Mechanical Behavior, Eckert 645

Figure 13.52 (a) X-ray diffraction patterns of (I) amorphous, (II) partially nanocrystallized Zr55Al10Cu30Ni5 samples; (b) bright-field TEM micrograph of the specimen corresponding to the diffraction pattern (II) after creep deformation at the onset of Tg (cf. curve II in Fig. 13.51 (b) ).

metallic glassy systems is still a matter of debate. The most likely explanation for this phenomenon was given by De Hey et al.[250] They measured in the tensile creep of ribbons of Pd-based amorphous alloys at temperatures below Tg an increase in the free volume produced by the continuously ongoing shear deformation. Consequently, the viscosity in the deformed specimen, i.e. within the volume subjected to shear, decreases and, thus, the stress level decays in a test under constant strain rate (Figs 13.51 (a) and (b) ). Recent experiments on Zr55Al10Cu30Ni5 bulk samples verified the same effect to result in an increase of the creep rate in constant stress creep tests.[252] Comparing deformation curves at different strain rates (Figs. 13.51 (a) and (b) ) reveals a weak trend that the peak stress decreases from the fully amorphous state (I) to the partially crystallized material (II), if at all. A peculiar behavior is observed for the partially nanocrystallized two-phase material at the lowest strain rate of ˙ε = 3 × 10−5 s−1 (Fig. 13.51(b) ). At this low strain rate the material reveals a significant stress increase after the stress plateau, even exceeding the initial peak stress value. This increase is attributed to the formation of additional nanocrystallites during the long-term thermal exposure upon slow compression. Obviously, the volume fraction of precipitated nanocrystals has increased substantially (Fig. 13.52). Whether there is a connected change in the viscosity of the remaining amorphous matrix which may superimpose this strengthening effect needs further clarification.

Properties

646

Figure 13.53 Strain rate—stress relationship (Norton plot) of fully amorphous and partially crystallized Zr55Al10Cu30Ni5 alloys.

In order to check the occurrence of viscous Newtonian flow the data points at the stress peak (full symbols) are adopted from tests such as displayed in Figures 13.51 (a) and (b) and plotted in Figure 13.53 on a log ˙ε − log σ scale for two temperatures, i.e. the onset of Tg (circles) and the point of inflection (squares). Figure 13.53 comprises results from different testing modes, namely constant strain rate tests (full symbols), constant stress (creep) tests (half filled symbols) and from viscosity measurements using parallel plate rheometry (open symbols). At both temperatures the data points can be successfully explained applying transition state theory to the flow stress which predicts a strain rate—stress relation of the form[253,254] ε˙ = ε˙ 0 sinh

σγ 0 Ω f MkBT

Eq. (13-6)

where γ0 is the local strain produced by the shear site of volume Ωf. Both, the product γ0 ·*Ωf as well as the reference strain rate ˙ε 0 are used as fitting parameters for adapting model curves (solid lines in Fig. 13.53) to the experiments. As expected from theory, Fig. 13.53 reveals a linear relationship between the strain rate and the stress in the low stress regime, revealing purely Newtonian viscous flow with a stress exponent n (=Δ log ˙ε /Δ log σ) = 1. At high stresses, transition to non-linear stressdependent flow occurs with n-values of 4 and higher.

13: Structure Formation and Mechanical Behavior, Eckert 647

Slowly Cooled Zr-Based Bulk Composite Materials Containing Ductile Phases As shown previously, fully glassy and partially crystallized Zr-based multicomponent bulk alloys are attractive candidates as advanced highstrength materials for structural applications. The combination of high strength with high elastic strain renders such alloys quite unique in comparison to conventional microcrystalline materials. However, one major drawback for their use in engineering applications is the often limited macroscopic plastic deformability at room temperature, despite the fact that some of these alloys show perfectly elastic—plastic deformation behavior. The limited ductility of glassy and nanostructured materials is due to the absence of strain hardening and strain rate hardening.[10,255] This inadequacy is due to the formation of highly localized shear bands.[255,256] These shear bands normally appear in a distorted region to accommodate the applied strain, forming thin regions in which very large strains are concentrated. Although the local plastic strain in a shear band can be quite high,[256] only a few shear bands become active prior to failure. Such catastrophic failure can be delayed by controlling the plastic instability via generation of multiple shear bands. Leng et al.[257] demonstrated the possibility of shear band confinement under tension by fabricating laminated composite specimens consisting of a layer of metallic glass bonded between two ductile metal layers. Such a composite microstructure is expected not only to be able to hinder the rapid propagation of shear bands but also to control the instabilities responsible for early failure. A recent study[258] on pure Cu has demonstrated a tremendous increase in the strength and ductility by tailoring the microstructure through cold working and subsequent recrystallization and abnormal grain growth in order to obtain a bimodal distribution of micrometer-size grains embedded in a nanocrystalline matrix. Moreover, a few high strength and ductile nanostructured composites have been reported in Zr-[167,169,183,259,260] and Tibase[182,183,261] alloy systems where a high strength amorphous/ nanostructured matrix is combined with a ductile dendritic solid solution as a toughening phase. Figure 13.54 displays the stress—strain curves of the uniaxial compression tests under quasistatic loading at room temperature for the Zrbased alloys with compositions (A) Zr66.4Cu10.5Ni8.7Al8Nb6.4 and (B) Zr73.5Cu7Ni1Al9.5Nb9 solidified through different casting routes. The corresponding mechanical data are listed in Table 13.3. Figure 13.54 and Table 13.3 clearly demonstrate that the castings of alloy A solidified through injection (AI3, AI5), centrifugal (ACF5), suction (AS5) and cold

648

Properties

Figure 13.54 Room temperature compression test results of differently prepared as-cast Zr–Cu–Ni–Al–Nb alloys. The sample identification is provided in Table 13.3.

crucible (ACC10) casting routes possess a lower ductility (εf ≈ 1.8–2.9%) but a higher yield strength (1261–1769 MPa) and yield strain (εy = 1.6–2.3%) than alloy B. The castings of alloy B (suction (BS5, BSA4), centrifugal (BCF5) and cold crucible (BCC10) ) demonstrate a wide variation in both ductility (εf ≈ 4.0–17.5%) and yield strength (1231– 1555 MPa) with an almost constant maximum compressive strength level of 1740–1770 MPa. The arc-melted ingot has a ductility of εf ≈ 7.5% and a relatively low yield strength of 1065 MPa. Despite these differences, all the specimens show yielding and work hardening behavior. Several tests were conducted for BCC10 samples[262] to ascertain the statistical variation in yield strength, maximum strength and fracture strain, which were estimated as 1196–1248 MPa, 1708–1770 MPa and 14–17.5%, respectively. The strength and the ductility of the composites depend on the dendrite volume fraction, as well as on the morphology and the fineness of the nanostructured eutectic matrix. The differently solidified samples of alloy A[263] have a relatively low volume fraction of dendrites (25–40 vol%) embedded in an amorphous (AI3) or nanostructured eutectic matrix and possess a high strength and a low ductility (Table 13.3). The slight increase in the value of elastic modulus (93–117 GPa) for ACF5 and AS5 is due to the presence of a relatively higher volume fraction of an unknown intermetallic phase. In contrast, the most easily tailorable alloy B[262] contains a higher volume fraction of dendrites (75–85 vol%) and the

— 84 82 117 93 87 89 80 84 82 89

35 μm(t) × 1 mm(w) 3 mm × 50 mm 5 mm × 50 mm 5 mm × 75 mm 5 mm × 75 mm 10 mm × 180 mm 5 mm × 75 mm 3.8 mm × 40 mm 5 mm × 75 mm 10 mm × 180 mm Button shape

Zr66.4Cu10.5Ni8.7Al8Nb6.4a Zr66.4Cu10.5Ni8.7Al8Nb6.4b Zr66.4Cu10.5Ni8.7Al8Nb6.4b Zr66.4Cu10.5Ni8.7Al8Nb6.4c Zr66.4Cu10.5Ni8.7Al8Nb6.4d Zr66.4Cu10.5Ni8.7Al8Nb6.4e Zr73.5Cu7Ni1Al9.5Nb9d Zr73.5Cu7Ni1Al9.5Nb9f Zr73.5Cu7Ni1Al9.5Nb9c Zr73.5Cu7Ni1Al9.5Nb9e Zr73.5Cu7Ni1Al9.5Nb9g

AR AI3 AI5 ACF5 AS5 ACC10 BS5 BSA4 BCF5 BCC10 INGOT

— 1769 1697 1654 1745 1261 1555 1314 1305 1231 1065

sy (MPa) — 2.1 2.3 1.6 2.0 1.6 1.9 1.9 1.8 1.7 1.4

ey (%)

— 1794 1840 1755 1922 1620 1740 1705 1771 1754 1405

smax (MPa)

— 2.3 2.8 1.8 2.6 2.9 4.0 10.8 10.7 17.5 7.5

ef (%)

(a) Melt spinning; (b) injection casting; (c) centrifugal casting; (d) induction melting and suction casting;(e) levitation melting and cold crucible casting; (f) arc melting and suction casting; (g) arc-melted ingot.

E (GPa)

As-solidified dimension

Composition (at%)

Alloy No.

Table 13.3 Compositions of the differently prepared Zr-base alloys, their dimensions and summary of the compression test data: Young’s modulus E, yield stress sy, yield strain ey, ultimate compression stress smax, and fracture strain ef

13: Structure Formation and Mechanical Behavior, Eckert 649

650

Properties

fineness of the matrix nanostructure (BS5: 23–50 nm; BCF5: 30–200 nm; BCC10: 200–400 nm) depends on the effective cooling rate[264] and the adopted casting method,[262] which in turn alters the mechanical properties of the alloy (Table 13.3). Obviously, the arc-melted ingot possesses a lower strength and ductility than the as-cast rods (Table 13.3) due to presence of an inhomogeneous microstructure.[265]

Consolidated Bulk Specimens from Amorphous Powders The easy deformation of metallic glasses at temperatures above Tg allows the easy consolidation of fully dense bulk specimens starting from atomized powders, utilizing the Newtonian flow behavior of the supercooled liquid (further details regarding the viscous flow behavior of amorphous powders and nanostructured composites based on metallic glasses will be given in Section 13.4.5). For this, the extrusion conditions, e.g. the extrusion temperature, the extrusion ratio and the pressure, have to be optimized in order to achieve a high density of the compact without excessive temperature rise due to deformation-induced heating. For example, gas atomized amorphous Zr55Al10Cu30Ni5 powder extruded at an extrusion ratio R = 5 at 673 K yields bulk compacts with a density of 98.7%.[141–143] The tensile strength of 1520 MPa, the fracture strain of 2% and the Young’s modulus of 80 GPa of the compacted material are nearly the same as the values measured for cast bulk and melt-spun samples. Moreover, the fracture surface of the extruded bulk sample consists of a welldeveloped vein pattern without indications for interparticle fracture,[142,143] revealing a good bonding between the initial powder particles. Similar data are possible for consolidated bulk specimens when using mechanically alloyed amorphous powders as starting materials.[211] Increasing the extrusion temperature to 693 K leads to partial crystallization, due to the heat generated by friction. The Young’s modulus of the partially crystallized nanostructured composite material increases to 117 GPa and the achieved density also further increases to 99.7%. However, the precipitation of the nanocrystalline compounds embrittles the material, leading to a reduced fracture strain of 0.9% and a fracture strength of 930 MPa. This behavior is consistent with the findings for partially crystallized cast bulk samples, also showing embrittlement when the volume fraction of precipitates exceeds a critical value (see above), and stresses the importance of carefully adjusting the extrusion parameters in order to optimize the mechanical properties of the compacts. In addition,

13: Structure Formation and Mechanical Behavior, Eckert 651 the samples have to be quenched from the extrusion temperature to reintroduce a significant amount of free volume for ductilization.[154] Nevertheless, optimized consolidation promises to allow for the production of high-strength Zr-based two-phase nanostructured materials for alloys exhibiting a successive stepwise crystallization behavior through the powder metallurgy route.

13.4.4 Ti-Based Alloys A similar composite approach to avoid catastrophic failure by controlling the plastic instability via generation and interaction of multiple shear bands as used for improving the ductility of Zr-based alloys has been successfully applied to Ti-based alloys. As an example, the mechanical properties of the Ti-based composites, whose microstructure has been discussed in Section 13.3.2, are reported here. The mechanical properties under uniaxial compression at an initial strain rate of 10−4 s−1 of the dendritic phase-reinforced nanostructured eutectic composites (T1, T2, T3;  2–3 mm as-cast rods) are listed in Table 13.4. The improvement of the mechanical properties (strength and room temperature plastic deformability) for alloy (T3) compared to alloy (T1) and other Ti-base bulk glassy alloys[172] can be clearly deduced from Table 13.4. This improvement is due to the presence of the micrometer-size toughening phase (β-Ti).[182,261]

Table 13.4 Room temperature compression test results of Ti—Cu—Ni—Sn—Ta alloys: Young’s modulus E; yield stress sy; yield strain ey; fracture stress sf; and plastic strain ep[182]

Alloy T1 T2 T3

T4 T5

Cast details

E (GPa)

sy (MPa)

ey (%)

sf (MPa)

ep (%)

 2 mm  3 mm arc melted  2 mm  3 mm arc melted arc melted arc melted

63 52 49 69 73 71 89 43

1800 1516 941 1755 1525 1037 1073 894

3.1 3.1 2.1 2.8 2.3 1.7 1.4 2.3

2251 1732 2468 2440 2282 2196 2214 4027

1.4 0.8 19.2 4.7 6.0 16.5 17.9 62.5*

* Specimen without failure.

652

Properties

Figure 13.55 Effect of composition dependent dendritic volume fraction in cast Ti–Cu–Ni–Sn–Ta alloys on Young’s modulus (), plastic strain () and yield strength ().

Figure 13.55 shows the variation in the mechanical properties of arcmelted ingots Ti—Cu—Ni—Sn—Ta alloys due to different (Ti + Ta) content, which ultimately changes the volume fraction of the toughening phase due to the shift of the alloy composition across the composite phase field from right to left, as depicted in Figure 13.25. It seems that the yield strength and the plastic deformation for the alloys T2—T5 are insensitive to the volume fraction of the dendritic phase. The increase of the elastic modulus for the composites T2—T5 is due to the decrease in the volume fraction of the nanostructured matrix. Energy dispersive X-ray spectroscopy (EDX) analysis suggests that there is no significant change in the composition of the dendritic phase, which varies only between 55 and 60 at% Ti, 2–4 at% Cu, 3–4 at% Ni, ~5 at% Sn and 26–33 at% Ta.[182] However, the near beta alloy (alloy T5, in the single bcc phase field of Figure 13.25) with fully β-Ti microstructure shows a drop in the elastic modulus to 43 GPa (Fig. 13.55, Table 13.4) and super-plastic-like room temperature compressibility of 62.5% plastic strain without failure.[182] The composition of the β-Ti phase was estimated to be Ti43.6Cu0Ni1.6Sn4.9Ta52.8.[182] Probably, a high amount of Ta (~50 at%) significantly promotes the formation of α′ and α″ martensite in the β-Ti phase.[266] This drastically decreases the elastic modulus and increases Poisson’s ratio. It has been suggested that the deformation mechanism and the pronounced cold workability of near beta Ti(Nb, Ta)-alloys is also linked to the formation of deformation twins on a nanometer lengthscale.[267]

13: Structure Formation and Mechanical Behavior, Eckert 653

Figure 13.56 Effect of composition dependent dendritic volume fraction in cast Ti–Cu–Ni–Sn–Nb alloys on plastic strain plastic strain () and yield strength ().

The Ti—Cu—Ni—Sn—Nb alloys with different (Ti + Nb) content also exhibit elastic deformation and pronounced yielding followed by plastic deformation with significant work-hardening.[184] It can be clearly seen from Figure 13.56 that there is an increase in the plastic deformability and a decrease in the yield strength with increasing volume fraction of dendrites in the different alloys. The important phenomenon noticed in these alloys is a significant change in the transition behavior from elastic to plastic deformation.[184] A higher volume fraction of nano/ultrafine-size matrix (alloy N1) yields a gradual elastic—plastic transition, whereas alloy N4 with fully coarse β-Ti phase shows a sharp transition. Very similar mechanical properties were also observed when substituting Nb by Mo.[184] The existence of the dendritic β-Ti phase together with a nanoeutectic matrix (Fig. 13.29(b) ) in alloy S1 represents a hypo-eutectic microstructure resulting in a yield strength of 1226 MPa with 2.5% plastic strain.[185] Increasing the Sn content from 4 to 9 at% shifts the alloy composition from a hypo- to a hyper-eutectic microstructure, as illustrated in Figure 13.29(c). This can also be visualized by considering the Ti—Sn(Nb) pseudo-binary phase diagram.[181] Alloy S2 has a yield strength of 758 MPa but shows a poor ductility of only about 1%.[185] For 14 at% Sn (S3), the composition of the alloy completely shifts to the intermetallic area, yielding different intermetallics in the as-solidified microstructure (Fig. 13.29(d) ), as also revealed by XRD (Fig. 13.29(a) ). S3 exhibits only elastic deformation and fails at only 300 MPa fracture stress.[185]

654

Properties

Regarding the “biocompatible” nanostructure—dendrite composites, the Co-(Ti60Cu14Co12Sn4Nb10) and Fe-containing (Ti74Cu8Fe6Sn2Nb10) alloys exhibit low elastic moduli (E) of 67 and 54 GPa, respectively, similar to (E = 55 GPa) for the well-known Ti—35Nb—7Zr—5Ta (wt%) alloy.[188] However, Ti—35Nb—7Zr—5Ta also has a very low yield strength (σy = 530 MPa) compared to the novel Ti-base nanostructured composites with biocompatible elements (σy = 1051 MPa for Ti74Cu8Fe6Sn2Nb10, and σy = 1050 MPa for Ti60Cu14Co12Sn4Nb10).[187] This is a significant achievement, since this type of nanostructured composite material[186,187,268] combines a low modulus and a doubled yield strength compared to conventional Ti-base microcrystalline near beta alloys used so far for biomedical purposes.[268] The combination of appreciable ductility (up to 61%) and strength (>1800 MPa), together with high elastic strain (εy ≈ 3%) for these alloys is promising for several engineering applications like high-performance springs, micro-gears, medical devices, sports equipment, wear-resistant materials etc. Synthesizing such a composite microstructure has paved the way to develop advanced composites by manipulating the microstructures as a function of both composition and casting conditions/cooling rate. Designing new nanostructured alloys with biocompatible elements (Cr, Co, Fe) that combine both high strength and low elastic modulus (54 GPa) is promising for a wide range of biomedical applications. Also, composites having a nanostructured matrix with micrometer-size ductile dendrites can be widely used to toughen other Ti-, Mg-, La-, Pd-, Fe- and Cu-base nanostructured materials.

13.4.5 Mechanically Attrited Composites The thermal stability investigations of as-milled amorphous powders and composites reveal a distinct glass transition followed by a wide supercooled liquid region. Together with the results obtained for the deformation behavior of cast bulk specimens at elevated temperatures, this points to an extended temperature region where the material exhibits a deformation regime governed by Newtonian flow. This viscous flow behavior is not only of fundamental scientific interest, but also opens the possibility to consolidate and shape such alloys and composites into bulk parts. A better insight into the flow behavior of the supercooled liquid can be derived from viscosity measurements using parallel plate rheometry at a constant heating rate and under isothermal conditions (details are given in ref. [269] ).

13: Structure Formation and Mechanical Behavior, Eckert 655

Flow Behavior of the Supercooled Liquid and Consolidation Figure 13.58 shows viscosity measurements for dispersoid-free Mg55Y15Cu30 and for composites with 5 vol% MgO, Cr2O3 or Y2O3 particles as typical examples for the viscosity data of mechanically alloyed powders. The measured viscosity, η, is an effective viscosity due to the presence of residual unreacted mateFigure 13.57 Viscosity values, η, of rial and the dispersion of the nanosized oxide particles in the mechanically alloyed Mg55Y15Cu30 powders without dispersoids and Mg—Y—Cu supercooled liquid. For with 5 vol% MgO, Cr2O3 or Y2O3 vs. all samples, the viscosity first Temperature.[208] decreases with increasing temperature upon passing through the glass transition into the supercooled liquid regime. Crystallization at higher temperatures strongly increases the viscosity. The observed course of the viscositiy data is consistent with the results obtained for cast Mg—Y— Cu bulk glasses.[84] However, looking more closely reveals that the mechanically attrited metallic glass composites show some differences compared to the dispersoid-free material. Slight differences were also observed in thermal stability investigations of these powders, and are related to (partial) dissolution of some of the dispersoids.[209] Nevertheless, the Mg55Y15Cu30 matrix alloy and the composites exhibit a similar temperature dependence of viscosity in the supercooled liquid region. As stated above, the measured viscosities are effective viscosities due to the combination of hard particles of unreacted material together with the oxide particles which are embedded in the supercooled liquid. An estimate for the contribution of the oxide particles to the effective viscosity is possible using the Einstein equation for the flow of mixtures[270] ηeff = η · (1 + 2.5 Vf)

Eq. (13-7)

where ηeff is the effective viscosity of the mixture, η the viscosity of the Mg55Y15Cu30 supercooled liquid and Vf the volume fraction of oxide dispersoids. In a first approximation η is assumed to be identical with or without dispersoids. According to Eq. (13-7), 5 vol% oxide particles

Properties

656

should increase ηeff by 12.5%. For all Mg—Y—Cu composites the viscosity changes are larger than this value. A similar effect was observed for Zr-based composites[213,247] and can be attributed to compositional changes of the supercooled liquid. Normalizing ηeff of composite samples with the viscosity η of the glassy matrix alloy gives the relative viscosity ηrel = ηeff/η

Eq. (13-8)

which directly reveals the changes in viscosity due to the nanocrystalline particles introduced into the glass. Figure 13.59 shows a plot of the minimum relative viscosity, ηrel,min, in the supercooled liquid just before the beginning of crystallization for Zr55Al10Cu30Ni5 blended with different Vf of nanocrystalline W particles. According to Eq. (13-8), volume fractions of up to 17.5 vol% are expected to increase the effective viscosity by less than 45% if the viscosity of the metallic liquid remains the same in the presence of the particles. However, the data show a considerably larger increase in ηrel,min for the W-containing composites than predicted

Figure 13.58 Changes in the minimum relative viscosity of the supercooled liquid, ηrel,min, for mechanically alloyed Zr55Al10Cu30Ni5 composites with different Vf in comparison with data for composites containing micrometer-sized ZrC particles[247] in the same glassy matrix. The expected values according to the Einstein equation (13.7) are also shown for comparison (dashed line).[213]

13: Structure Formation and Mechanical Behavior, Eckert 657 by Eq. (13-7) (dashed line in Fig. 13.59). Similar trends were also reported for cast bulk Zr55Al10Cu30Ni5 metallic glass containing micrometer-sized ZrC particles[247] (these data are also included in Fig. 13.59 for comparison) and for partially crystallized Zr41.2Ti13.8Cu12.5Ni10Be22.5.[271] In the latter alloy this was attributed to a significant change in stoichiometry of the glass towards a composition with a higher equlibrium viscosity at a given temperature. Comparing the present viscosity data with those for cast bulk composites indicates that the viscosity increase with increasing Vf is mainly related to slight changes in the composition of the glassy matrix phase due to (partial) dissolution of particles. The viscosity increase due to alloying can be described in the framework of the free volume theory of viscous flow.[272–274] Density fluctuations in the supercooled liquid relate to the viscosity according to[272] η = η0 (T ) ⋅ exp

⎛ bv m ⎞ ⎝ vf ⎠

Eq. (13-9)

with the initial viscosity, η0, the average free volume per atom, vf, and the critical volume for flow, bvm.[272] Taking bvm to be independent of temperature and considering a temperature-dependent free volume vf[273] describes the viscosity as a function of temperature very well.[275] Assuming slight changes in packing density and short-range order as a result of dissolution of particles suggests that the dissolved atoms reduce the amount of free volume of the glass, leading to a viscosity increase. Therefore, the viscosity changes of metallic glass composites can be rationalized by considering changes in the free volume of the glassy matrix together with a minor contribution from the particles hindering the flow of the supercooled liquid. The free volume model also describes the temperature dependence of the equilibrium viscosity. Rewriting Eq. (13-9) yields the Vogel– Fulcher–Tammann (VFT) equation[274,276] η = η0 · exp[D* · T0/(T − T0) ]

Eq. (13-10)

where D* is the fragility parameter and T0 is the VFT temperature where the barriers with respect to flow would go to infinity. D* is a measure for the fragility of the glass in terms of the fragility concept, classifying material according to the temperature dependence of their kinetics in the liquid and supercooled liquid state:[276] the larger D*, the stronger is the glass.

Properties

658

Taking A = ln η0, B = bvm and D* = B/T0 as constants gives a modified form of the VFT relation[274] ln η = A + B/(T − T0)

Eq. (13-11)

which was used for fitting the data (A, B and T0 are fitting parameters). The VFT fit yields D* = 27 and T0 = 363 K for the particle-free matrix alloy. Similiar fragility parameters were found for other bulk metallic glass forming alloys.[195,269,277—279] The determined VFT-temperatures are much lower than the calorimetric Tg observed in the DSC which is characteristic for strong liquid behavior.[275] Although the fragility concept is strictly valid only for homogeneous liquids,[276] one can describe the flow behavior of the nanostructured composites in an analogous way. The presence of nanocrystals in the matrix causes only a small viscosity change according to the Einstein relation (see above). Such a small change can be neglected when compared to the overall change in viscosity upon the changes in free volume of the matrix. VFT fits for the composites, therefore, primarily describe the viscous flow behavior of the matrix. The VFT parameters derived for the Zr-based Wcontaining composites are D* = 27 − 28 and T0 = 336 − 359 K for the samples with different volume fractions of particles, thus being almost the same as the values found for the matrix alloy. There is no clear correlation between the VFT parameters and the amount of particles present in the supercooled liquid. Hence, the observed changes in the effective viscosity do not significantly affect the T0 values or the fragility parameters D* for samples with different Vf. This indicates that the changes in effective viscosity do not drastically alter the overall flow behavior of the material, and that the composites behave, overall, as equally strong liquids as the fully glassy Zr55Al10Cu30Ni5 alloy. In Figure 13.60 the viscosity data are compared with other glassforming alloys in a fragility plot.[271,272,280] This plot enables a true comparison of different systems because the viscosity is normalized to the temperature where all the materials exhibit the same viscosity of 1012 Pas,[276] which is taken as the laboratory glass transition temperature Tg. Strong glasses such as SiO2 are characterized by an Arrhenius-type temperature dependence of the viscosity. They exhibit a small VFT temperature far below the glass transition temperature and high melt viscosities.[276,280] In contrast, fragile glasses have low melt viscosities and a VFT temperature close to Tg. The fragility plot reveals that the Zr55Al10Cu30Ni5 alloy as well as the composite samples behave closer to the strong than to the fragile glasses. Similar data are known for other bulk metallic glass-

13: Structure Formation and Mechanical Behavior, Eckert 659

Figure 13.59 Fragility plot comparing mechanically alloyed Zr55Al10Cu30Ni5 composites with different volume fraction of W particles, Vf, with other “strong” and “fragile” glasses.[213]

Figure 13.60 X-ray diffraction pattern for a consolidated Zr55Al10Cu30Ni5 composite with 4 vol% W particles taken from the cross-sectional surface of a bulk sample. The inset shows the corresponding DSC scan (heating rate 40 Kmin−1).[213]

forming alloys,[84,280] which are also much more viscous than “conventional” metallic glasses like Au77.8Ge13.8Si8.4. However, the changes in effective viscosity for the composites do not drastically affect the overall flow behavior of the material, thus enabling the consolidation of dense bulk composite specimens from mechanically alloyed powders by hot pressing at temperatures above Tg. This is exemplified in Figure 13.61 for a consolidated Zr55Al10Cu30Ni5 composite with 4 vol% W particles. X-ray diffraction and TEM[213] gave no hint of

660

Properties

crystallization during compaction. The features of the DSC scans and the heat release upon crystallization of the bulk samples are essentially the same as for the as-milled powders. The bulk samples are about 99.5% dense and can be further deformed or shaped above Tg without crack formation or crystallization. Similar results were obtained for Mg- and Febased alloys.[204,209]

Room Temperature Mechanical Properties The mechanical behavior of consolidated bulk composite specimens prepared from mechanically attrited powders was investigated by room temperature Vickers microhardness, Hv, measurements. No cracks were observed radiating from the corners or sides of the Vickers indentation, indicating a rather high fracture toughness for the material. High hardness values were determined for a variety of Zr- and Mg-based alloys and composites. For example, a hardness of 5.6 ± 0.1 GPa was determined for Zr55Al10Cu30Ni5 glassy specimens. This value increases to about 6.1 ± 0.2 GPa when introducing 17.5 vol% nanocrystalline W particles. This hardness increase is similar as for cast Zr55Al10Cu30Ni5 samples containing micrometer-sized ZrC particles,[247] or for partially crystallized Zr— Al—Ni—Cu—Ag[281] or Zr—Cu—Pd—Al glasses[148] with nanoscale precipitates. Using the relation Hv = 3σy (ref. [213]) and neglecting residual porosity in the bulk samples, the yield strength, σy, is estimated to be about 1870 MPa for Zr55Al10Cu30Ni5 which increases to about 2000 MPa for a volume fraction of 17.5% nanocrystalline W particles. This is in good agreement with the results from room temperature compressive tests on Zr57Al10Cu20Ni8Ti5 composite alloys as discussed in the context of Figure 13.49. Also for Mg55Y15Cu30-based composites the hardness increases considerably upon blending with nanosized oxide particles. The Hv values vary between 3.45 GPa for the dispersoid-free material and 3.68–3.88 GPa for the samples with 5 vol% of different oxide particles, corresponding to an estimated yield strength of about 1150 MPa for Mg55Y15Cu30 and of 1.23–1280 MPa for the composites. These data are larger than the hardness values of 2.0–3.2 GPa for as-quenched amorphous ribbons and are comparable to the hardness of 3.9–4.3 GPa reported for partially or fully crystallized ribbons of the same or comparable composition.[205,206] This suggests that both the nanometer-sized crystals of unreacted constituents and oxide additions increase the mechanical strength of the mechanically alloyed Mg-composites.

13: Structure Formation and Mechanical Behavior, Eckert 661 To get a better insight into the general trend of the hardness data for different metallic glass composites containing nanoscale particles, the measured hardness data of the composites were normalized to the hardness of the particle-free glassy matrix alloy according to Hv,norm = Hv/Hv,0

Eq. (13-12)

where Hv is the measured hardness of the composite and Hv,0 is the hardness of the glassy matrix alloy. This normalizing allows direct comparison of hardness changes between different composites with different matrices and different types of particles. In contrast to the results obtained for composites containing micrometer-sized particles,[247] the normalized hardness of the mechanically alloyed composites containing nanosized particles indicates a steady (linear) increase with Vf (Fig. 13.62). While the hardness increase for the Zr55Al10Cu30Ni5-based composites is rather moderate, a strong effect is found for the Mg55Y15Cu30-based and the Zr65Al7.5Cu17.5Ni10-based composites containing nanoscale oxide particles. The TEM microstructure investigations revealed that the mechanically alloyed composite samples contain uniformly distributed nanocrystals embedded in the glassy matrix. It is hardly conceivable that any dislocation motion takes place in the composites and, therefore, no hardening in the different phases of the composite is expected.[7,8,282] Assuming no special interaction between the nanoparticles and the glassy phase except the force balance, finite element analysis of the unit cell model[224] suggests that the overall hardness of such a composite can be described by a rule of mixtures based on the volume fraction and the hardness of each phase Hv = Vf,m · Hv,m + Vf,W · Hv,W

Eq. (13-13)

where Vf,i is the volume fraction of the respective phase and Hv,i is the hardness of the glassy matrix alloy and the nanocrystalline particles, respectively. As shown in Figure 13.62, the normalized Hv,norm data fit well to a linear relation as expected from a rule of mixtures (dashed and solid lines in Fig. 13.62). This indicates that the room temperature hardness of the nanoscale composites is determined by both phases, suggesting that nanocrystalline particles are more effective for improving the room temperature strength of metallic glasses than micrometer-sized reinforcements. However, this simple approach neglects effects like partial dissolution of particles into the glassy matrix or composition changes due to the

662

Properties

Figure 13.61 Changes in normalized Vickers hardness, Hv,norm, for mechanically alloyed Zr55Al10Cu30Ni5-, Zr65Al7.5Cu17.5Ni10- and Mg55Y15Cu30-based composites as a function of volume fraction of different nanosized particles, Vf.

formation of reaction layers/concentration gradients around the particles. Moreover, possible interactions between the shear bands propagating in the glassy matrix and the particles are not considered. Such phenomena are supposed to contribute to the strengthening effect of nanocrystalline metallic precipitates in the amorphous matrix produced by partial crystallization.[148,281] Furthermore, it does not elucidate why the effectiveness of strengthening strongly differs for the various alloy composite systems investigated here, cf. the different slopes of the dashed and the solid line in Figure 13.62. Hence, a simple rule of mixtures approach can only give a first estimate for the resulting strength. Further investigations regarding the strengthening mechanisms of metallic glass composites are necessary for a better understanding of their mechanical properties. This will be a topic of future work, also with respect to their tensile and compression deformation behavior at different strain rates and temperatures.

13.5 Summary and Outlook In this overview, phenomenological results concerning the formation of nanostructured two- or multiphase materials in Al-, Mg-, Zr- and Ti-based alloy systems and the resulting mechanical properties were presented.

13: Structure Formation and Mechanical Behavior, Eckert 663 Such nanostructured materials may contain crystalline, quasicrystalline or amorphous phases and are characterized by an ultrafine microstructure, where the different phases are intimately mixed on a nanoscale. As proceessing techniques, both rapid quenching from the melt or solid state reaction can be utilized. These synthesis routes may directly lead to a twophase nanostructure. In other cases, additional heat treatment has to be employed in order to create or optimize the desired nanostructure, such as in the case of devitrification of metallic glasses. Whereas crystallization of bulk glass-forming alloys can directly yield bulk nanostructured samples with the desired microstructure, rapidly quenched ribbons, gas atomized powders or mechanically attrited powders and composites have to be subsequently consolidated to achieve dense bulk specimens. When the precursor material is a metallic glass, this can be easily done at relatively low temperatures and pressure by using the viscous flow of the supercooled liquid. At least for two-phase materials containing nanoscale crystalline or quasicrystalline particles in a residual amorphous matrix, quenching to room temperature is favorable for achieving a decent ductility. However, special care has to be taken with respect to clean processing and accurate control of consolidation parameters in order to achieve dense bulk specimens without inducing microstructural coarsening or possible phase transitions of the metastable phases from which the unique microstructure of these materials are built up. Hence, despite the possibility of producing bulk material from powders or ribbons, it is highly desirable to obtain such nanostructures directly in bulk from, e.g. through casting. In particular, the multicomponent glass-forming systems are attractive candidates for creating two-phase nanostructures because of their castability as large-scale bulk specimens and parts, and the ease of inducing nanostructure formation by precipitation of second-phase particles by simple partial devitrification of the amorphous phase/supercooled liquid. The mechanical properties of multicomponent two- or multiphase alloys with nanoscale microstructure are very encouraging regarding the combination of high tensile strength and good ductility at room temperature. For example, Al-, Mg-, Zr- or Ti-based alloys containing amorphous/nanocrystalline, amorphous/quasicrystalline or nanocrystalline/ crystalline phases can exhibit very high room temperature strength together with good ductility. At elevated temperatures, two-phase mixtures containing a residual glassy phase can be deformed by viscous flow at rather low temperatures. In addition, some of the nanocrystalline/crystalline or nanocrystalline/quasicrystalline nanostructure variants indicate superplastic behavior at rather high strain rates. Hence, these materials

664

Properties

seem to be superior to single-phase nanocrystalline materials which typically exhibit little ductility in tension for grain sizes less than about 25 nm. This class of materials provides many interesting topics for study of microstructure—property relations, offering both the possibility of discovering and developing new materials and properties, and a way to test models and understanding of mechanical deformation in ultrafine composites consisting of different phases with different mechanical properties, such as high strength in one phase and good ductility in the other. For example, materials with extremely small grain size and high density of second phase particles, i.e. a small interparticle spacing, provide many challenges to the understanding of deformation theory. Also, procedures for arresting or blunting deformation bands to increase strength, ductility or toughness have to be further developed and coupled with a detailed knowledge of the actual microstructure of the material, including investigations regarding the nature of the interfaces between the different phases. This is expected to further improve the materials properties and to lead to a basic understanding of how to design new alloys for specific applications. The development of multicomponent alloys with nanoscale microstructure as technological materials not only requires a basic understanding of the role of chemistry and structure in determining properties, but also an understanding of how to create such structures during large-scale fabrication. Such basic understanding is now developing and interest in these nanostructured materials is growing. Fabrication processes for large volume production are already being developed, and with the careful optimization of procedures for preparing specific materials for specific applications, the importance of this new class of materials is bound to grow.

Acknowledgements The author gratefully acknowledges the support of the German Science Foundation (DFG Schwerpunktprogramm “Unterkühlte Metallschmelzen” and “Quasikristalle”; Grants Ec 111/7, Ec 111/8, Ec 111/9, and He 1872), of the EU within the framework of the research and training networks on Bulk Metallic Glasses (HPRN-CT-2000-00033) and Ductile BMG Composites (MRTN-CT-2003-504692), of the Alexander von Humboldt-Foundation, and of the German Academic Exchange Service (DAAD). Special thanks are given to many of my coworkers over the years, in particular B. Bartusch, K. Buchholz, M. Calin, J. Das, S. Deledda, A. Gebert, A. Gümbel, W. Gude, G. He, M. Heilmaier, N. Ismail,

13: Structure Formation and Mechanical Behavior, Eckert 665 K.B. Kim, A. Kübler, U. Kühn, U. Kunz, S. Kuszinski, W. Löser, N. Mattern, H. Lehmann, A. Reger-Leonhard, N. Schlorke-de Boer, F. Schurack, H. Schulze, S. Scudino, M. Seidel, M. Stoica, L.Q. Xing, W. Xu, P. Yu, L.C. Zhang, W.Y Zhang, and Z.F. Zhang who have all contributed in one way or another to the results presented in this overview. Finally, I would like to thank R. Busch, A.L. Greer, K. Hono, A. Inoue, J.Z. Jiang, W.L. Johnson, C.C. Koch, U. Köster, Y. Li, E. Ma, N. Nishiyama, J.H. Perepezko, K. Samwer, L. Schultz, D.J. Sordelet, M.L. Sui, W.H. Wang, B.C. Wei and A.R. Yavari for many valuable discussions.

References 1. Gleiter, H., Prog. Mater. Sci. 33:223 (1989). 2. Siegel, R.W., in: Mechanical Properties and Deformation Behavior of Materials Having Ultrafine Microstructures, (M. Nastasi, D.M. Parkin, and H. Gleiter, eds.), pp. 509, NATO ASI Series, Vol. 233, Kluwer, Dordrecht (1993). 3. Edelstein, A.S., and Cammarata, R.C. (eds), Nanomaterials: Synthesis, Properties and Applications, IOP Publishing, Bristol (1996). 4. Andres, R.P., Averback, R.S., Brown, W.L., Brus, L.E., Goddard III, W.A., Kaldor, A., Louie, S.G., Moscovits, M., Peercy, P.S., Riley, S.J., Siegel, R.W., Spaepen, F., and Wang, Y., J. Mater. Res. 4:704 (1989). 5. Hall, E.O., Proc. Phys. Soc. London B 64:747 (1951). 6. Petch, N.J., J. Iron Steel Inst. 174:25 (1953). 7. Weertman, J.R., and Averback, R.S., in: Nanomaterials: Synthesis, Properties and Applications, Chapter 13, (A.S. Edelstein and R.C. Cammarata, R.C., eds.), pp. 323, IOP Publishing, Bristol (1996). 8. Morris, D.G., Mechanical Behaviour of Nanostructured Materials, Materials Science Foundations, Vol. 2, Trans Tech Publ., Zürich (1998) 9. Weertman, J.R., Farkas, D., Hemker, K., Kung, H., Mayo, M., Mitra, R., and Van Swygenhoven, H., MRS Bull. 24(2):44 (1999). 10. Koch, C.C., Morris, D.G., Lu, K., and Inoue, A., MRS Bull. 24(2):55 (1999). 11. Hadjipanayis, G.C., and Siegel, R.W. (eds), Nanophase Materials: Synthesis—Properties—Applications, Kluwer, Dordrecht (1994). 12. Klement W. Jr., Willens, W.H., and Duwez, P., Nature 187:869 (1960). 13. Jones, H., Rapid Solidification of Metals and Alloys, The Institute of Metallurgists, London (1982). 14. Anantharaman, T.R., and Suryanarayana, C., Rapidly Solidified Metals: A Technological Overview, Trans Tech Publ, Zürich (1987). 15. Suryanarayana, C., and Jones, H., in: Materials Science and Technology, Vol. 15, (R.W. Cahn, P. Haasen and E.J. Kramer, eds.), pp. 55, VCH Verlagsgesellschaft, Weinheim (1991). 16. Groeber, H., and Hanemann, H., Arch. Eisenhüttenwesen 11:199 (1937). 17. Serita, Y., Ishikawa, T., and Kimura, F., J. Jpn. Inst. Light Metals 20:1 (1970).

666

Properties

18. Hinesly, C.P., and Morris, J.G., Metall. Trans. 1:1476 (1970). 19. Armstrong, G.R., and Jones, H., in: Solidification and Casting of Metals, pp. 454, The Metals Society, London (1979). 20. Inoue, A., Mater. Trans. JIM 36:866 (1995). 21. Johnson, W.L., in: Metastable, Mechanically Alloyed and Nanocrystalline Materials, (R. Schulz, ed.), pp. 35, Mater. Sci. Forum, Vol. 225–227, Trans Tech Publ., Zürich (1996). 22. Benjamin, J.S., Sci. Am. 234:40 (1976). 23. Benjamin, J.S., Metall. Trans. 1:2943 (1970). 24. Schultz, L., and Eckert, J., in: Glassy Metals III, Topics in Applied Physics, Vol. 72, (H.J. Güntherodt and H. Beck, eds.), pp. 69, Springer, Berlin (1994). 25. Koch, C.C., in: Materials Science and Technology, Vol. 15, (R.W. Cahn, P. Haasen and E.J. Kramer, eds.), pp. 193, VCH Verlagsgesellschaft, Weinheim (1991). 26. Shingu, P.H. (ed.), Mechanical Alloying, Mater. Sci. Forum, Vol. 88–90, Trans Tech Publ., Zürich (1992). 27. Yavari, A.R. (ed.), Mechanically Alloyed and Nanocrystalline Materials, Mater. Sci. Forum, Vol. 179–181, Trans Tech Publ., Zürich (1995). 28. Schulz, R. (ed.), Metastable, Mechanically Alloyed and Nanocrystalline Materials, Mater. Sci. Forum, Vol. 225–227, Trans Tech Publ., Zürich (1996). 29. Fiorani, D., and Magini, M. (eds.), Synthesis and Properties of Mechanically Alloyed and Nanocrystalline Materials, Mater. Sci. Forum, Vol. 235–238, Trans Tech Publ., Zürich (1997). 30. Baró, M.D., and Suriñach, S. (eds.), Mechanically Alloyed, Metastable and Nanocrystalline Materials, Mater. Sci. Forum, Vol. 269–272, Trans Tech Publ., Zürich (1998). 31. Calka, A., and Wexler, D. (eds.), Metastable, Mechanically Alloyed and Nanocrystalline Materials, Mater. Sci. Forum, Vol. 312–314, Trans Tech. Publ., Zürich (1999); J. Metastable and Nanocrystalline Materials 2–6 (1999). 32. Fecht, H.J., in: Nanomaterials: Synthesis, Properties and Applications, (A.S. Edelstein and R.C. Cammarata, eds.), pp. 89, Chapter 5, IOP Publishing, Bristol (1996). 33. Schwarz, R.B., in: Mechanically Alloyed, Metastable and Nanocrystalline Materials, (M.D. Baró and S. Suriñach, eds.), pp. 665, Mater. Sci. Forum, Vol. 269–272, Trans Tech. Publ., Zürich (1998). 34. Koch, C.C., Nanostructured Mater. 2:109 (1993). 35. Eckert, J., Schultz, L., and Urban, K., Appl. Phys. Lett. 55:117 (1989). 36. Eckert, J., Schultz, L., and Urban, K., Acta Metall. Mater. 39:1497 (1991). 37. Eckert, J., Schultz, L., and Urban, K., Europhys. Lett. 13:349 (1990). 38. Eckert, J., Holzer, J.C., Krill III, C.E., and Johnson, W.L., J. Mater. Res. 7:1751 (1992). 39. Hellstern, E., Fecht, H.J., Fu, Z., and Johnson, W.L. J. Appl. Phys 65:305 (1989). 40. Fecht, H.J., Hellstern, E., Fu, Z., and Johnson, W.L. Metall. Trans A 21:2333 (1990). 41. Benghalem, A., and Morris, D.G., Acta Metall. Mater. 42:4071 (1994).

13: Structure Formation and Mechanical Behavior, Eckert 667 42. Eckert, J., Holzer, J.C., Krill III, C.E., and Johnson, W.L., J. Appl. Phys. 73:2794 (1993). 43. Dougherty, G.M., Shiflet, G.J., and Poon, S.J., Acta Metall. Mater. 42:2275 (1994). 44. Seidel, M., Eckert, J., and Schultz, L., J. Appl. Phys. 77:5446 (1995). 45. Yang, J.S.C., and Koch, C.C., J. Mater. Res. 5:325 (1990). 46. Schlump, W., and Grewe, H., in: New Materials by Mechanical Alloying Techniques, (E. Arzt and L. Schultz, eds.), pp. 307, DGM Informationsgesellschaft, Oberursel (1989). 47. Shingu, P.H., Ishihara, K.N., and Kuyama, J. in: Proc. 34th Japan Congress on Materials Research, pp. 19, Society of Materials Science, Kyoto (1991). 48. Swinkels, F.B., Wilkenson, D.S., Arzt, E., and Ashby, M.F., Acta Metall. 31:1829 (1983). 49. Gutmanas, E.Y., Trusov, L.I., and Gotman, I., Nanostructured Mater. 4:893 (1994). 50. Sadangi, R.K., Kear, B.H., and McCandlish, L.E., Nanostructured Mater. 2:563 (1993). 51. Cline, C.F., and Hopper, R.W., Scr. Metall. 11:1137 (1977). 52. Schwarz, R.B., Kasiraj, P., Vreeland, T., and Ahrens, T.J., Acta Metall. 32.1243 (1984). 53. Jain, M., and Christman, T., Acta Metall. Mater. 42:1901 (1994). 54. Suryanarayana, C., Int. Mater. Rev. 40.21 (1995). 55. Vance, R.R., and Courtney, T.H., Scr. Metall. Mater. 26:1435 (1992). 56. He, L., and Ma, E., Mater. Sci. Eng. A 204:240 (1995). 57. Shaik, G.R., and Milligan, W.W., Metall. Mater. Trans. A 28:895 (1997). 58. Sanders, P.G., Fougere, G., Thompson, L.J., Eastman, J.E., and Weertman, J., Nanostructured Mater. 8:243 (1997). 59. Morris, D.G., and Morris, M.A., Acta Metall. Mater. 39:1763 (1991). 60. Birringer, R., Mater. Sci. Eng. A 117:33 (1989). 61. Gertsman, V.Y., and Birringer, R., Scr. Metall. Mater. 30:577 (994). 62. Hahn, H., Logas, J., and Averback, R.S., J. Mater. Res. 5:609 (1990). 63. Averback, R.S., Höfler, H.J., and Tao, R., Mater. Sci. Eng. A 166:169 (1993). 64. Zener, C., quoted by Smith, C.S., Trans. AIME 175:47 (1948). 65. Lu, K., Nanostructured Mater. 2:643 (1993). 66. Grundmann, U., Heilmaier, M., and Schultz, L., Scr. Mater. 37:103 (1997) 67. Heilmaier, M., Saage, H., and Eckert, J., Mater. Sci. Eng. A 239–240:652 (1997). 68. Lu, K., Mater. Sci. Eng. R 16:161 (1996). 69. Köster, U., and Herold, U., in: Glassy Metals I, Topics in Applied Physics, Vol. 46, (H.J. Güntherodt and H. Beck, eds.), pp. 225, Springer, Berlin (1981). 70. Köster, U., in: Synthesis and Properties of Mechanically Alloyed and Nanocrystalline Materials, (D. Fiorani and M. Magini, eds.), pp. 377, Mater. Sci. Forum, Vol. 235–238, Trans Tech Publ., Zürich (1997). 71. Köster, U., Rüdiger, A., and Meinhardt, J., in: Advances in Nanocrystallization, (A.R. Yavari, ed.), pp. 9, Mater. Sci. Forum, Vol. 307, Trans Tech Publ., Zürich (1997); J. Metastable and Nanocrystalline Materials 1:9 (1999).

668

Properties

72. Köster, U., Spassow, T., and Sutton, M., in: Proc. 3rd Int. Workshop on NonCrystalline Solids, Matalascañas 1991, (A. Conde, C.F. Conde and M. Millán, eds.), pp. 149, World Scientific, Singapore (1992). 73. Marcus, M., and Turnbull, D., Mater. Sci. Eng. 23:211 (1976). 74. Ohlberg, S.M., Golob, H.R., and Stickler, D.W., in: Nucleation and Crystallization in Glasses and Melts, pp. 55, The American Ceramic Society, Columbus, OH (1962). 75. Ramsden, A.H., and James, P.F., J. Mater. Sci. 19:1406 (1984). 76. Huston, E.L., Cahn, J.W., and Hilliard, J.E., Acta Met. 14:1053 (1966). 77. Chou, C.-P., and Turnbull, D., J. Non-Cryst. Solids 17:169 (1975). 78. Chen, H.S., Mater. Sci. Eng. 23:151 (1976). 79. Tanner, L.E., and Ray, R., Scr. Met. 14:657 (1980). 80. Pelton, A.R., and Tanner, L.E., in: Rapidly Quenched Metals, (S. Steeb and H. Warlimont, eds.), pp. 747, North-Holland Physics Publisher, Amsterdam (1985). 81. Busch, R., Schneider, S., and Johnson, W.L., Appl. Phys. Lett. 67:1544 (1995). 82. Schneider, S., Geyer, U., Thiyagarajan, P., Busch, R., Schulz, R., Samwer, K., and Johnson, W.L., in: Metastable, Mechanically Alloyed and Nanocrystalline Materials, (R. Schulz, ed.), pp. 59, Mater. Sci. Forum, Vol. 225–227, Trans Tech Publ., Zürich (1996). 83. Schneider, S., Geyer, U., Thiyagarajan, P., and Johnson, W.L., in: Synthesis and Properties of Mechanically Alloyed and Nanocrystalline Materials, (D. Fiorani and M. Magini, eds.), pp. 337, Mater. Sci. Forum, Vol. 235–238, Trans Tech Publ., Zürich (1997). 84. Busch, R., Liu, W., and Johnson, W.L., J. Appl. Phys. 83:4134 (1998). 85. Liu, W., Johnson, W.L., Schneider, S., Geyer, U., and Thiyagarajan, P., Phys. Rev. B 59:11755 (1999). 86. Liu, W., Ph.D. Thesis, California Institute of Technology, Pasadena (1998). 87. Busch, R., and Johnson, W.L., Appl. Phys. Lett. 72:2695 (1998). 88. Kim, Y.H., Inoue, A., and Masumoto, T., Mater. Trans. JIM 31:747 (1990). 89. Chen, H., He, Y., Shiflet, G.J., and Poon, S.J., Scr. Metall. 25:1421 (1991). 90. He, Y., Shiflet, G.J., and Poon, S.J., Acta Metall. Mater. 43:83 (1995). 91. Inoue, A., Yamamoto, M., Kimura, H.M., and Masumoto, T., J. Mater. Sci. Lett. 6:194 (1987). 92. Inoue, A., Ohtera, K., Tsai, A.P., and Masumoto, T., Jpn. J. Appl. Phys. 27:L479 (1988). 93. He, Y., Poon, S.J., and Shiflet, G.J., Science 241:1640 (1988). 94. Inoue, A., and Masumoto, T., in: Encyclopedia of Materials Science and Engineering, Vol. 2, (R.W. Cahn, ed.), pp. 660, Pergamon Press, Oxford (1990). 95. Inoue, A., Nakazato, K., Kawamura, Y., and Masumoto, T., Mater. Sci. Eng. A 179/180:654 (1994). 96. Kimura, H., Sasamori, K., and Inoue, A., Mater. Trans. JIM 37:1722 (1996). 97. Inoue, A., Mater. Sci. Eng. A 179/180:57 (1994). 98. Inoue, A., and Kimura, H.M., in: Synthesis and Properties of Mechanically Alloyed and Nanocrystalline Materials, (D. Fiorani and M. Magini, eds.), pp. 873, Mater. Sci. Forum, Vol. 235–238, Trans Tech Publ., Zürich (1997).

13: Structure Formation and Mechanical Behavior, Eckert 669 99. Inoue, A., and Kimura, H.M., in: Quasicrystals, (J.M. Dubois, P.A. Thiel, A.P. Tsai and K. Urban, eds.), pp. 495, Mater. Res. Soc. Symp. Proc., Vol. 553, Materials Research Society, Warrendale, PA (1999). 100. Inoue, A., Ohtera, K., Tsai, A.P., and Masumoto, T., Jpn. J. Appl. Phys. 27:L280 (1988). 101. Tsai, A.P., Inoue, A., and Masumoto, T., Metall. Trans. A 19:1369 (1988). 102. Inoue, A., Amiya, K., Yoshii, I., Kimura, H.M., and Masumoto, T., Mater. Trans. JIM 35:485 (1994). 103. Inoue, A., Ohtera, K., Tsai, A.P., Kimura, H.M., and Masumoto, T., Jpn. J. Appl. Phys. 27:L1579 (1988). 104. Inoue, A., Kita, K., Ohtera, K., and Masumoto, T., J. Mater. Sci. Lett. 7:1287 (1988). 105. Inoue, A., Kimura, H.M., and Kita, K., in: New Horizons in Quasicrystals, A.I. Goldman, D.J. Sordelet, P.A. Thiel and J.M. Dubois, eds.), pp. 256, World Scientific, Singapore (1997). 106. Ohtera, K., Inoue, A., Terabayashi, T., Nagahama, H., and Masumoto, T., Mater. Trans. JIM 33:775 (1992). 107. Kim, Y.H., Inoue, A., and Masumoto, T., Mater. Trans. JIM 32:599 (1991). 108. Inoue, A., Watanabe, M., Kimura, H.M., Takahashi, F., Nagata, A., and Masumoto, T., Mater. Trans. JIM 33:723 (1992). 109. Watanabe, M., Inoue, A., Kimura, H.M., Aiba, T., and Masumoto, T., Mater. Trans. JIM 34:162 (1993). 110. Inoue, A., Watanabe, M., Kimura, H.M., and Masumoto, T., Sci. Rep. Res. Inst. Tohoku Univ. A 38:138 (1993). 111. Kimura, H.M., Sasamori, K., Watanabe, M., Inoue, A., and Masumoto, T., Mater. Sci. Eng. A 181/182:845 (1994). 112. Inoue, A., Kimura, H.M., Sasamori, K., and Masumoto, T., Mater. Trans. JIM 36:6 (1995). 113. Kimura, H.M., Inoue, A., Sasamori, K., Kim, Y.H., and Masumoto, T., Mater. Trans. JIM 36:1004 (1995). 114. Inoue, A., Kimura, H.M., Sasamori, K., and Masumoto, T., Mater. Trans. JIM 37:1287 (1996). 115. Inoue, A., Kimura, H.M., Sasamori, K., and Masumoto, T., Mater. Trans. JIM 35:85 (1994). 116. Schurack, F., Börner, I., Eckert, J., and Schultz, L., in: Metastable, Mechanically Alloyed and Nanocrystalline Materials, (A. Calka D., and Wexler, eds.), pp. 49, Mater. Sci. Forum, Vol. 312–314, Trans Tech Publ., Zürich (1999); J. Metastable and Nanocrystalline Materials 2–6:49 (1999). 117. Schurack, F., Eckert, J., and Schultz, L., Nanostructured Mater. 12:107 (1999). 118. Inoue, A., Kimura, H.M., and Sasamori, K., Advanced Mater. 4:91 (1998). 119. Inoue, A., Kimura, H.M., and Masumoto, T., Nanostructured Mater. 7:363 (1996). 120. Kimura, H.M., Sasamori, K., and Inoue, A., Mater. Trans. JIM 37:1722 (1996). 121. Inoue, A., Ohtera, K., Kita, K., and Masumoto, T., Jpn. J. Appl. Phys. 27:L2248 (1988).

670

Properties

122. Inoue, A., Kohinata, M., Ohtera, K., Tsai, A.P., and Masumoto, T., Mater. Trans. JIM 30:378 (1989). 123. Kim, S.G., Inoue, A., and Masumoto, T., Mater. Trans. JIM 31:929 (1990). 124. Kato, A., Horikiri, H., Inoue, A., and Masumoto, T., Mater. Sci. Eng. A 179/180:707 (1994). 125. Inoue, A., and Masumoto, T., Mater. Sci. Eng. A 173:1 (1993). 126. Shibata, T., Kawanishi, M., Nagahara, J., Inoue, A., and Masumoto, T., Mater. Sci. Eng. A 179/180:632 (1994). 127. Inoue, A., Bulk Amorphous Alloys, Materials Science Foundations, Vol. 4, Trans Tech Publ., Zürich (1998). 128. Inoue, A., and Masumoto, T., Mater. Sci. Eng. A 133:6 (1991). 129. Eckert, J., Schlorke, N., and Schultz, L., in: Synthesis/Processing of Lightweight Materials II, (C.M. Ward-Close, F.H. Froes, D.J. Chellman and S.S. Cho, eds.), pp. 383, The Minerals, Metals and Materials Society, Warrendale, PA (1997). 130. Liu, W., and Johnson, W.L., J. Mater. Res. 11:2388 (1996). 131. Inoue, A., Nakamura, T., Nishiyama, N., and Masumoto, T., Mater. Trans. JIM 33:937 (1992). 132. Inoue, A., Nishiyama, N., Kim, S.G., and Masumoto, T., Mater. Trans. JIM 33:360 (1992). 133. Kato, A., Suganuma, T., Horikiri, H., Kawamura, Y., Inoue, A., and Masumoto, T., Mater. Sci. Eng. A 179/180:112 (1994). 134. Inoue, A., Horikiri, H., Kato, A., and Masumoto, T., Mater. Trans. JIM 35:79 (1994). 135. Uoya, A., Shibata, T., Higashi, K., Inoue, A., and Masumoto, T., J. Mater. Res. 11:2731 (1996). 136. Inoue, A., Zhang, T., and Masumoto, T., Mater. Trans. JIM 31:425 (1990). 137. Inoue, A., Zhang, T., Nishiyama, N., Ohba, K., and Masumoto, T., Mater. Trans. JIM 34:1234 (1993). 138. Lin, X.H., Johnson, W.L., and Rhim, W.K., Mater. Trans. JIM 38:473 (1997). 139. Xing, L.Q., Eckert, J., Löser, W., and Schultz, L., Appl. Phys. Lett. 73:2110 (1999). 140. Gebert, A., Eckert, J., and Schultz, L., Acta Mater. 46:5475 (1998). 141. Kawamura, Y., Kato, H., Inoue, A., and Masumoto, T., Appl. Phys. Lett. 67:2008 (1995). 142. Kato, H., Kawamura, Y., and Inoue, A., Mater. Trans. JIM 37:70 (1996). 143. Kawamura, Y., Kato, H., and Inoue, A., Advanced Mater. 4:4861 (1998). 144. Eckert, J., Mattern, N., Zinkevitch, M., and Seidel, M., Mater. Trans. JIM 39:623 (1998). 145. Köster, U., Meinhardt, J., Roos, S., and Liebertz, H., Appl. Phys. Lett. 69:179 (1996). 146. Frankwicz, P.S., Ram, S., and Fecht, H.-J., Mater. Lett. 28:77 (1996). 147. Chen, M.W., Inoue, A., Sakurai, T., Ping, D.H., and Hono, K., Appl. Phys. Lett. 74:812 (1999). 148. Fan, C., and Inoue, A., Mater. Trans. JIM 38:1040 (1997). 149. Fan, C., Takeuchi, A., and Inoue, A., Mater. Trans. JIM 40:42 (1999).

13: Structure Formation and Mechanical Behavior, Eckert 671 150. Eckert, J., Mattern, N., Seidel, M., and Schultz, L., in: Structure and Dynamics of Glasses and Glass Formers, (A. Angell, K.L. Ngai, J. Kieffer, T. Egami, and G.U. Nienhaus, eds.), pp. 465, Mater. Res. Soc. Symp. Proc., Vol. 455, Materials Research Society, Warrendale, PA (1997). 151. Mattern, N., Roth, S., Bauer, H.-D., Henninger, G., and Eckert, J., Mater. Sci. Eng. A, 304–306:311 (2001). 152. Xing, L.Q., Eckert, J., Löser, W., Schultz, L., and Herlach, D.M., Philos. Mag. A 79:1095 (1999). 153. Fan, C., Louzguine, D.V., Li, C., and Inoue, A., Appl. Phys. Lett. 75:340 (1999). 154. Inoue, A., Zhang, T., Saida, J., Matsushita, M., Chen, M.W., and Sakurai, T., Mater. Trans. JIM 40:1137 (1999). 155. Inoue, A., Zhang, T., Saida, J., Matsushita, M., Chen, M.W., and Sakurai, T., Mater. Trans. JIM 40:1181 (1999). 156. Schurack, F., Eckert, J., and Schultz, L., Mater. Sci. Eng. A, 294–296:164 (2000). 157. Haas, V., Cho, M.I., Ishii, H., and Inoue, A., in: Advances in Nanocrystallization, (A.R. Yavari, ed.), pp. 203, Mater. Sci. Forum, Vol. 307, Trans Tech Publ., Zürich (1997); J. Metastable and Nanocrystalline Materials 1:203 (1999). 158. Haas, V., Cho, M.I., Ishii, H., and Inoue, A., Nanostructured Mater. 12:829 (1999). 159. Büchler, E.H., Watanabe, E., and Kazama, N.S., Int. J. Non-Equilibrium Processing 10:35 (1997). 160. Inoue, A., Kato, A., Zhang, T., Kim, S.G., and Masumoto, T., Mater. Trans. JIM 32:609 (1991). 161. Leonhard, A., Xing, L.Q., Heilmaier, M., Gebert, A., Eckert, J., and Schultz, L., Nanostructured Mater. 10:805 (1998). 162. Xing, L.Q., Eckert, J., Löser, W., and Schultz, L. Appl. Phys. Lett. 74:664 (1999). 163. Chen, M.W., Zhang, T., Inoue, A., Sakai, A., and Sakurai, T., Appl. Phys. Lett. 75:1697 (1999). 164. Saida, J., Matsushita, M., Zhang, T., Inoue, A., and Sakurai, T., Appl. Phys. Lett. 75:3497 (1999). 165. Fan, C., Ott R.T., and Hufnagel, T.C. Appl. Phys. Lett. 81:1 (2002). 166. Hirano, T., Kato, H., Matsuo, A., Kawamura, Y., and Inoue, A. Mater. Trans. JIM 41:1454 (2000). 167. Hays, C.C., Kim, C.P., and Johnson, W.L., Phys. Rev. Lett. 84:2901 (2000). 168. Szuecs, F., Kim, C.P., and Johnson, W.L., Acta. Mater. 49:1507 (2001). 169. Kühn, U., Eckert, J., Mattern, N., and Schultz, L. Appl. Phys. Lett. 80:2478 (2002). 170. Eckert, J., Kühn, U., Mattern, N., He, G., and Gebert, A., Intermetallics 10:1183 (2002). 171. Pekarskaya, E., Kim, C.P., and Johnson, W.L. J. Mater. Res. 16:2513 (2001). 172. Zhang, T., and Inoue, A., Mater. Trans. JIM 39:1001 (1998). 173. Zhang, T., and Inoue, A., Mater. Trans. JIM 40:301 (1999). 174. Kim, Y.C., Kim, W.T., and Kim, D.H., Mater. Trans. JIM 43:1243 (2002). 175. Kim, Y.C., Kim, W.T., and Kim, D.H., Mater. Sci. Eng. A 375:127 (2004).

672

Properties

176. Hays, C.C., and Glade, S.C., Appl. Phys. Lett. 80:3096 (2002). 177. Peker, A., and Johnson, W.L., Appl. Phys. Lett. 63:2342 (1993). 178. He, G., Löser, W., Eckert, J., and Schultz, L., J. Mater. Res. 17:3015 (2002). 179. Kato, H., and Inoue, A., J. Alloy Comp. 375:171 (2004). 180. Dai, Q.L., Sun, B.B., Sui, M.L., He, G., Li, Y., Eckert, J., Luo, W.K., and Ma, E., J. Mater. Res. 19:2557 (2004). 181. Massalski, T.B., Okamoto, H., Subramanian, P.R., and Kasprzak, L., Binary Alloy Phase Diagrams, 2nd edn, ASM International, Ohio, USA (1992). 182. He, G., Löser, W., and Eckert, J., Acta Mater. 51:5223 (2003). 183. Eckert, J., He, G., Das, J., and Löser, W., Mater. Trans. JIM 44:1999 (2003). 184. He, G., Eckert, J., Löser, W., and Hagiwara, M., Acta Mater. 52:3035 (2004). 185. He, G., Eckert, J., and Hagiwara, M., Metall. Mater. Trans. A 35:3605 (2004). 186. He, G., and Hagiwara, M., Mater. Trans. JIM 45:1555 (2004). 187. He, G., and Hagiwara, M., Mater. Trans JIM 45:1120 (2004). 188. Long, M., and Rack, H.J., Biomater. 19:1621 (1998). 189. Niinomi, M., Metall. Mater. Trans. A 33:477 (2002). 190. Rho, J.Y., Roy, M.E., Tsui, T.Y., and Pharr, G.M., J. Biomed. Mater. Res. 45:48 (1999). 191. Niinomi, M., Biomater. 24:2673 (2003). 192. Sagel, A., Wunderlich, R.K., Perepezko, J.H., and Fecht, H.-J., Appl. Phys. Lett. 70:580 (1997). 193. Eckert, J., Mater. Sci. Eng. A 226–228:364 (1997). 194. Schlorke, N., Eckert, J., and Schultz, L., Mater. Sci. Eng. A 226–228:425 (1997). 195. Eckert, J., Seidel, M., Kübler, A., Klement, U., and Schultz, L., Scr. Metal. Mater. 38:595 (1998). 196. Benameur, T., Inoue, A., and Masumoto, T., Mater. Trans. JIM 35: 451 (1994). 197. Benameur, T., Inoue, A., and Masumoto, T., Nanostructured Mater. 4:303 (1994). 198. Seidel, M., Eckert, J., Bauer, H.-D., and Schultz, L., in: Grain Size and Mechanical Properties–Fundamentals and Applications, (M.A. Otooni, R.W. Armstrong, N.J. Grant and K. Ishizaki, (eds.), pp. 239, Mater. Res. Soc. Symp. Proc., Vol. 362, Materials Research Society, Warrendale, PA (1995). 199. Kawamura, Y., Inoue, A., Sasamori, K., Katoh, A., and Masumoto, T., Mater. Trans. JIM 34:969 (1993). 200. Tsai, A.P., Aoki, K., Inoue, A., and Masumoto, T., J. Mater. Res. 8:5 (1993). 201. Mizutani, U., Takeuchi, T., and Fukunaga, T., Mater. Trans. JIM 34:102 (1993). 202. Schurack, F., and Eckert, J., unpublished results. 203. Asahi, N., Mater. Sci. Eng. A 226–228:67 (1997). 204. Schlorke, N., Eckert, J., and Schultz, L., in: Mechanically Alloyed, Metastable and Nanocrystalline Materials, (M.D. Baró and S. Suriñach, eds.), pp. 761, Mater. Sci. Forum, Vol. 269–272, Trans Tech Publ., Zürich (1998).

13: Structure Formation and Mechanical Behavior, Eckert 673 205. Schlorke, N., Eckert, J., and Schultz, L., in: Magnesium Alloys and Their Applications, (B.L. Mordike and K.U. Kainer, eds.), pp. 533, DGM Informationsgesellschaft, Frankfurt (1998). 206. Inoue, A., Kohinata, M., Tsai, A.P., and Masumoto, T., Mater. Trans. JIM 30:378 (1990). 207. Eckert, J., Schlorke, N., and Schultz, L., in: Proc. 1998 Powder Metallurgy World Congress, Granada, Vol. 1, pp. 311, European Powder Metallurgy Association, Shrewsbury (1998). 208. Eckert, J., Schlorke-de Boer, N., Weiß, B., and Schultz, L., Z. Metallkde. 90:908 (1999). 209. Seidel, M., Ph.D. Thesis, University of Technology Dresden (1998). 210. Kuhnast, F.A., Held, O., Regnier, F., and Illeková, E., Mater. Sci. Eng. A 226–228:463 (1997). 211. Eckert, J., Seidel, M., Schlorke, N., Kübler, A., and Schultz, L., in: Synthesis and Properties of Mechanically Alloyed and Nanocrystalline Materials, (D. Fiorani and M. Magini, eds.), pp. 113, Mater. Sci. Forum, Vol. 235–238, Trans Tech Publ., Zürich (1997). 212. Lin, X.H., and Johnson, W.L., J. Appl. Phys. 78:6514 (1995). 213. Eckert, J., Kübler, A., and Schultz, L., J. Appl. Phys. 85:7112 (1999). 214. Eckert, J., L., in: Metastable, Mechanically Alloyed and Nanocrystalline Materials, (A. Calka and D. Wexler, eds.), pp. 3, Mater. Sci. Forum, Vol. 312–314, Trans Tech Publ., Zürich (1999); J. Metastable and Nanocrystalline Materials 2–6:3 (1999). 215. Kübler, A., Gebert, A., Eckert, J., and Schultz, L., J. Appl. Phys. 83:3438 (1998). 216. Inoue, A., Mater. Sci. Eng. A179/180:57 (1994). 217. Kübler, A., Eckert, J., and Schultz, L., Nanostructured Mater. 12:443 (1999). 218. Moelle, C., Lu, I.-R., Sagel, A., Wunderlich, R.K., Perepezko, J.H., and Fecht, H.-J., in: Mechanically Alloyed, Metastable and Nanocrystalline Materials, (M.D. Baró and S. Suriñach, eds.), pp. 47, Mater. Sci. Forum, Vol. 269–272, Trans Tech Publ., Zürich (1998). 219. Zhou, C.R., and Xu, J., J. Non-Cryst. Solids 297:131 (2002). 220. Coble, R.L., J. Appl. Phys. 34:1679 (1963). 221. Greer, A.L., Zhong, Z.C., Jiang, X.Y. Rutherford, K.L., and Hutchings, I.M., in: Chemistry and Physics of Nanostructures and Related Non-Equilibrium Materials, (E. Ma, B. Fultz, R. Shull, J. Morral and P. Nash, eds.), pp. 3, The Mineral, Metals and Materials Society, Warrendale, PA (1997). 222. Inoue, A., Nakazato, K., Kawamura, Y., Tsai, A.P., and Masumoto, T., Mater. Trans. JIM 35:95 (1994). 223. Chalmers, B., Principles of Solidification, pp. 64, John Wiley and Sons, New York, NY (1964). 224. Kim, H.S., Suryanarayana, C., Kim, S.J., and Chun, B.S., Powder Metall, in press 225. Greer, A.L., Curr. Opin. Solid State & Mater. Sci. 2:412 (1997). 226. He, Y., Dougherty, G.M., Shiflet, G.J., and Poon, S.J., Acta Metall. Mater. 41:337 (1993). 227. Inoue, A., Curr. Opin. Solid State & Mater. Sci. 2:305 (1997).

674

Properties

228. Yokoyama, Y., Inoue, A., and Masumoto, T., Mater. Trans. JIM 34:135 (1993). 229. Takeuchi, S., in: Quasicrystals, (J.M. Dubois, P.A. Thiel, A.P. Tsai and K. Urban, eds.), pp. 283, Mater. Res. Soc. Symp. Proc., Vol. 553, Materials Research Society, Warrendale, PA (1999). 230. Wittmann, R., Urban, K., Schandl, M., and Hornbogen, E., J. Mater. Res. 6, 1165 (1991). 231. Feuerbacher, M., Klein, H., Schall, P., Bartsch, M., Messerschmidt, U., and Urban, K., in Quasicrystals, (J.M. Dubois, P.A. Thiel, A.P. Tsai and K. Urban, eds.), pp. 307, Mater. Res. Soc. Symp. Proc., Vol. 553, Materials Research Society, Warrendale, PA (1999). 232. Feuerbacher, M., Metzmacher, C., Wollgarten, M., Urban, K., Baufeld, B., Bartsch, M., and Messerschmidt, U., Mater. Sci. Eng. A 233:103 (1997). 233. Sordelet, D.J., Kim, J.S., and Bessier, M.F., in: Quasicrystals, (J.M. Dubois, P.A. Thiel, A.P. Tsai and K. Urban, eds.), pp. 459, Mater. Res. Soc. Symp. Proc., Vol. 553, Materials Research Society, Warrendale, PA (1999). 234. Sánchez, A., Garcia de Blas, F.J., Algaba, J.M., Alvarez, J., Valles, GarciaPoggio, M.C., and Agüero, A., in: Quasicrystals, (J.M. Dubois, P.A. Thiel, A.P. Tsai and K. Urban, eds.), pp. 447, Mater. Res. Soc. Symp. Proc., Vol. 553, Materials Research Society, Warrendale, PA (1999). 235. Inoue, A., Horio, Y., Kim, Y.H., and Masumoto, T., Mater. Trans. JIM 32:669 (1992). 236. Unsworth, W., Mater. Product. Technol. 4:359 (1989). 237. Gjsetland, H., Light Metal Age 47(7):19 (1989). 238. Gilbert, C., Ritchie, R.O., and Johnson, W.L., Appl. Phys. Lett. 71:476 (1997). 239. Kawamura, Y., Kato, H., Inoue, A., and Masumoto, T., Appl. Phys. Lett. 67:2008 (1995). 240. Kawamura, Y., Shibata, T., Inoue, A., and Masumoto, T., Appl. Phys. Lett. 69:1208 (1996). 241. Inoue, A., Zhang, T., and Masumoto, T., Mater. Trans. JIM 36:391 (1995). 242. Bruck, H.A., Christman, T., and Johnson, W.L., Scr. Metall. 30:429 (1994). 243. Xing, L.Q., Eckert, J., and Schultz, L., Nanostructured Mater. 12:503 (1999). 244. Xing, L.Q., Bertrand, C., Dallas, J.-P., and Cornet, M., Mater. Sci. Eng. A 241:216 (1998). 245. Spaepen, F., Acta Metall. 25:407 (1977). 246. Kim, Y.H., Inoue, A., and Masumoto, T., Mater. Trans. JIM 31:747 (1990). 247. Kato, H., and Inoue, A., Mater. Trans. JIM 38:793 (1997). 248. Inoue, A., Zhang, T., and Kim, Y.H., Mater. Trans. JIM 38:749 (1997). 249. Leonhard, A., Heilmaier, M., Eckert, J., and Schultz, L., in: Creep Behavior of Advanced Materials for the 21st Century, (R.S. Mishra, A.K. Mukherjee and K. Linga Murty, eds.), pp. 41, The Minerals, Metals and Materials Society, Warrendale, PA (1999). 250. De Hey, P., Sietsma, J., and Van den Beukel, A., Acta Mater. 46:5873 (1998). 251. Kawamura, Y., Shibata, T., Inoue, A., and Masumoto, T., Mater. Trans. JIM 40:335 (1999). 252. Reger-Leonhard, A., Heilmaier, M., and Eckert, J., unpublished results.

13: Structure Formation and Mechanical Behavior, Eckert 675 253. 254. 255. 256. 257. 258. 259. 260. 261. 262. 263. 264. 265. 266. 267. 268. 269. 270. 271. 272. 273. 274. 275.

276. 277. 278. 279. 280.

281. 282.

Taub, A.I., Acta Metall. 28:633 (1980). Taub, A.I., and Luborsky, F.E., Acta Metall. 29:1939 (1981). Jia, D., Ramesh, K.T., and Ma, E., Scripta Mater. 42:73 (1999). Wei, Q., Jia, D., Ramesh, K.T., and Ma, E., Appl Phys Lett 81:1240 (2002). Leng, Y., and Courtney, T.H., Metall Trans A 21:2159 (1990). Wang, Y., Chen, M., Zhou, F., and Ma, E., Nature 419:912 (2002). Eckert, J., Kühn, U., Mattern, N., He, G., and Gebert, A., Intermetallics 10:1183 (2002). Das, J., Löser, W., Kühn, U., Eckert, J., Roy, S.K., and Schultz, L., Appl Phys Lett 82:4690 (2003). He, G., Eckert, J., Löser, W., and Schultz, L., Nature Mater. 2:33 (2003). Das, J., Güth, A., Klauß, H.J., Mickel, C., Löser, W., Eckert, J., Roy, S.K., and Schultz, L., Scripta Mater. 49:1189 (2003). Das, J., Roy, S.K., Löser, W., Eckert, J., and Schultz, L., Mater. Manufact. Process 19:423 (2004). Löser, W., Das, J., Güth, A., Klauß, H.J., Mickel, C., Kühn, U., Eckert, J., Roy, S.K., and Schultz, L., Intermetallics 12:1153 (2004). Das, J., M.Tech. Thesis, Indian Institute of Technology, Kharagpur, India, April (2003). Jeong, H.W., Kim, S.E., Hyun, Y.T., Lee, Y.T., and Park, J.K., Mater. Sci. Forum 475–479:2291 (2005). Eckert, J., Das, J., He, G., Calin, M., and Kim, K.B., Mater. Sci. Eng. A in press (2006). He, G., Eckert, J., Dai, Q.L., Sui, M.L., Löser, W., Hagiwara, M., and Ma, E., Biomater. 24:5115 (2003). Bakke, E., Busch, R., and Johnson, W.L., Appl. Phys. Lett. 67:3260 (1995). Einstein, A., Ann. Physik (Leipzig) 34:592 (1911). Waniuk, T.A., Busch, R., Masuhr, A., and Johnson, W.L., Acta Mater. 46:5229 (1998). Taub, A.I., and Spaepen, F., Acta Metall. 28:1781 (1980). Grest, G.S., and Cohen, M.H., Adv. Chem. Phys. 48:455 (1981). Tsao, S.S., and Spaepen, F., Acta Metall. 33:881 (1985). Wilde, G., Görler, G.P., Jeropoulos, K., Willnecker, R., and Fecht, H.-J., in: Mechanically Alloyed, Metastable and Nanocrystalline Materials, (M.D. Baró and S. Suriñach, eds.), pp. 541, Mater. Sci. Forum, Vol. 269–272, Trans Tech Publ., Zürich (1998). Angell, C. A., Science 267:1924 (1995). Kübler, A., Eckert, J., Gebert, A., and Schultz, L., J. Appl. Phys. 83:3438 (1998). Weiss, M., Moske, M., and Samwer, K., Appl. Phys. Lett. 69:3200 (1996). Zappel, J., and Sommer, F., J. Non-Cryst. Solids 205–207:494 (1996). Busch, R., Bakke, E., and Johnson, W. L., in: Synthesis and Properties of Mechanically Alloyed and Nanocrystalline Materials, (D. Fiorani and M. Magini, eds.), pp. 327, Mater. Sci. Forum, Vol. 235–238, Trans Tech Publ., Zürich (1997). Inoue, A., Zhang, T., and Kim, Y.H., Mater. Trans. JIM 38:749 (1997). Ke, M., Hackney, S.S., Milligan, W.W., and Aifantis, E.C., Nanostructured Mater. 5:689 (1995).

14 Nanostructured Electronics and Optoelectronic Materials Raphael Tsu Department of Electrical and Computer Engineering, University of North Carolina, Charlotte, USA Qi Zhang Chief Scientist, Advanced Photonix, Inc., 305 County Road YZ, Dodgeville, USA

14.1 Introduction The availability of highly controlled deposition techniques and methods for nanometer-scale lithographic and device fabrications stimulated the growth of interest in the study of quantum size effects in the electronic and optical properties of nanostructures, ranging from one-dimensionally confined superlattices to three-dimensionally confined quantum dots. In what follows, rather than covering a wide range of topics, we focus on semiconductors, more precisely, silicon. A fairly in-depth treatment of several topics is included, such as the dielectric constant, the capacitance, doping and exciton binding energies of a nanoparticle, as well as several examples of applications such as porous silicon, nanoscale silicon particles embedded in an oxide matrix, a superlattice of amorphous silicon sandwiched between thin oxides and a superlattice of crystalline silicon sandwiched between adsorbed oxygen monolayers, as well as some quantum devices. The first part deals with the physics of nanostructured materials, and the second part deals with possible devices such as light emitting diodes (LEDs) and quantum field effect transistors (QD-FETs).

14.2 Physics of Nanostructured Materials 14.2.1 Quantum Confinement: Superlattices and Quantum Wells Superlattices and quantum wells were introduced as man-made quantum structures to engineer the quantum states for electrical and Carl C. Koch (ed.), Nanostructured Materials: Processing, Properties, and Applications, 2nd Ed., 677–718 © 2007 William Andrew, Inc.

677

678

Properties

optical applications.[1,2] In twenty-five or more years, thousands of papers have been published. By 1997, more than 465 patents had been awarded on topics relating to the application of microelectronic and optoelectronic devices and techniques for producing superlattice materials. In retrospect, the idea relies heavily on the availability of good heterojunctions, lattice matched systems,[3] and later, the strained layered systems.[4] It is appropriate to briefly discuss what is needed for the realization of a man-made quantum system. To realize quantum states in a given geometry, the size must be smaller or comparable to the coherence length of electrons, in order to exhibit quantum interference. This requirement eliminates doping as an effective means to achieve confinement, except at low temperatures,[5] because doping comes from charge separation which results in barriers generally far exceeding the coherence length of electrons at room temperatures. On the other hand, band-edge alignment of a heterojunction provides abrupt barrier height. This short-range potential is the consequence of higher order multiples in the atomic potentials. A new type of superlattice was proposed, the Epilayer Doping Superlattice (EDS), consisting of, for example, a couple of layers of Si in AlP.[6] The idea is fundamentally different from atomic plane-doped or δ-doped superlattices[7] where only a small fraction of a plane is occupied by doping or substitution. Another type of superlattice designed to incorporate extremely localized interaction, most promising for silicon, was introduced in 1993,[8] consisting of an effective barrier for silicon, formed by a suboxide with a couple of monolayers of oxygen atoms. This system, as a barrier for silicon, has been experimentally realized.[9] Localized interaction in a man-made quantum system is not new; for example, resonant tunneling involving localized defects has been reported.[10] Recently, superlattices with extremely localized interaction in what we call the Hetero-Epilattice Superlattice (HES) has been successfully fabricated by sandwiching thin silicon epitaxial layers between monolayers of adsorbed oxygen.[11] What is most remarkable is the lack of stacking fault defects. A variation of this HES shows electroluminescence.[12] The HES, subsequently changed to SAS, Semiconductor-atomic-superlattice,[13] can finally promote silicon to join the ranks of devices with quantum effects, thus far almost totally dominated by III–V semiconductors, except in the Si/Ge case where carrier confinement is in the germanium.[14]

14.2.2 Dielectric Constant of Nanoscale Silicon As the physical size approaches several nanometers, reduction in the static dielectric constant, ε, becomes significant. Basically, ε measures

14: Nanostructured Electronics, Tsu and Zhang

679

screening which is not effective in a nanoparticle. Specifically, quantum confinement increases the separation of the energy states causing a lowering of the transitions, which sum to the value for ε. A modified Penn model, taking into account the quantum confinement induced discrete energy states, was applied to a sphere and to a wire. The calculated sizedependent ε is consistent with the wave vector dependent ε(q).[15] This work was initially motivated by the need to understand the self-limiting mechanism in the electrochemically etched porous silicon. The argument goes as follows: A significant reduction of ε results in a large increase in the binding energy of dopants rendering the disappearance of extrinsic dopings.[16] Reference 16 was distributed during the Materials Research Society (MRS) meeting in 1992 which led to the more involved numerical calculations of Wang and Zunger[17] and Lannoo et al.[18] What followed was somewhat familiar. A more complete version of Ref. 16 was rejected by the reviewer who insisted that Rayleigh scattering should be included! We had to rewrite our manuscript and resubmit the results in the present form of Ref.15. Strictly speaking, the dielectric function is only definable in an unbounded region of space. The wave-vector dependent, ε(q), has been derived for semiconductors such as Si, Ge and GaAs.[19] The use of this ε(q) is essential for the calculation of screened shallow impurity potentials. Replacing the free electron momentum by knp = αnp/a with αnp being the nth roots of the spherical Bessel function, jp(ka) = 0, we obtain the size-dependent dielectric constant ε(a) ε(a) = 1 + (εb − 1)/[1 + (ΔE/Eg)2]

Eq. (14-1)

in which εb is the bulk dielectric constant, Eg is the energy of the center of the imaginary part of the dielectric function, and ΔE is given by πEF/kF a. Taking the values for silicon, εb = 12, Eg = 4 eV, and filling the energy bands up to EF of 4 × 5 × 1022 valence electrons per cm3, or EF = 12.6 eV, the computed ε(a) is shown in Figure 14.1. Note that our computed results when using εb = 11.3 agree almost perfectly with that of ε(q) from Ref. 19 by putting q = 2π/2a (from Fourier transform), and agree well with those of Refs 17 and 18. This simple treatment gives a good physical insight, apart from the fact that the approach is far simpler to incorporate into other calculations, for example, the binding energy of a dopant in a nanoparticle, which we shall treat next. Before we leave this subject, we point out that the size-dependent dielectric constant of a quantum wire using this model gives the same expression with the spherical Bessel function replaced by Bessel function,

680

Properties

Figure 14.1 Plot of size-dependent static dielectric constant ε(a) vs. the radius a of silicon sphere in angstroms. The solid line is from the modified Penn model with εb = 12; the dashed-dot line is from the same model with εb = 11.3; the crosses are from Ref. 18; the longer dashes are from Ref. 17; and the shorter dashes are from Ref. 16.

as long as we use an isotropic effective mass. This size-dependent dielectric constant should be important in other physical situations, such as electron–phonon interaction, polaritons, optical properties and especially applying to doping of shallow impurities.

14.2.3 Doping of a Nanoparticle The doping of a quantum dot is an important issue. Since electrochemically etched porous silicon exhibits quantum confinement in photoluminescence,[20] the quantum size effect on doping, including interactions with induced charges at the dielectric discontinuity, requires investigation. Fundamentally, quantum confinement pushes up the allowed energies resulting in an increase in the binding energy, Eb, of shallow impurities such as the cases of quantum wells[21] and superlattices.[22] Theoretical treatments of the dielectric constant in quantum confined systems[23,24] show that a significant reduction takes place when the width of the quantum well is below 2 nm. Qualitatively, quantum confinement cuts down the motion of electrons, resulting in a reduction in screening. Using the Bohr model for shallow dopants, the binding energy is inversely proportional to the square of the dielectric constant, a reduction in the static dielectric constant greatly increases the binding energy to the extent that

14: Nanostructured Electronics, Tsu and Zhang

681

Figure 14.2 Donor binding energy vs. dot radius in angstroms for several values of the dielectric constant of the matrix.

most nanoparticles show no extrinsic doping. In a quantum dot of radius a, the measured ε(a) for porous silicon[16] was in fair agreement with the calculation given by Eq. (14-1) in Section 14.2.2. However, preliminary calculated binding energy[25] for dopants points out that this reduction in ε plays a small role in the final results because the larger increase is due to the induced polarization charges at the boundary of the dielectric discontinuity.[26] With ε1 and ε2 denoting the dielectric constants of the particle and the matrix, for ε1 > ε2, the induced charge of the donor is of the same sign resulting in an attractive interaction with the electron of the dot, pushing deeper the ground state energy of the donor resulting in an appreciable increase in Eb. For ε1 < ε2, the opposite is true; Eb is much reduced allowing possible extrinsic conductivity at room temperatures. Figure 14.2 shows the donor binding energy versus several values of the dielectric constant for the matrix, 1 for air or vacuum, 6 for water within the Helmholtz layer, etc. Note that at a dot radius of 2 nm, the former gives Eb = 0.8 eV, while the latter gives 0.2 eV, making it possible to show extrinsic doping at room temperatures.[27] The lack of extrinsic doping as the particle size is reduced by electrochemical etching serves as a limiting factor on the size reduction in etching in the dark. If etching is performed in the presence of light, electron–hole generation can lead to continuous etching without limitation.[27] In electroluminescent (EL) diodes, it should be important to match the dielectric constants to facilitate extrinsic doping. In fact, this may be the reason why in porous silicon (Psi) and EL

682

Properties

devices, the use of the SiC/PSi/Si pn junction[28] seems to work well; because the dielectric constant of SiC matches that of PSi, allowing the formation of a pn-junction. Extrinsic doping forms the backbone of all solid state devices with pnjunctions. We have shown that almost all shallow levels become deep in nanoscale particles, and induced charges at the dielectric interface between the quantum dot and its matrix affect, to a large degree, the binding energy of the dopants. Therefore, nanostructured materials in optoelectronic applications require more thought.

14.2.4 Excitonic Binding and Recombination Energies Electrochemically etched porous silicon displays visible luminescence.[20] The role of quantum confinement in the porous silicon luminescence is established by the increase of the optical absorption gap,[29] and by the decrease of the Raman phonon frequency with the increase of the peak luminescence energy.[30] The quantum confinement effects in silicon nanocrystallites have been treated by tight-binding, effective mass, pseudopotential and first-principles local density approximations. However, in order to take into account the induced electrostatic polarization due to dielectric mismatch at the silicon crystallite boundary with the external medium, recombination and binding energies of excitons in silicon quantum dots may be calculated within essentially the same framework as the calculation of the binding energy of a quantum dot treated in Section 14.2.3.[31] The peak recombination energy in silicon quantum dots is quite insensitive to the nature of the external medium due to approximate cancellation of the polarization terms in the recombination energy of the excitons involved. No such cancellation is present for the binding energy of the excitons. Excitons in silicon crystallites surrounded by vacuum are electrostatically bound by about 1 eV. When immersed in water, the binding energy is dramatically decreased compared to vacuum as the doping case considered in the last section. Recombination and binding energies of the excitons confined in silicon nanocrystallites are calculated within the effective mass approximation. This approximation has been already applied to evaluate the one- and twoelectron ground state energies,[32] donor binding energy,[27] excitonic energy,[34] and absorption coefficient.[35] The envelope wave functions of both electron and hole are determined by the kinetic energy of each particle, which dominates the properties of excitons in quantum dots in the

14: Nanostructured Electronics, Tsu and Zhang

683

Figure 14.3 Exciton binding energy, solid, and recombination energy, dashed, vs. dot radius in nm for several values of the dielectric constants of the matrix.

size the range of several nanometers. The electrostatic terms, Coulomb interaction, polarization interaction and electron–hole polarization selfenergies are treated by perturbation, in a variational calculation. As before,[27] the binding energy is defined as the minimum energy required to break an exciton by removing both electron and hole from the same silicon quantum dot and placing them in separate silicon nanocrystals. The modified Penn model is used for the dielectric constant of the dot, see Section 14.2.1. Figure 14.3 shows the calculated values of the exciton binding energy, solid; and the recombination energy, dashed; vs dot radius for various ε values of the matrix. Note that all curves for the recombination energy are close to each other, due to cancellation effects discussed previously. Note that, for a radius of 1.5 nm, the exciton binding energy ranges from 82 meV (ε2 = 80) to 1.03 eV (ε2 = 1), a change of more than an order of magnitude. For the case of the same dielectric constants of the dot and the matrix, where only the Coulomb interaction is present, the exciton binding energy is 0.16 eV, over ten times higher than the bulk value of 14.7 meV. This major increase of the Coulomb interaction part of the exciton binding energy is caused by the increased overlap of the electron–hole wave functions. For all four values of the dielectric constant of the external matrix, the exciton binding energy is much greater than the characteristic thermal energy at room temperature, therefore, excitons confined to a quantum dot are well bound and stable irrespective of the surrounding matrix. Since the self-polarization and the polarization terms

684

Properties

are very large, any theory not including the dielectric mismatch between the dot and the environment cannot be taken seriously. We need to have a strong appreciation in the magnitude of the exciton binding energy for a quantum dot with a radius below 15 nm. Excitons simply cannot be broken in the usual sense! The activation energy of the break-up of excitons, estimated from the slope of the luminescence decay in air with increasing temperature, is approximately 100–120 meV,[35] which is far less than the exciton electrostatic binding energy (1 eV in vacuum), and is thus related to the turning-on of some non-radiative recombination channels. The photoluminescence in porous silicon at room temperature is due to recombination of excitons confined in silicon nanocrystals whose effective diameters are approximately 3 nm for nanowires and 3 nm for nanodots. The transition is still phonon assisted as in the bulk that involves an electron from the bottom of the conduction band and a hole from the top of the valence band, thus separated in the momentum space. In a well passivated system without surface trap states, a non-radiative channel may involve electron tunneling out of the quantum dot.[36] It is instructive to compare exciton radiative recombination in silicon nanocrystallites and in bulk Si, both at room temperature. In the bulk, it is far likelier that the exciton will be broken by a phonon than to encounter the right phonon for phonon assistance. The exciton break up is facilitated by the quasi-continuum of available states in both valence and conduction bands. The electron and hole liberated from the exciton by a phonon, fly apart, thus disabling radiative recombination. The quasicontinuum of valence and conduction band states is modified into a discrete set of energy levels due to quantum confinement. Since the thermal phonons, without sufficient energy to break up these excitons, allow excitons enough time to wait for the right phonon with the necessary momentum, phonon-assisted radiative recombination occurs. In short, it was assumed that nanoscale particles allowed the relaxation of momentum conservation, or even suggested that the band structures may be made direct by nanostructuring. Now we understand the situation quite well: short of nanostructuring to a dimension of well under 1 nm, indirect band structure still dominates optical transitions. The apparent increase in the observed luminescent efficiency is due to the long-lived excitons due to quantum confinement. The direct gap in a silicon nanocrystallite is located at 2.9 eV.[37] Therefore, some of the observed weak and fast blue luminescence in non-oxidized Si may be from this component.

14: Nanostructured Electronics, Tsu and Zhang

685

14.2.5 Capacitance in a Nanoparticle The effects of charge accumulation in quantum confinement have been under intensive study involving quantum dots,[38] and energy states of a silicon nanoparticle.[39] It is possible to introduce as many electrons as one would like into a classical capacitor until breakdown occurs from the ever increasing voltage. The situation is quite different for a nanoscale capacitor where each electron introduced into a quantum capacitor must satisfy the Pauli exclusion principle. In particular, since it takes kinetic energy to localize an electron, as the dimension shrinks to nanometer regime, only a small number of electrons can be put into the system before its energy is higher than the barriers serving to confine them. The problem is quite complex even in a classical calculation since a Green’s function method must be used to calculate the potentials due to the presence of an electron, or electrons with all the interactions with induced charges on the boundaries of the dielectric mismatch. For a silicon sphere embedded in an oxide matrix, ε of silicon is 12 while that of the oxide is 4. We shall sketch the salient procedures used in the calculation.[32] For a sphere, the use of the electrostatic Green’s function allows the identification of the potential energy terms in both one- and two-electron cases. Even in the one-electron case, an electron interacts with its induced bound surface charge density, resulting in the self-polarization. Note that this term goes to zero without the dielectric discontinuity. In the two-electron case, there are four terms: self polarization terms for each electron, Coulomb interaction, and polarization interaction. The polarization term arises because one electron induces a bound surface charge density, which interacts with the second electron. There is one important consideration worthy of discussion. Evaluation of the electrostatic energy requires that one specifies electronic positions. For a single electron, the position in the center gives the minimum polarization energy. However, for two electrons, the classical ground-state energy is determined by the minimum of the competing repulsive components. Coulomb and polarization interactions push the electrons apart towards the well boundary, while the selfpolarization terms push the electrons away from the boundary, closer to each other. The separation of the two electrons are obtained by a minimization of the total electrostatic energy. Table 14.1 lists the calculated values for one electron in a sphere of radius a, Ec1, and for two electrons, Ec2, and Δc, the difference. For the quantum-mechanical calculation, the zero order spherical Bessel function is used for both the one-and two-electron cases for the

Properties

686 Table 14.1 Classically Calculated One- and Two-Electron Electrostatic Energies

a (nm) E1c (eV) Ec2 (eV) Δc (eV)

1 0.12 0.60 0.48

2 0.06 0.30 0.24

3 0.04 0.20 0.16

4 0.03 0.15 0.12

Table 14.2 One- and Two-Electron Ground State Energies From Quantum-Mechanics

a (nm) E1 (eV) E2 (eV) Δqm (eV) C(qm)/C(classical)

1 1.59 3.64 2.05 0.23

2 0.43 1.09 0.66 0.36

3 0.21 0.57 0.36 0.44

4 0.12 0.37 0.25 0.48

unperturbed ground state wave functions which apply to the Hamiltonian having only the kinetic energy term with infinite barrier boundary condition. All of the energies for the self-energy, the Coulomb energy and the polarization energy are computed using perturbation similar to the treatment of the helium ground state. Below a = 1 nm, the validity of the effective mass formulation is questionable, and beyond a = 4 nm, the simple ground state wave function used is not a good approximation (higher order spherical Bessel functions need to be used for the perturbation calculation). The calculated results are tabulated in Table 14.2 for the total energies of the one- and two- electron cases, with the capacitance defined by E2 − E1 = Δ = e2/ 2C. In Table 14.2, it is noted that the capacitance C(qm) for a spherical particle of radius a = 3 nm is less than half the value of the capacitance for the classical particle. Physically, it is not meaningful to consider only electrostatically what happens when adding or subtracting an electron from the confining system because, quantum-mechanically, electrons have kinetic energies occupying their states. The kinetic energy scales with a−2, while all the other terms scale with a−1, therefore, our C(qm)

14: Nanostructured Electronics, Tsu and Zhang

687

does not scale with a, unlike the calculated C(qm) for large systems where the kinetic term is negligibly small.[40,41] Apart from the larger size regime and the inclusion of more electrons, there is an important difference between the results presented here and those of Refs 40 and 41. It is important to include the dielectric mismatch between the quantum dot and the matrix,[32] it is a formidable task to even extend this treatment to three- or four electron cases. Our results should have important applications in nanoscale electronics, particularly in devices with few electrons. Qualitatively at temperatures when the coherent length of the electrons is exceeded by the size, quantum treatment is not necessary and may be incorrect for the capacitance. Classical formulation using the Poisson equation should be used. Recently we have extended the part with classical treatment in Ref. 32 to N > 2, for N electrons confined inside a dielectric sphere of e1 with e1 > e2, the dielectric constant of the matrix. Our formulation involves the minimization of the total interaction energy, polarization energy and the direct Coulomb repulsion between the between ith and the jth electrons and the self polarization energy of the ith electron. Because the configurations of the electrons are found to be unique for each electron added to the system, we modified the obvious extension of the case for the capacitance of two electrons, C2, with E1 and E2 to CN with EN and EN+1, by defining a new “mono-phasic” capacitance C(N) ≡ (Ne)2/EN

Eq. (14-2)

leading to a capacitance agreeing with that of a single electron as well as with N electrons for N large, approaching the metallic limit of the Gauss model. At first we thought that the configuration of N > 2 could follow the so-called Platonic solids with N = 4,6,8,12,20, because we discovered that symmetry plays the most important role. However, our preliminary results extending N to 12 showed that what is more important is the role of the shell model in the Periodic Table of elements. This totally new aspect was uncovered with detailed computation by LaFave beyond the work,[42] when extending N to 200, establishing that not only the shell model, but the whole Periodic Table itself shares detailed features with our classical Poisson equation approach to the problem of the N electrons in a dielectric sphere. We have shown in this example that the effect of the discreteness of the electronic charge, even in the classical domain, may be more important to nanoelectronics than quantum phenomena.

688

Properties

14.2.6 Structure, Bonds and Coordinations of Si Nanostructure: Porous Si and Si Clusters Structure Since the discovery of fairly efficient photoluminescence in the visible spectrum by Canham[20] in PSi, an electrochemically etched silicon, the debate is on as to the origin of the luminescence: quantum confinement,[29][43] a-SiH,[44] siloxene derivatives,[45] surface states,[46] etc. Our assessment is that, whenever careful researchers disagree, often several of these mechanisms are present. Which one dominates depends on special preparation resulting in differences in structures. More details and recent thoughts on this subject, as well as many other aspects of porous Si, may be found in the book cited,[47] and a review article.[48] Since structure, bonds and coordination are the starting points of any physical models, we shall deal with it from our point of view, not necessarily from a vantage point. More inclusive discussions are found in Refs 47 and 48. From the correlation of the Raman shift with the upshifts of the photoluminescence (PL) peak, it was concluded that the red luminescence originates from nanostructures in the order of 2–3 nm.[30] Subsequently, using high-resolution cross-sectional transmission electron microscopy (TEM),[49] silicon nanocrystallites of 3–5 nm were identified in typical porous silicon.

Bonds and Coordinations Freshly produced red, yellow and green emitting porous Si specimens have been studied by near edge and extended X-ray absorption fine structure (NEXAFS and EXAFS).[50] The emission peaks are at 690, 580 and 520 nm, which almost cover the full visible range. The correlation between the coordination numbers of the first, second and third Si neighbor shells, from Fourier transform fitting of EXAFS with both emission peak energies and optical bandgaps estimated by PLE (photoluminescence excitation), suggests that the nanostructures of the PSi are nanowires, rather than clusters of nanocrystallites, for porous silicon samples prepared with low resistive silicon wafer. Two types of quantum nanowire with one and oneplus a fractional dimensionality are proposed to interpret the correlation. Moreover, the order factors of the theoretical fit suggest that nanowires of the freshly produced PSi have crystalline cores. To better understand the role of the structure of porous silicon in quantum confinement, many methods, including soft X-ray absorption,[51] XAS,[52]

14: Nanostructured Electronics, Tsu and Zhang

689

TEM,[49] and EXAFS,[53][54] have been used. It was found that electrochemical etching of silicon into porous silicon produces different structures depending on the resistivity of the silicon wafer used. This phenomenon may be understood using the self-limiting model of electrochemical etching in the dark.[28] In this model, etching requires the holes to come to the surface. Highly extrinsically doped silicon allows this to take place, while highly resistive materials allow the normally very slow etching in HF to compete with this low etching process resulting in the break-up of a thicker wire into clusters. In brief, highly resistive wafers result in more cluster-like PSi, while lower resistivity wafers produce more wire-like PSi. This is a good example of how EXAFS can clearly distinguish the structures. The coordination number and distances of Si—Si shells from Ref. 50 are given in Tables 14.3 and 14.4. Figure 4 gives the dependence of dimensionality and size with PSi “color.” See figure caption for details. Although more than one mechanism is likely to be involved in the luminescence of PSi, EXAFS and theoretical fits show a relatively simple picture for freshly produced PSi with PL peaks from 690 nm to 520 nm, which is nearly the full range we can obtain using only anodization processes. The nanocores of the PSi are crystalline and quantum confinement is the only mechanism widening the bandgap at this stage. A nanowire network (one-plus a fractional dimensionality) for red and yellow PSi, and

Table 14.3 The Coordination Numbers of Si—Si Shells

red PSi yellow PSi green PSi c-Si a-Si

1st NSi—Si

2nd NSi—Si

3.80 ± 0.15 3.65 ± 0.13 3.0 ± 0.2 4.0 ± 0.1 4.0 ± 0.1

7.42 ± 0.44 5.74 ± 0.38 3.01 ± 0.49 12.00 ± 0.35 0

3rd NSi—Si 8.15 ± 6.47 ± 3.98 ± 12.00 ± 0

Table 14.4 The Distances of Si—Si Shells of PS

red PSi yellow PSi green PSi

1st R

2nd R

3rd R

2.34 2.34 2.34

3.81 3.81 3.79

4.49 4.50 4.55

0.75 0.67 0.73 0.58

690

Properties

Figure 14.4 Dependence of dimensionality and size with PSi “color.” (a) Bandgaps of the three types of emitting PSi: Eg1 from PLE and Eg2 from PL peaks. (b) Theoretical calculation of confined bandgap energy of PSi of wires and dots by LCAO. (c) First shell NSi—Si of red, yellow and green PSi from EXAFS fits. (d) Dimensionality and size dependence with the NSi-Si from (i) bare wire: curves 3 -cylinder (100), 4 and 5–two types of hexagonal columns (111); (ii) wire network (one-plus-a-fraction dimension): curves 1–(111) and 2–(100), details see Ref. 48. As shown by the guiding lines, the PSi in this study favor the nanowire network and wire structures, with average diameters of 2.2, 1.9 and 1.3 nm, respectively for red, yellow and green PSi.

nanowire (one dimensional or less) for green PSi, are suggested. The mean core sizes are 2.2, 1.9 and 1.3 nm respectively for red, yellow and green PSi. It was pointed out in Refs 48 and 50 that the quantum wire nanostructure is better suited for PSi application in EL quantum dots. Strong and stable blue photoluminescence, visible to the naked eye under 0.4 μW of 300 nm and 2.7 μW of 370 nm excitation, has been observed for samples of Si clusters embedded in SiO2 matrices, prepared by rf co-sputtering followed by N2 annealing at 800°C. Si K-edge EXAFS and NEXAFS strongly suggest that Si nanoclusters have crystalline cores after annealing.[55]

14: Nanostructured Electronics, Tsu and Zhang

691

Figure 14.5 Schematic of one of the first PSi LED. (Schematic of one of the first PSi LED based on Ref. 55.)

The Si—Si bond length in the annealed sample is 2.35 ± 0.02 Å, as expected, whereas the Si—O distance is 1.58 ± 0.02 Å instead of the expected 1.62 Å. Our EXAFS data does not show direct evidence for crystallinity: the fitted Debye–Waller factors of the first shell Si—Si bonds at about 0.006 ± 0.003 Å2 suggest a structure between amorphous and crystalline for which the factors are 0.004 and 0.007 Å2, respectively. Alternatively, EXAFS suggests that the structure is composed of clusters which have crystalline cores surrounded by disordered bonds. Either way, the absence of any features corresponding to second shell Si—Si—Si is not surprising, considering the small first shell Si—Si contribution and the rapid fall-off of EXAFS with distance. The formation of more Si—Si bonds in the Si-clustered samples is responsible for the increases in PL efficiency in the blue.

14.3 Applications 14.3.1 Porous Silicon Since Canham’s discovery,[20] there have been continuous reports of porous silicon-based diodes (LEDs), as well as photodetectors, opticallogic gates, etc. The first PSi LEDs were those of Richter et al.[55] and Koshida and Koyama.[56] Figure 14.5 shows the device structure, which represents a typical early LED work. Its EL spectrum peaked at ∼680 nm.

692

Properties

This LED had a structure of semi-transparent metal–porous silicon layer–p-type silicon–Al electrode with an external quantum efficiency of only 10−5% and with very limited device lifetime. Since then, the efficiency has been much improved to ∼0.2%, close to a practically useful level of ∼1% of industrial requirement.[57–59] These devices are represented by a structure of indium tin oxide (ITO)–p+ PSi layer–n− substrate, Al/poly-ascontact–n+ PSi layer–p− substrate, and a p+ nn+ PSi structure, respectively. Both the PL and EL spectra of PSi are normally broad, similar to other types of nanostructures and nanoclusters. A dramatic reduction of the spectral width of EL and the radiative decay time is accomplished by combining micro-cavities to enhance LED performance,[60] leading to the possibility[61] of integration of PSi LED with standard bipolar circuitry. There are modest improvements in device lifetime.[62,63] The former involves encapsulating the porous silicon in aluminum and aluminum oxide, whereas the latter involves oxidizing the porous silicon surface to prevent further oxidization and to form stable passive layers with radiative centers. The longest reported device lifetime is several weeks either under DC operation or continuous pulse operation. Although oxide passivated structures show better stability, efficiencies are generally significantly lower. The fundamental shortcoming of porous silicon (they are extremely reactive and fragile) seems to be difficult to overcome, except recently it was shown that gently etched PSi has much improved morphology.[64] A photodetector with a structure of Al/RTO (rapid thermal oxidized) PSi/p—Si/Al exhibited higher responsivity at 350 nm than a UV-enhanced Si photodiode with an external quantum efficiency of 75% at 740 nm.[65] A large optically induced absorption change in PSi has been demonstrated in an all-optical logic gates (invert- or NOR-gate function).[66] The porosity of PSi can be selectively obtained under different etching conditions. As a result, the refractive index of the material can be tuned. This enables the fabrication of distributed Bragg reflectors by stacking alternating quarter-wavelength-thick layers of higher-and lower-porosity material. It has been reported that broadband mirrors entirely made from porous silicon for CW and mode-locked Ti:sapphire lasers and for a tunable dye laser.[67]

14.3.2 Photoluminescence in nc-Si/SiO2 Superlattices To overcome the problem of structural robustness associated with the porous silicon, it was proposed that nanoparticles of silicon with sizes in

14: Nanostructured Electronics, Tsu and Zhang

a

693

b

Figure 14.6 (a) PL intensity vs. photon energy for a nine-period Si/IAG superlattice annealed in oxygen and hydrogen at 850°C; and (b) cross-section TEM of the sample.

the range of several nanometers sandwiched between thin oxide layers to form a superlattice may solve the problem of mechanical robustness while retaining the features of quantum confinement as in porous silicon.[68] In Ref. 55, the name IAG-superlattice was introduced, for Interface Adsorbed Gas-superlattice. This name originates from the scheme that oxygen is introduced via surface adsorption in order to prevent the formation of a very thick oxide. In this scheme, silicon up to 12 nm thick is deposited either in the amorphous phase or crystalline phase, followed by the in-situ growth of a thin oxide. This arrangement is repeated to acquire the desired volume of interaction. In practice, the maximum period of the superlattice structure tried was only nine periods. The reason to keep relatively few periods is dictated by the need to further passivate interface defects, annealing in the presence of gas mixtures. In particular, annealing in H2 or H2 + O2 gives better results as shown in Figure 14.6(a) with PL peaks at 1.7 eV and 2.34 eV. Figure 14.6(b) shows a cross-section TEM of the structure. It was found in surface Auger that the oxygen peaks coincide with the silicon dips, indicating that the structure indeed consists of silicon separated by regions with high oxygen content.[69] Moreover, the 2.34 eV peak is attributed to surface effects. This brings up an important

694

Properties

point in all nanostructured materials. In devices dictated by bulk, surface or interface regions are considered undesirable. As the particle size shrinks to the nanometer regime, surface or interface regions become significant or even dominate the “bulk,” thus we need to reorient our views so that surface or interface regions are the focus of our considerations. The grain size of the silicon nanoparticles was found to be approximately 3 nm using Raman scattering and checked by TEM.[68] The mechanism controlling the grain size is quite involved. We only touch the salient points here. Basically, unlike the amorphous-crystalline phase transition in bulk, in very thin structures, the phase transition is controlled by proximity effects rather than simple temperature.[70] These considerations prevented us from using extremely thin silicon layers as recently demonstrated in the work of Lockwood et al.[71] We postpone the discussion on what happens when we use a much-reduced thickness for the silicon layers until the treatment on the epitaxial Si/O superlattices, or Si/O SAS, for semiconductor atomic superlattice.

14.3.3 Luminescence from Clusters Elemental semiconductors Ge,[72] Si,[73] and C,[74] embedded in an SiO2 matrix exhibited fairly strong and stable PL, with peaks ranging from IR to blue.[73,74] Samples for LED were fabricated consisting of 45 nm thick polycrystalline Si films deposited initially as amorphous films by e-beam evaporation onto 70 nm SiO2 films thermally grown on n+ Si substrates, followed by Ge implantation, to create a supersaturated solid solution of Ge in the SiO2 film with approximately uniform Ge (∼5 nm in diameter) concentration of 5%. The samples were subsequently annealed at 600°C, 1 × 10−6 torr, for 40 minutes to induce precipitation. The EL spectrum was broad and peaked at 1.2–1.4 eV. Samples of silicon clusters can be prepared by sputtering SiO2 onto silicon wafers without additional heating. After annealing at 800°C for 20–30 minutes in N2, the typical PL spectra show typical quantum size effect in Figure 14.7(a)–(c) with increasing Si to oxide ratio. A more detailed discussion on the EXAFS characterization and mechanisms of cluster size control may be found in Refs 70 and 73. Stable blue and unstable UV PL from C clusters embedded in SiO2 matrix has been observed.[74,75] The EXAFS analyses for O and C are more complex than Si. Applying the K-edge EXAFS to the system, it was found that the blue PL was related to C nanoclusters with local π-bonds. The unstable UV PL is thought to be related to silicon.

14: Nanostructured Electronics, Tsu and Zhang

695

Figure 14.7 PL of silicon clusters in SiO2 matrix. The intensites of PLs are related to the cluster densities found by TEM.

Recent effort of Er-doped Si clusters embedded in a SiO2 matrix has drawn a lot of attention because the demand for silicon based photonics and the standard telecommunication wavelength of 1.54 μm contributed by Er ions. For example, EL at 1.54 μm with a quantum efficiency of ∼1% at room temperature has been demonstrated in MOS-type LED.[74] An optical gain value of 0.7 cm−1 at 1535 nm was reported in nanocrystal/ Er-doped SiO2 waveguides.[75]

14.3.4 Semiconductor/Atomic/Superlattice: Si/O Superlattice We introduce the term Hetero-Epilattice-Superlattice, HES, to describe a superlattice system consisting of epitaxial layers of silicon, for example, sandwiched between adsorbed disorder species. The terminology has subsequently been changed from HES to SAS for semiconductor/Atomic/ Superlattice. Several years ago, in search of a barrier system for silicon, where a lattice matched heterojunction is lacking except in the SixGe1−x system,[78] it was proposed that perhaps the best and simplest way to build a barrier onto silicon is to utilize the concept of strain layer superlattice with sufficiently thin silicon layers.[8] Subsequently, it was realized that the best way to limit the thickness of the oxide which introduces disorder, is to limit the supply of oxygen by surface adsorption.[9] This is so because after a monolayer coverage of oxygen on a clean silicon surface, further oxygen adsorption is not possible without substantial heating to

696

Properties

Figure 14.8 A scheme of a Si/O quantum well using a Si/O superlattice as the barrier. (From Tsu, Ref. 8, with permission.)

drive in the oxygen via diffusion. This method is, therefore, in the realm of self-organized crystal growth.[79] SAS is the outgrowth of the originally proposed barrier for silicon described in Ref. 8. Basically, the concept of a strain-layer superlattice is that, with a sufficiently thin epitaxial layer, the strain energy in each layer is below the energy needed for the growth of point defects or dislocations. It is important to note that dislocations have an activation energy for nucleation and a lower activation energy for growth. Therefore, in principle, it is possible to greatly exceed the energy requirement without actually generating defects. Figure 14.8 shows the scheme of a strain-layer epitaxial Si/O barrier system. Because the width b << w, Eb1 and Eb2 are much higher than E1 and E2, the quantum well states. This is the basis of a strain-layer barrier. The effective barrier height may be as high as 1 eV, so that it is possible to design the separation E2 − E1, depending on the width w, to be much greater than kBT at room temperature. The thickness at which SiO2 can be tolerated for continuous epitaxy is the key for this kind of superlattice. What is under debate is whether the monolayer of SiO2 on Si is ordered or disordered. Epitaxial growth of silicon may be continued after interruption with oxide growth.[80] The explanation involves the natural seeding provided by even the reconstructed Si(100) at low-pressure chemical vapor deposition (CVD) with low flow rate. Based on grazing angle X-ray diffraction, Rabedeau et al.[81] found evidence for low coverage 2 × 1 epitaxial structure at the SiO2/Si interface for dry oxides grown on ordered Si surfaces at room temperature, however, the 2 × 1 structure does not survive thermal annealing. Subsequently, the Si/O superlattice was formed and I–V measurements showed a barrier height of more than 1 eV.[9] However, more careful theoretical and experimental work showed that the barrier height is 0.5 eV,[82] which is sufficiently high for a variety of applications in electronic and

14: Nanostructured Electronics, Tsu and Zhang

697

optoelectronic applications. Basically, the method involves the adsorption of oxygen onto a clean Si surface followed by Si deposition of 1–2 nm thick, using the in-situ reflection high energy electron diffraction (RHEED) pattern for the monitoring of the 2 × 1 surface reconstruction as a measure of the restoration of a clean silicon surface before the next growth step. Recently, more detailed work about defects and degree of epitaxy, has been carried out on the hetero-epitaxy with oxygen exposure.[11,13,83] Figure 14.9 shows the typical in-situ RHEED (reflection high energy electron diffraction) pattern, used to monitor epitaxy. The 2 × 1 dimerization, the half order, is clearly shown in Figure 14.9(a) from surface reconstruction of the Si(100). Exposure to 10 L of oxygen at 30°C does not eliminate the 2 × 1 structure (shown in Fig. 14.9(b)), until 1.1 nm of Si is deposited (shown in Fig. 14.9(c)), where the transformation from a 2D to a 3D pattern is quite evident. A second exposure at 10 L does not substantially alter the RHEED (shown in Fig. 14.9(d)), however, the 2 × 1 reappears after 8 nm of Si is deposited (shown in Fig. 14.9(e)). This series of in-situ RHEED demonstrates that epitaxial growth is continued beyond the adsorbed oxygen layer. A four-time repeated sequence with 10 L oxygen exposure followed by 1.1 nm of Si deposition at 550°C has been formed and the 2 × 1 reconstruction was brought back with the silicon layer of 12 nm, serving to restore the epitaxial structure as well as a capping for protection. Similar structures were fabricated for a class of electroluminescent devices. Figure 14.10(a) shows the emission of visible light from the edge of an Al contact, yellowishgreen to the naked eye, and Figure 14.10(b) shows a spectrum of a peak at 2 eV with additional higher energy contributions. (The spectrum for EL covers a range from red to near UV shown in Figure 14.10(b), therefore, the appearance of yellowish green is due to the partially transparent gold contact used as an electrode.) It is important to point out that this SAS EL device is extremely stable and robust. With the application of 6–8 V, this EL diode has operated continuously for almost one year, representing, to our knowledge, the longest operation of a silicon nanostructured EL device to date. In fact, our life test has lasted for a year without detectable degradation![13,83] Estimate of the conduction band-edge offset using HOMO-LUMO is ∼1 eV for monolayers of oxygen in between adjacent 1.1 nm of epitaxial silicon layers. This is fairly close to the PL and EL observed in these devices.[13,84] Estimate of shear strain and tensile strain from DFT gives 1% and 10% respectively.[13] Since the strain energy is spread out several atomic layers on both sides of the oxygen monolayer, the layer period

698

Properties

Figure 14.9 RHEED of Hetero-Epilattice Superlattice showing the reestablishment of epitaxy. The 2 × 1 reconstruction in (a) is not destroyed after oxygen exposure in (b), however few monolayers of Si deposition transforms the 2D pattern to 3D in (c), and (d). Full restoration is achieved after 8 nm of Si, shown in (e).

must be at least doubled before it becomes possible to clearly show up the periodic structure of the layers in HRTEM. For this reason, we have tripled the period and successfully obtained good image of a periodic structure.[13,85] However, these samples did not show more than a few percent lower in the emission peak, indicating that the states responsible for the observed emission is likely to be due to Si—O complex, because the simple picture of a particle in a box should have lowered the emission

14: Nanostructured Electronics, Tsu and Zhang

a

699

b

Figure 14.10 EL of LED using HES. (a) photograph of EL; (b) EL spectrum of a LED. The EL (solid curve) and PL (dotted curve, excited by 457.9 nm line of argon ion laser) spectra measured at room temperature of sample 1, left, at reverse bias V = 14 V and sample 2, right, at reverse bias V = 20 V.

energy by a factor of four![13] It was suggested to us that it may be possible to obtain similar results by simply interrupting the deposition without introducing oxygen adsorption. Well, there was no sign of any kind of PL or EL, indicating that even if defects were introduced, emission of light is definitely associated with introduction of oxygen.[13] What is most revealing is the significant increase (more than ten-fold) in luminescence intensity in EL after annealing in O2 + H2 at 400°C for 30 minutes. That is also the reason why we did not significantly increase the number of Si/O periods, because if we did, the structure might be too thick for efficient annealing. The bottom line is simple: O—Si complex is responsible for the optical activity, while H passivates the dangling bonds, the same old theory of passivation. It is quite possible that the Si/O superlattice serves as a diffusion stopper for oxygen diffusion in the annealing process, substantially increasing the Si—O complexes which are responsible for the observed emission with a fairly broad spectrum covering from red to near UV as shown in Figure 14.10(b).

14.3.5 Amorphous Silicon/Oxide Superlattice Lu and Lockwood reported strong visible PL from amorphous-Si/SiO2 superlattice.[71,86] The PL peak can be shifted from 1.7 eV to 2.3 eV when the thickness of amorphous silicon layer in the superlattice is 3 nm to 1 nm. The visible light emission was explained in terms of quantum

700

Properties

confinement of electrons in the two-dimensional silicon layers. It is appropriate to discuss the difference of this a-Si/SiO2 and the HES (SAS for semiconductor-atomic-superlattice in Ref. 13) discussed in more detail in Section 14.3.4. As long as the amorphous silicon thickness is below that of the coherence length, from a quantum phenomenon point of view, it is not so important whether the confined electrons can interfere in an amorphous layer or in a crystalline layer. However, amorphous bonding allows more flexibility resulting in a larger variety of bonding defects which ultimately manifest in increased recombination and scattering centers, in comparison to the crystalline case. Nevertheless, hydrogenated amorphous silicon photovoltaic materials have been steadily improved. Any premature judgment is somewhat risky at this stage. Comparing with the SAS presented in the last section, one can infer that at least the amorphous case does not allow efficient carrier injection as the Si—O superlattice devise.[13]

14.3.6 nc-Si in an Oxide Matrix Resonant tunneling involving discrete quantum states in nanocrystalline-Si (nc-Si) with a-SiO2 barriers[87,88] gives rise to very rich features, which include Coulomb blockade and avalanche multiplication[89] that results in slow oscillations in time due to the presence of negative resistance,[90] or 1/f-like noise spectrum due to trapping.[13] Basically, the idea is to create sufficiently small particles of silicon in the nanometer regime, tunneling via these quantized energy states should give rise to sharp features. The observed current peaks have line widths approximately given by kBT.[88] There the mystery arises. In the structure for resonant tunneling, a layer of silicon consisting of nanoscale particles embedded in an SiO2 matrix sandwiched between two oxide barriers, there should be a size distribution resulting in broad features, or even wiping out any conductance peaks. The answer to the mystery lies in the fact that nucleation and subsequent growth of the nanocrystallites are under some constraint conditions. The most important one is the thickness of the layer sandwiched between something else. The effects of a constraint condition have been discussed in relation to the control of particle size via crystallization from the amorphous phase presented in Section 14.3.2. We take this opportunity to reiterate that crystal growth in an extremely thin layer dominated by proximity effects is a new frontier of crystal growth. Without going into too much detail in this section, it is important to point out various features normally one does not encounter. In forming

14: Nanostructured Electronics, Tsu and Zhang

701

Figure 14.11 Theoretical and experimental current vs bias voltage. (Taken from Ref. 80.)

many nanoscale particles in a layer for I–V measurements, these nanoparticles are all in parallel. Imagine, one tries to measure the transconductance of one hundred transistors having slightly different features connected in parallel. Fortunately, the situation is not entirely hopeless as presented in Ref. 88. Basically, the observed multitudes of current jumps may arise from very small difference in the particle size. Due to IR drop in the substrate, they appear at fairly well separated voltages. A simple theory, including the avalanche multiplication triggered by the resonant tunneling through each particle or a group of particles having approximately the same particle size, can indeed fit the measured conductance data as shown in Figure 14.11, taken from Ref. 89. Results indicate another interesting point: particle size variation is in almost discrete steps. In retrospect, it seems to be obvious that in a small nanoscaled particle, size increase must be dictated by the size of the basic unit cell. Another interesting effect is the sharpening of the resonant conductance peak with light. This mystery has been expounded in great detail in Ref. 13. By going through many unpublished data, it was concluded that the conductance jumps were due to gG0, where g is a degeneracy factor and G0 = e2/h, while the slow oscillations were due to trapping.[13] Using phase sensitive

Properties

702

I–V measurements, it was discovered that the drastic sharpening of the peak with light of photon energy above the bandgap may be due to the filling of charged traps.[90] At this stage, it is natural to ask how such a complicated system may be applied to devices. As discussed, the formation of these nanocrystallites needs new studies in crystallization in a constraint situation, taking into account the proximity of hetero-materials on each side of the layer. In dealing with the presence of many particles having slightly different sizes in parallel, the situation is compounded by the presence of a large series resistance in the substrate with deep depletion. This series resistance results in voltages depending on the number of particles in parallel even if the same energy state is involved. It should be realized that these complications arise because one is engaged in a general study, not targeted to an actual device. In an actual device, the fabrication technique has advanced to the point that only a handful of particles are in parallel, and with better basic understanding, the particle size distribution can be better controlled. Thus, it should be quite possible to utilize these features in the design of multiple logic, functional devices,[91] and even in the emerging field of quantum computing.[92]

14.3.7 Electronic Applications of Si/O Superlattices In Section 14.3.4, we discussed the formation of the Silicon/Oxygen Superlattice. We shall point out some of the possible incorporations of this technology into future electronic devices. A brief summary is outlined to give a better grip of what it is and how to make use of it. • It is a superlattice consisting of alternate layers of nanometer scale silicon layers sandwiched between adsorbed oxygen monolayers. • The epitaxy of silicon is continued beyond the adsorbed disordered layer of oxygen, forming a structurally perfect superlattice, as if no disordered layers were present. • Electrically, a barrier consisting of two adsorbed monolayers of oxygen with a thin layer of silicon, 1–2 nm thick, shows a barrier height of 0.5 eV in I–V measurement. • The silicon deposition with thickness in excess of 3 nm on an adsorbed oxygen monolayer is epitaxy- and defect-free. Therefore, such a system is ready to play an important role in the design of high speed and low power transistors.

14: Nanostructured Electronics, Tsu and Zhang

703

Figure 14.12 A typical high resolution cross-section TEM with 10 Langmuirs of adsorbed oxygen.

Figure 14.12 shows a typical high-resolution cross-section TEM with 10 L of adsorbed oxygen.[94] The presence of clusters of oxygen with dimensions approximately 3 nm is quite evident. It is important to recognize the fact that the presence of oxygen clustering in the oxygen adsorbed layer does not invalidate a barrier; because electrons are de Broglie waves, they cannot leak out of the “hole” with dimensions smaller than a wavelength. Figure 14.13 shows a couple of examples of I–V. Note that there is a 1000 : 1 difference in the measured currents between the forward and reverse directions, acting as a good conductor and good insulator, respectively. This feature may be applied to 3D integration of ICs. The SAS is still in its research stage, nonetheless our results should open the door for a wide variety of future applications.

704

Properties

14.3.8 Single Electron Transistor There are a wide variety of quantum devices.[95] The single electron transistor, SET, is selected because the subject matter is somewhat close to the subjects we have presented thus far. The observation with current transport of discrete electronic states in a quantum dot set the stage of SET.[96] An excellent tutorial treatment may be found in an article in Physics Today.[97] Basically, SET promises extremely high Figure 14.13 Measured I–V density with ultrasmall power dissipation characteristic of a barrier and the possibility of room temperature formed by the superlattice. operation. The application of SET includes Dashed lines shows a case with 50 L of adsorbed oxygen. devices of nanoscaled memory, capacitance, logic gates, etc. Effects of single electron charging were observed in metal particles many years ago.[98] Single electron charging in a field effect transistor (FET) was observed more than ten years ago.[38] However, the first experimental study of a new nanometer FET with a single barrier in a onedimensional channel was presented by Chou and Wang.[99] Not only is it not absolutely necessary to confine the electrons by two barriers, but a single barrier, under appropriate conditions, can result in single electron charging effects. In 1996 Tiwari et al. reported a memory device utilizing silicon nanocrystallites.[100] The device was fairly stable with no degradation in performance after more than 109 cycles of operation. It is impressive, though far short of the desired cyclability of 1013. Figure 14.14 shows the schematic cross-section and band profiles during injection (write cycle), storage, and removal (erasure) of an electron in the device. A thin tunneling oxide (1.1–1.8 nm thick) separates the inversion surface of an nchannel silicon FET from a distributed film of nanocrystals of silicon, 5 nm crystallites, at a density of 1 × 1012 cm−2, covering the entire surface channel region. A thicker tunneling oxide (4.5 nm or higher) separates the nanocrystals from the control gate of the FET. Injection of an electron occurs from the inversion layer via direct tunneling when the control gate is forward biased relative to the source and drain. The resulting stored charge screens the gate charge and reduces the conduction in the inversion layer.

14: Nanostructured Electronics, Tsu and Zhang

705

Figure 14.14 (a) A Schematic cross-section and (b) band diagram during injection, (c) storage, and (d) removal of an electron from a nanocrystal. (From Fig. 1 of Tiwari et al. of Ref. 100 with permission.)

The nanostructure made by the combination of electron beam lithography and reactive ion etching can also lead to the other approach to singleelectron MOS memory,[101] operable at room temperature.[102] The structure consists of a narrow channel MOS FET with a width (∼10 nm) smaller than the Debye screening length of a single electron, and a nanoscale polysilicon dot (7 × 7 nm) as the floating gate embedded between the channel and the control gate, as shown in Figure 14.15. Even without adding a tunnel oxide between the channel and the polysilicon floating gate, the potential barrier still exists because of the thin native oxide at the grain boundary. As reported, this approach limited the fluctuation of the actual transistor channel width, the floating-gate size, and tunnel barrier thickness in Ref. 101, as well as the broad size distribution of silicon nanocrystals in Ref.100. The storage of one electron leads to a discrete shift in the threshold voltage. A staircase relationship between the charging voltage and the threshold shift, exemplifies a self-limiting charging process. Single-electron nano-capacitance devices utilize nanometer-size metal and semiconductor particles as the active device elements.[103,104] For sufficiently small (∼2–4 nm in diameter) particles, the single-electron charging energies that characterize a particle are much greater that kBT at room

706

Properties

Figure 14.15 Schematic of a single-electron MOS memory that has a narrow silicon channel and a nanoscale polysilicon dot as the floating gate. The crosssection view illustrates the floating gate and the channel region. (From Fig. 1 of Yano et al. of Ref. 101 with permission.)

temperature, suggesting the possibility of operation of metal nanocrystalbased single-electron devices at ambient temperature. For example, Markovich et al.[105] reported a technique for fabricating single-electron capacitance devices from a two-dimensional close-packed organically functionalized Ag nanocrystal layer. Evidence of a Coulomb blockade, a step structure reminiscent of a Coulomb staircase, and memory charging effects were obtained. The step structures were reproducible from device to device over multiple scans and different frequencies. The authors argued that each of the steps reflected an increase in capacitance of the device due to collective single-electron (or hole) charging of the particles in the monolayer film. It is noted that the increase of capacitance as particle size is reduced below a couple of nanometers discussed in Section 14.2.5 is neglected here. Figure 14.16 shows the basic device structure, its equivalent circuit, and working function, in which PMMA (polymethylmethacrylate) serves as the dielectric spacer layer. Ohshima and Kiehi proposed a new approach to digital circuitry based on phase bi-stability in locked single-electron tunneling oscillations.[106–108] The scheme is based on binary logic states associated with the tunneling phase in single-electron tunnel junctions pumped at twice the tunneling frequency and activated by clocking the dc bias. Figure 14.17 shows the proposed circuit in which tunneling junctions with capacitance, C, are pumped by a common ac source, Vp. The pump at ωp supplies the drive for locking the oscillations and the circuit timing reference. The basic operation principle is the generation of single-electron tunneling oscillations at ωset phase-locked to the pump, creating a multi-stability due to the indeterminate phase relationship between ωset and ωp = n ωset, where n was 2. Phase information is clicked from gate to

14: Nanostructured Electronics, Tsu and Zhang

707

gate by sequentially clocking the dc bias in each gate. When the dc bias is raised, the gate is activated and subharmonic locking is established. Since the gates are capacitively coupled, the final state is influenced by the states of the previous gate. Therefore, Tunneling Phase Logic, TPL, the synchronization of single-electron tunneling events with the pump to produce a phase bi-stability, provides the properties needed for potential applications in digital logic circuitry. Self-assembled Si quantum dots have been tested for resonant tunneling devices operating at room temperature.[109] The size of the hemispherical Si quantum dots grown by low pressure CVD, 5–20 nm in diameters and 1–10 nm in height, leads to strong carrier confinement in the Si nanodot allowing the observation of Figure 14.16 (a) Cross-sectional negative differential conductance schematic of a device (not to (NDC) at room temperature. The scale); (b) equivalent RC circuit; (c) tunneling device consists of a doubleequivalent energy-level diagram. In barrier of n+-Si(100)/SiO2 (3-nm(b), C1 represents the thick)/Si nanodots/native SiO2. This is Al/Al2O3–nanocrystal junction capacitance, R is the same an important development, therefore, junction’s tunneling resistance, and this work, in our views, should be C2 represents the repeated at lower temperatures for a nanocrystal–PMMA–Al junction capacitance. In (c), regions 1 and 5 better NDC. Si nanopillars can be utilized to are Al electrode layers, region 2 is an Al2O3 layer (1 nm), region 3 is make a single-electron transistor that the metal nanocrystal monolayer operates using a nanometre-scale (5 nm), and region 4 is the PMMA vibrating arm.[110] The transistor is a insulating spacer layer (30–40 nm). type of device known as a nano(Reprinted by permission of the electromechanical system, or NEM. publisher.) The transistor features a silicon arm about 200 nanometres long and only tens of nanometres across. By applying an ac voltage with a frequency that matches the resonant frequency of the arm—in this case between 350

708

Properties and 400 megahertz—they were able to make the arm vibrate between the electrodes. This resulted in a flow of electrons from the source to the island, from where the electrons then tunnelled towards the drain electrode.

Figure 14.17 TPL of three coupled stages. Ultrasmall tunnel junctions (double-box symbols) with capacitance C are pumped by a common ac source and dc biased by clocked sources through a load resistance. (Reprinted by permission of the publisher.)

14.3.9 Quantum Dot Laser

The potential application of selfassembled quantum dots (QD) in optoelectronics is notable. This insitu growth approach utilizes the island growth, Stranski–Krastanow (SK) mode, in highly strained hetero-epitaxial systems, such as InGaAs on GaAs. The defect-free self-assembled quantum dots can be simply fabricated by using conventional molecular beam epitaxy (MBE), metal-organic vapor phase epitaxy (MOVPE), or metal-organic chemical vapor deposition (MOCVD). The size distributions of these dots are uniform to within 10%, with densities as high as 1011 cm−2. The PL efficiencies of the dots are greater than, or equal to, that of the reference quantum well.[111] By using these self-assembled quantum dots as an active layer, low-threshold edge-emitting lasers have been fabricated. Soon afterwards, the QD vertical-cavity surface-emitting laser (VCSEL) appeared. VCSEL, using QDs in the active region, is more attractive, since it has the advantage of controlling both the electron and photon modes in a microcavity structure. The matching of the cavity mode with narrow bandwidth light emission from QDs results in a high-performance light source with very low threshold current. The bandwidth, which critically determines the temperature characteristics of the VCSEL, can be designed by controlling the dot size distribution. Therefore, the QD VCSEL may represent an optimum optical device utilizing QD. Saito et al.[112] reported an InGaAs QD VCSEL when operated under continuous-wave current of 32 μA at room temperature. The structure of VCSEL is shown in Figure 14.18, grown by MBE. Bottom 18-period, ndoped, AlAs/GaAs distributed Bragg reflectors (DBRs) and a spacer layer

14: Nanostructured Electronics, Tsu and Zhang

709

Figure 14.18 Quantum dot VCSEL with 10-period InGaAs dots in the active region. On the average, the dots were 28 nm in diameter and 3 nm in height with a density of 2 × 1010 cm−2, covering 12% of the surface. (From Fig. 1 of Kiehi et al. of Ref. 106 with permission.)

of AlGaAs were first grown on a GaAs substrate. The thicknesses of AlAs and GaAs layers in the DBR were 64.9 and 51.5 nm, respectively. The interface layer between them was a graded composition AlGaAs layer (18 nm). Next, 10 periods of In0.5Ga0.5As dot/Al0.25Ga0.75As layer were grown at 520°C. The dots were obtained by alternate depositions of GaAs, 0.2 monolayer (ML) and InAs, 0.2 ML, including a 2 second pause. The tendency of the flattening of the AlGaAs cover layer on the dots results in the subsequent growth being as good as the first layer. After the growth of the InGaAs dot active layer, a p-type AlGaAs spacer layer and top 14.5period DBRs were grown at 600°C. A reflectivity spectrum of an as-grown wafer showed that the DBR mirror center and resonant cavity were positioned at a wavelength of around 960 nm. With a 3-nm height of the QD, effects of quantum confinement dominate over the thermal energy at room temperature. Huffaker et al.[113] recently reported a low threshold, oxide-confined (7 μm aperture) VCSEL based on a self-assembled QD active region. The operating condition was at room temperature under the pulsed threshold of 560 μA. The schematic cross-section is shown in Figure 14.19. The structure consists of a lower DBR of twenty-six n-type AlAs/GaAs quarter-wave pairs, a half-wave cavity spacer consisting of lower n-type and upper p-type Al0.99Ga0.01As layers cladding the QD active region, and a single upper p-type GaAs quarter-wave layer followed by heavily p-type Al0.75Ga0.25As and GaAs contact layers. The upper DBR is completed by six additional quarter-wave pairs of MgF/ZnSe deposited using electron

710

Properties

Figure 14.19 Schematic illustration of the oxide-confined VCSEL containing the In0.50Ga0.35Al0.15As active region. (From Huffaker et al. of ref. 113 with permission.)

beam evaporation. The last heavily p-doped layers are designed to be selectively etched from the VCSEL cavity prior to the MgF/ZnSe deposition. The QD active region is formed by depositing five monolayers of In0.50Ga0.35Al0.15As sandwiched between undoped GaAs and graded AlGaAs barriers. The substrate temperature is held at ∼620°C for growth of the DBRs and cavity spacer, and reduced for the QD deposition and growth to ∼500°C. A 20 s pause after the five monolayer deposition is used to allow the QD formation, monitored with RHEED. The 15% Al on the column III lattice sites is added to shift the room temperature QD emission wavelength to ∼0.98 μm. Atomic force microscope (AFM) images showed the In0.50Ga0.35Al0.15As QDs formed with the six monolayer deposition to have lateral sizes of roughly 20 nm diameter with a high density of ∼1011 cm−2. A purple (420 nm) laser diode, using self-assembled QDs in the active layer, has been investigated by Narukawa et al.[114] This active layer consists of seven periods of undoped In0.20Ga0.80N (3 nm)/In0.05Ga0.95N (6 nm) MQW, sandwiched between GaN wave guiding layers (0.1 μm in each) and Al0.15Ga0.85N cladding layers (0.4 μm in each). Using cross-section TEM and energy-dispersive X-ray (EDX), the authors found that the radiative recombination was from quantum self-assembled QDs. These QDs were in the In-rich region in the wells and had sizes from 3 to 5 nm with a peak at 3 nm in diameter. Similar to the structure of QD laser, a simple approach to LED is the self-organized island formation,[115,116] with three monolayers of InAs islands sandwiched between 5.5 nm of GaAs layers. These LEDs emit light over a broad spectrum, with a typical linewidth of 120 nm, peaked between 1000 and 1100 nm.

14: Nanostructured Electronics, Tsu and Zhang

711

The modulation bandwidth of conventional 1.0–1.3 μm self-organized In(Ga)As quantum dot (QD) lasers is limited to ∼6–8 GHz due to hot carrier effects arising from the predominant occupation of wetting layer/barrier states by the electrons injected into the active region at room temperature. With the assistance of tunnel injection and p-doping, highperformance In(Ga)As undoped and p-doped tunnel injection selforganized QD lasers emitting at 1.1 and 1.3 μm. Undoped 1.1 μm tunnel injection lasers have ∼22 GHz small-signal modulation bandwidth and a gain compression factor of 8.2 × 10−16 cm3.[117] Using molecules as electronic components is a new direction in the science and technology of nanometre-scale systems. The ultimate limit would be a device where electrons hop on to, and off from, a single atom between two contacts. The concept has been examined with two related molecules containing a Co ion bonded to polypyridyl ligands, attached to insulating tethers of different lengths.[118] Changing the length of the insulating tether alters the coupling of the ion to the electrodes, enabling the fabrication of devices that exhibit either single-electron phenomena, such as Coulomb blockade effect.

14.4 Challenges in Quantum Dot Devices Quantum well, QW, structures with planar contacts do not present problems with input/output, I/O. However, devices with quantum dots, QDs, run into some serious problems with I/O.[119] In addition, robustness and redundancy must be solved before any system can move from a laboratory experiment to industrial application. For example, the smallest MOSFET involves more than 1000 electrons under the gate. As device size is reduced, several limits are beyond scaling. Doping difficulties may be circumvented using carrier injection. In fact one frequently comes across the term quantum transport. In reality transport involves diffusive action, while quantum tunneling is dominated by phase coherency. As long as tunneling dictates the transfer of charge and energy excitation in general, who would want to dope the QDs? One still needs to avoid the formation of Schottky barrier by using n+ or p+ between the QDs and the metal contact. And as we have shown in the section on doping, this transition region must be much lager than the size of the QDs, and therefore many problems such as statistical uniformity of dopants and the binding energy of dopants, etc. may be avoided. Now, if the contacts are big, what is the advantage of scaling down the active device size? The reason is dictated by speed and power dissipation.

712

Properties

For optical and optoelectronic applications, the situation is more under control. Because the optical wavelength is much larger, efficient coupling is dictated by sizes at least a significant fraction of the optical wavelengths involved. Thus, the situation is quite similar to the case of QWs. Then why would we need QDs? We need quantum dots as clusters forming an effective material with QDs embedded in a matrix. For example, optically induced transparency may be achieved many orders of magnitude below what is possible in bulk solids.[120] Thus, QDs serve to enrich the possible parameters needed in a given device.

14.5 Epilogue As device size is reduced, several limits are beyond scaling. For example, barriers produced by doping cannot be scaled down due to the solid solubility limit of dopants. The abruptness of the heterojunctions allows scaling down to nanometer regime, however, quantum effects set in. Not all quantum effects are good for device performance. Lowering the carrier mobility resulting in degrading the carrier coherence length can eliminate quantum effects, but power dissipation and speed will be sacrificed. The game plan is, therefore, to utilize the quantum effects in seeking new designs of devices. Fundamentally, quantum effects are the manifestation of the wave nature of electrons, which controls constructive and destructive interferences, resulting in new flexibilities and possibilities. Quantized systems have discrete energy–momentum relations, which allow new applications. The reduction in the dielectric screening, the capacitance, the increase in dopant binding energies resulting in the difficulty in doping nanoparticles, may represent undesirable features. Usually lower capacitance translates into a reduction in the RC time constant, however, the characteristic time may not be dictated by RC, rather by tunneling time. The concept of charging a capacitor needs to be modified to include the recognition that it is not the storing of a charge, rather the storing of an electron, that is involved, and that the electron has kinetic energy in a quantum system. The widely separated energy states of electrons in a nanoparticle lead to steps or peaks in current. Therefore, with better understanding and designs, many undesirable features can be turned around to our advantage. The subject matter is huge. Our review is, thus, quite limited, mainly focused on a few systems that we have worked on during the past years.

14: Nanostructured Electronics, Tsu and Zhang

713

Acknowledgments Part of the review describes work performed by the authors under the auspices of the US Army Research Office, Office of the Naval Research, and National Science Foundation. Qi Zhang also wishes to acknowledge the support of his work from CCLRC Darsbury Laboratories and BESSY.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

Esaki, L., and Tsu, R., IBM J. Res. Devel., 14:61 (1970). Tsu, R., and Esaki, L., Appl. Phys. Letts., 22:562 (1973). Cho, A.Y., Appl. Phys. Lett., 19:467 (1971). Matthews, J.W., and Blakeslee, A.E., J. Cryst. Growth, 32:265 (1976). Dohler, G.H., Kunzel, H., Olego, D., Ploog, K., Ruden, P., Stoltz, H., and Abstreiter, G., Phys. Rev. Lett., 47:864 (1981). Tsu, R., Mat. Res. Soc. Sypm. Proc., 102:219 (1988). Zrenner, A., Reisinger, H., Koch, F., Ploog, K., 17th Int. Conf. Phys. Semi. (D. Chadi, and W., Harrison, eds.), p. 325, Springer, Berlin (1985). Tsu, R., Nature, 364:19 (1993). Tsu, R., Filios, A., Lofgren, C., Cahill, D., van Nostrand, J., and Wang, C.G., Solid-State Electronics, 40:221 (1996). Dellow, M.W., Beton, P.H., Langerak, C.J.G.M., Foster, T.J., Main, P.C., Eaves, L., Henini, M., Beaumont, S.P., and Wilkinson, C.D.W., Phys. Rev. Lett., 68:1754 (1992). Tsu, R., Filios, A., Lofgren, C., Dovidenko, K., and Wang, C.G., Electrochem. Solid State Lett. 1 (2) 80 (1998). Zhang, Q., Lofgren, C., Filios, A., and Tsu, R., Physica B, 8:365 (2000), Tsu, R., Superlattice to Nanoelectronics, Elsevier, (2005). Wegscheider, W., Eberl, K., Cerva, H., and Oppolzer, H., Appl. Phys. Letts., 55:448 (1989). Tsu, R., Basic, D., and Ioriatti, L., J. Appl. Phys., 82:1327 (1997). Tsu, R., Ioriatti, L., Harvey, J.F., Shen, H., and Lux, R., Mat. Res. Soc. Sypm. Proc., 283:2093 (1993). Wang, L.W., and Zunger, A., Phys. Rev. Lett., 73:1039 (1994). Lannoo, M., Delerue, C., and Allan, G., Phys. Rev. Lett., 74:3415 (1995). Walter, J.P., and Cohen, M.L., Phys. Rev. B, 2:1821 (1970). Canham, L.T., Appl. Phys. Lett., 57:1046 (1990). Bastard, G., Phys. Rev. B, 24:4714 (1981). Ioriatti, L., and Tsu, R., Surf. Sci., 174:420 (1986). Tsu, R., and Ioriatti, L., Superlatt. Microstruct., 1:295 (1985). Kahen, K.B., Leburton, J.P., and Hess, K., Superlatt. Microstruct., 1:289 (1985). Tsu, R., and Babic, D., Nato Workshop on Optical Properties of Low Dimensional Silicon Structures, p. 203, CNET, France, Kluwer, Dordrect (1993).

714

Properties

26. Tsu, R., and Babic, D., Appl. Phys. Lett., 64:1806 (1994). 27. Tsu, R., and Babic, D., Porous Silicon Science and Technology, Les Houches (J.C., Vial, and J., Derrien, eds.), p. 111, Springer Verlag, Berlin (1994). 28. Matsumoto, et al., in: Proc. 1992, Internat. Conf. Solid State Devices and Materials, p. 478, Tsukuba, Japan. 29. Lehmann, V., and Gosele, U., Appl. Phys. Lett., 58:856 (1991). 30. Tsu, R., Shen, H., and Dutta, M., Appl. Phys. Lett., 60:112 (1992). 31. Babic, D., and Tsu, R., Superlattices Microstruct., 22:582 (1997). 32. Babic, D., Tsu, R., and Greene, R.F., Phys. Rev. B, 45:14150 (1992). 33. Takagahara, T., and Takeda, K., Phys. Rev. B, 46:15578 (1992). 34. Hybertsen, M.S., Phys. Rev. Lett., 72:1514 (1994). 35. Harvey, J.F., Shen, H., Lux, R.A., Dutta, M., Pannlapati, J., and Tsu, R., Mat. Res. Soc. Symp. Proc., 256:175 (1992). 36. Vial, J.C. et al. Phys. Rev. B, 45:14171 (1992). 37. Chelikowsky, J.R., Wagener, T.J., Weaver, J.H., and Jin, A., Phys. Rev. B, 40:9644 (1989). 38. Meirav, U., Kastner, M.A., and Wind, S.J., Phys. Rev. Lett., 65:771 (1991). 39. Ye, Q.Y., Tsu, R., and Nicollian, E.H., Phys. Rev. B, 44:1806 (1991). 40. Macucci, M., Hess, K., and Iafrate, G.J., Phys. Rev. B, 48:17354 (1993). 41. Iafrate, G.J., Hess, K., Krieger, J.B., and Macucci, M., Phys. Rev. B, 52:10737 (1995). 42. Zhu, J., and Tsu, R., Capacitance of few Electron dielectric Spheres, to be published in Microelectronic J. 2006. 43. Cullis, A.G., and Canham, L.T., Nature, 353:335 (1991). 44. Vasques, R.P., Fathauer, R.W., George, T., and Ksendzov, A., Appl. Phys. Lett., 60:1004 (1992). 45. Brandt, M.S., Fuchs, D.H., Stutzmann, M., Weber, J., and Cardona, M., Solid State Comm., 81:307 (1992). 46. Ito, T., Ohta, T., Arakaki, O., and Hiraki, A., Mat. Res. Soc. Symp., 256:127 (1992). 47. Feng, Z.C., and Tsu, R., (eds.), Porous Silicon, World Scientific, Singapore (1994). 48. Cullis, A.G., Canham, L.T., and Calcott, P.D.L., J. Appl. Phys., 82:909 (1997). 49. Cole, M.W., Harvey, J.F., Lux, R.A., Eckart, D.W., and Tsu, R., Appl. Phys. Lett., 60:2800 (1992). 50. Zhang, Q., and Bayliss, S.C., J. Appl. Phys., 79:1351 (1996); Bayliss, S.C., Zhang, Q., and Hutt, D.A., Phys. Stat. Sol. B, 190:85 (1995). 51. Buren, T.V., Gao, Y., Tiedje, T., Dahn, J.R., and Way, B.M., Appl. Phys. Lett., 60:3013 (1992). 52. Naudon, A., Goudeau, P., Halimaoui, A., Lambert, B., and Bomchil, G., Appl. Phys. Lett., 75:780 (1994). 53. Berbezier, I., and Halimaoui, A., J. Appl. Phys., 74:5421 (1993). 54. Sham, T.K., Jiang, D.T., Coulthard, I., Lorimer, J.W., Feng, X.H., Tan, K.H., Frigo, S.P., Rosenberg, R.A., Houghton, D.C., and Bryskiewicz, B., Nature, 367:331 (1993). 55. Pichter, A., Steiner, P., Kozlowski, F., and Lang, W., IEEE Electron Device Lett., p. 691 (1991).

14: Nanostructured Electronics, Tsu and Zhang

715

56. Koshida, N., and Koyama, H., Appl. Phys. Lett., 60:347 (1992). 57. Loni, A., Simons, A.J., Cox, T.I., Calcott, P.J.D., and Canham, L.T., Electronics Letters, 31:1288 (1995). 58. Tsybeskov, L., Moore, K.L., Hall, D.G., and Fauchet, P.M., Phys. Rev. B, 54:R8361 (1996). 59. Linros, J., and Lilic, N., Appl. Phys. Lett., 66:3048 (1995). 60. Pavesi, L., Gardini, R., and Mazzoleni, C., Solid State Com., 97:1 (1996). 61. Hirschman, K.D., Tsybeskov, L., Duttagupta, S.P., and Fauchet, P.M., Nature, 384:338 (1996). 62. Lazarouk, S., Jaguiro, P., Katsouba, S., Masini, G., La Monica, S., Maiello, G., and Ferrari, A., Appl. Phys. Lett., 68:2108 (1996). 63. Tsybeskov, L., Duttagupta, S.P., Hirschman, K.D., and Pauchet, P.M., Appl. Phys. Lett., 68:2058 (1996). 64. Filios, A., Hefner, S., and Tsu, R., JVST B, 14:3431 (1996). 65. Tsai, C., Li, K.H., Campbell, J.C., and Tasch, A., Appl. Phys. Lett., 62:2818 (1993). 66. Matsumoto, T., Hasegawa, N., Tamaki, T., Ueda, K., Futagi, T., Mimura, H., and Kanemitsu, Y., Jpn. J. Appl. Phys., 33:L35 (1994). 67. Zheng, W.H., Reece, P., Sun, B.Q., and Gal, M., Appl. Phys. Lett., 84:3519 (2004). 68. Tsu, R., Morais, J., and Bowhill, A., Mat. Res. Soc. Symp. Proc., 358:825 (1995). 69. Morais, J., Doctoral Thesis, UNICAMP, Department of Physics, Campinas, Brazil, (1995); Also in Morais, J., Lander, R., and Tsu, R., to be published. 70. Tsu, R., Filios, A., Lofgren, C., Ding, J., Zhang, Q., Morais, J., and Wang, C.G., Proc. ECS. 97–11, Quantum Confinement: Nanoscale Materials, Devices and System, (M. Cahay, ed.) Montreal (May 4–7, 1997). 71. Lu, Z.H., Lockwood, D.J., and Barlbeau, J.M., Nature, 378:258 (1995). 72. Maeda, Y., Phys. Rev. B, 51:1658 (1995). 73. Zhang, Q., Bayliss, S.C., and Hutt, D.A., Appl. Phys. Lett., 66:1977 (1995). 74. Zhang, Q., Bayliss, S.C., and Frentrup, W., Solid State Commun., 99:883 (1996); Zhang, Q., Bayliss, S.C., Al-Ajili, A., and Hutt, D.A., Nucl. Instr. and Meth. in Phys. Res. B, 97:329 (1994). 75. Hayashi, S., Kataoka, M., and Yamamoto, K., Jpn. J. Appl. Phys., 32:L247 (1993). 76. Iacona, F., Pacifici, D., Irrera, A., Miritello, M., Franzò, G., Priolo, F., Sanfilippo, D., Di Stefano, G., and Fallica, P.G., Appl. Phys. Lett., 81:3242 (2002). 77. Lee, J., Shin, J., and Park, N., J. Lightwave Technol., 23:19 (2005). 78. Ismail, K., Meyerson, B.S., and Wang, P.J., Appl. Phys. Lett., 55:973 (1991). 79. Fukuda, M., Nakagawa, K., Miyazaki, S., and Hirose, M., Appl. Phys. Lett., 70:2291 (1997). 80. Meakin, D., Stobbs, M., Stoemenis, J., and Economou, N.A., Appl. Phys. Lett., 52:1053 (1988). 81. Rabedeau, T.A., Tidswell, I.M., Pershan, P.S., Berk, J., and Freer, B.S., Appl. Phys. Lett., 59:706 (1991). 82. Ding, J., and Tsu, R., Appl. Phys. Lett., (Oct. 1997). 83. Tsu, R., Phys. Stat. Sol (a) 180:333 (2000).

716

Properties

84. Tsu, R., Mat. Res. Soc. Proc. 592:351 (2000). 85. Osman Gurdal, HRTEM unpublished. 86. Baribeau, J.M., Lockwood, D.J., and Lu, Z.H., Mater. Res. Soc. Symp. Proc., Vol. 382, Pittsburgh (1995). 87. Tsu, R., SPIE J., 1361:313 (1990). 88. Ye, Q.Y., Tsu, R., and Nicollian, E.H., Phys. Rev. B., 44:1806. 89. Boeringer, D.W., and Tsu, R., Phys. Rev. B, 51:13337 (1995). 90. Tsu, R., Li, X.L., and Nicollian, E.H., Appl. Phys. Lett., 65:842 (1994). 91. Ding, C., and Tsu, R., Mat. Res. Soc. Symp. Proc., 378:757 (1995). 92. Capasso, F., Mohammed, K., and Cho, A.Y., IEEE Quantum Electronics, QE 22:1853 (1986). 93. Lloyd, S., Scientific Am., Vol. 140 (1995). 94. Tsu, R., Dovidenko, K., and Lofgren, J.C., ECS Proc., 99–22:294 (1999). 95. For a good recent review, the readers are referred to “Semiconductor Quantum Devices”, Marc Cahay, in: Adv. Electronics and Electron Phys., 89:93 (1994). 96. Reed, M.A., Randall, J.N., Aggarwal, R.J., Matyi, R.J., Moore, T.M., and Wetsel, A.E., Phys. Rev. Lett., 60:535 (1988). 97. Kastner, M., Phys. Today, Vol. 24 (Jan. 1993). 98. Zeller, T.R., and Giaever, I., Phys. Rev., 20:1504 (1968). 99. Chou, S.Y., and Wang, Y., Appl. Phys. Lett., 61:1591 (1992). 100. Tiwari, S., Rana, F., Hanafi, H., Hartstein, A., Crabbe, E., and Chan, K., Appl. Phys. Lett., 68:1377 (1996). 101. Yano, K., Ishii, T., Hashimoto, T., Kobayashi, T., Murai, F., and Seki, K., IEEE Trans. Electron Devices, 41:1628 (1994). 102. Guo, L., Leobandung, E., and Chou, S.Y., Appl. Phys. Lett., 70:850 (1997). 103. Covin, V.L., Schlamp, M.C., and Alivisatos, A.P., Nature Vol. 370:354 (1994). 104. Schmidt, H., Medeiros-Ribiero, G., Oestreich, M., Petroff, P.M., and Dohler, G.H., Phys. Rev. B, 54:11346 (1996). 105. Markovich, G., Leff, D.V., Chung, S.W., Soyez, H.M., Bunn, B., and Heath, J.R., Appl. Phys. Lett., 70:3107 (1997). 106. Kiehi, R.A., and Ohshima, T., Appl. Phys. Lett., 67:2494 (1995). 107. Ohshima, T., and Kiehl, R.A., J. Appl. Phys., 80:912 (1996). 108. Ohshima, T., Appl. Phys. Lett., 69:4059 (1996). 109. Fukuda, M., Nakagawa, K., Miyazaki, S., and Hirose, M., Appl. Phys. Lett., 70:2291(1997). 110. Scheible, D.V., and Blick, R.H., App. Phys. Lett., 84:4632 (2004). 111. Leonard, D., Pond, K., and Petroff, P.M., Phys. Rev. B, 50:11687 (1994). 112. Saito, H., Nishi, K., Ogura, I., Sugou, S., and Sugimoto, Y., Appl. Phys. Lett., 69:3140 (1996). 113. Huffaker, D.L., Baklenov, O., Graham, L.A., Streetman, B.G., and Deppe, D.G., Appl. Phys. Lett., 70:2356 (1997). 114. Narukawa, Y., Kawakami, Y., Funato, M., Fujita, S.Z., Fujita, S.G., and Nakamura, S., Appl. Phys. Lett., 70:981 (1997). 115. Tabuchi, M., Noda, S., and Sasaki, A., in Science and Technology of Mesoscopic Structures, (S. Namba, C., Hamaguchi, and T., Ando, eds.), Springer, Tokyo (1992).

14: Nanostructured Electronics, Tsu and Zhang

717

116. Solomon, G.S., Larson, M.C., and Harris, J.S., Jr., Appl. Phys. Lett., 69:1897 (1996). 117. Fathpour, S., Mi, Z., and Bhattacharya, P., J. Phys. D: Appl. Phys., 38:2103 (2005). 118. Park, J., Pasupathy, A.N., Goldsmith, J.I., Chang, C., Yaish, Y., Petta, J.R., Rinkoski, M., Sethna, J.P., Abruna, H.D., McEuen, P.L., and Ralph, D.C., Nature, 417:722 (2002). 119. Tsu, R., Microelectronics J., 34:329 (2003). 120. Kang, K., Daneshvar, K., Tsu, R., Microelectronic J., 35:629 (2004).

Index

Note: page numbers in italics refer to figures A ab initio quantum mechanical calculations 296 abnormal grain growth instability 179, 275, 345, 647 absorption coefficient 682 absorption desorption cycle 403–5, 410, 411 absorption plateau 407 acid/base reaction 4 acrylonitrile-butadiene-styrene, ABS 128 action of deformation 125 activation cycle 408–10 energy 143, 181–2, 184, 196, 197, 199, 203, 272, 273, 337, 351, 352, 355, 356, 389, 575, 684, 696 low 180–2, 186, 197–8, 351, 352, 355, 356, 389, 575, 684, 696 values 181, 182, 184, 197 enthalpy 343, 345, 348 process 376, 404, 405, 409, 410, 415 active site electronic state 366–8 additional heat treatment 663 adhesion 99 force of 180 adsorbates 176, 185, 190, 214 adsorbed oxygen 387, 389, 391, 393, 703, 703, 704 equilibrium concentration 387 layer 394, 697 monolayer 677, 678, 702 epitaxy and defect free 702 adsorption, activated carbon 402 AFM see atomic force microscopy

Ag nanocrystal layer 706 Ag-Fe 125, 572 Ag2O/C mixtures 128 agglomerated powders 8, 192, 194, 195, 198, 208, 216 agglomerates 6–8, 10, 25, 55, 63, 66, 98, 109, 110, 125, 188, 191–4, 334, 340, 354 agglomeration 6–9, 21, 64, 66, 78, 83, 174, 179, 185, 187, 194, 200, 203, 246, 551 and dispersion 6–9 of particles 6, 7, 9, 15, 26, 28, 79, 81, 108 agglomerations 549 air-sensitive chemicals 14 Al alloys 70, 126, 129, 145, 158, 159, 190, 208, 210, 211, 213, 215 fcc-Al 583, 599 matrix 584, 598, 600, 619, 632 nano-grains 587, 618, 630, 631 phase 584–6, 618, 621, 632 particles 82, 581, 582, 584, 630, 631 peaks 126 phases 582, 585, 621 Al-based alloys 582, 590, 598–9, 603, 633, 642 amorphization 620 amorphous 617–18, 619 bulk 598, 634 containing quasicrystalline particles 598–9 high-strength 581, 630 mechanical alloying 617 nanostructured 581–2 two-phase 629–30

719

720 Al-based alloys (cont’d) six types 582 systems 581 Al-metalloid systems 581 Al-rich alloys 585, 619 amorphous 583, 618–19 Al2O3 powders 59, 203, 208 non-agglomerated 195 Al88Ni9Ce2Fe1 alloy 630 Al91Mn7Fe2 squeeze-cast alloys 634 Al92Mn6Ce2 squeeze-cast alloys 634 all-optical logic gates 692 alloys composition 15, 250, 160, 492, 504, 575, 580, 581, 584–5, 588, 592, 594, 596, 605, 607, 609, 611, 612, 614–16, 623, 625, 630, 631, 633, 634, 638, 652, 653, 653, 657 equiaxed 239 mechanical milling 129, 371, 405, 553 with quasicrystalline phases 632–4 single-step process 634 systems high tensile strength 587 AlN phase normal grain growth 22 AlNxBN100−x 22 AlO4 23 alternate crystal structures 176–7 alternative processing techniques 598 alumina 21, 55, 91, 92, 95, 98, 99, 100, 101, 112, 179, 200, 202, 203, 206, 212, 215, 374, 464, 465–7, 470 alumina–titania coating 98–9 aluminum alloys 321 different strengthening mechanisms 630 high-strength 629 superplastic 639 ambient milling temperature 571, 618 ammonia detection improved sensitivity 390

Index amorphization 120, 126, 144, 151, 152, 159, 160, 180, 181, 307, 587, 618–21, 624 amorphous bonding 700 Cu 32 layer 407, 700 materials crystallized 251, 351, 487, 489, 493, 577, 582 metals 540 microstructure nanogranular 488, 489, 496, 497, 526, 582, 587, 631 precursor 340, 513 crystallization 270, 378, 487, 598 powders 20 ribbon 513, 630, 631 silicon photovoltaic materials 700 superlattice 677, 699–700, 574, 575, 576 structure 126, 152, 210, 212, 252, 256, 489, 508, 509 thin films 410 TiO2 particles 16, 20 amorphous/Al particle interface 631 amorphous alloys 130, 348, 351, 372, 379, 409, 488, 489, 493, 502, 519, 533, 575, 583, 588, 590, 592, 595, 600, 618, 630, 631, 637, 638, 645 Al—Ni—Si Al-based material 581, 583, 590 containing nanoscale particles, strength of 637–8 deformation behavior 372, 629, 635–6 devitrification 593 Fe—B—Si 347, 487, 531 amorphous/amorphous interface crystals heterogeneous nucleation 579 amorphous-crystalline phase transition 694

Index amorphous/nanocrystalline phases 663 amorphous phase 71, 120, 315, 316, 340, 351, 395, 487–9, 492–7, 499–501, 508–510, 512–15, 517, 518, 520–2, 524–7, 534, 571, 574, 581–4, 587–91, 596, 600–2, 604, 618, 619, 621–3, 627, 631, 637, 638, 641, 643, 663, 693, 700 broad diffraction maxima 623 crystallization 270, 348, 381, 526, 575–9, 590 formation 159 intergranular 332 maximum Fe concentration 488, 489 metastable 177, 492, 576 network 582, 587, 631 partial devitrification 590, 630, 663 separation 577–9 single 510, 525, 578, 583 thermal stability 492, 497, 499, 618 amorphous phase/supercooled liquid devitrification 663 amorphous powders consolidated bulk specimens from 650–1 amorphous/quasicrystalline phases 663 amorphous silicon/oxide superlattice 699–700 amorphous-Si/SiO2 superlattice 699 amorphous single-phase alloys 583 angle boundaries 146, 147, 149, 150, 151, 236, 343, 347 anhydrous γ-Al2O3 176 anion vacancies 370 anisotropic intensity distributions 131 surface energies 175, 179 anisotropy 15, 31, 32, 179, 181, 216, 243, 244, 281, 348, 441–7, 450, 452, 462, 463, 466, 467, 505, 514, 516, 527, 528, 539 problem 175

721 annealing 21, 33, 141, 142, 159, 215, 235, 246, 262, 262, 270, 304–6, 338, 339, 341, 342, 344, 347, 348, 370, 376–9, 383, 395–7, 407, 408, 466, 470, 492, 494, 500, 500, 505–7, 514, 526, 527, 528, 529, 557, 574, 579–81, 583, 584, 590, 593–8, 601, 603–6, 619, 626, 631, 634, 642, 690, 693, 694, 696, 699 strengthening effect 552 time 140, 274, 573, 574 anodization processes 689 anomalous small-angle x-ray scattering (ASAXS) 580, 601 antiferromagnetic oxide layers 248 applications catalytic 23, 372, 379 commercial 23, 108, 277, 560 potentials 350 APS see atmospheric plasma spraying aqueous borohydride 9 chemistry 5 electrodeposition 30–2 electroless deposition 31, 32 electroplating 30–2 media 8, 9 organic 8 methods 9–10 reactions 4 Archimedes’ principle 200 Arrhenius plots 181, 341, 346, 350, 353 as-cast specimen 609 x-ray diffraction maxima 606 as-deposited Fe-(Hf, Zr, Re)-O films soft magnetic properties 527 as-deposited Fe-Hf-O films sputtered 524 as-deposited Fe-M-O films magnetic properties 527 as-grown wafer reflectivity spectrum 709

722 as-milled amorphous powders thermal stability investigations 654 as-quenched amorphous ribbons 660 as-sputtered structure 520–5 as-synthesized AlN powders 16, 21 as-synthesized particles 7, 15, 19, 29 ASAXS see anomalous small-angle x-ray scattering Ashby model 208 ASTM B-117 salt spray test 266 ASTM G28 267 ASTM G35 267 ASTM G44 267 ASTM G48 267 ASTM G65 111 atmospheric contamination 122 atmospheric plasma spraying (APS) 95, 103, 106–11, 107, 109, 110 atom-probe field ion microscopic data 492 atom-probe field ion microscopy technique 497 atomic bonding 165 diffusion 331, 332, 341 force 175 homogeneity 21 interactions 302, 321, 323 mechanisms 174, 179, 216 ordering processes 348 plane-doped superlattices 678 rearrangement 312 scale mixing 4, 125 structure 3, 119, 142, 164, 165, 376 transport 331, 342 atomic force microscopy (AFM) 175, 462, 710 study 175 atomic level simulations 293, 294, 537, 560 atomic potential energies and forces 296–7 atomic-scale mixing 125 atomistic simulations 293, 335 atomization 235, 568, 569, 586, 590, 591, 598, 599, 633 techniques 568, 569

Index atomized powders 586, 590–2, 599, 635, 636, 650, 663 amorphous extruding 583, 584 consolidation 618 autocatalytic electroless process 31 avalanche multiplication 700, 701 B backscattered electron images 146, 613, 614 ball milling 48, 122, 129, 135, 137, 151, 157, 250, 409, 411, 416, 555, 556, 557, 558, 560, 620 device 163 ball mills 120, 121, 569 high-energy 121 ball-milling process 120, 123, 126, 128, 129, 131, 133, 136, 570 high-energy 48 ball-powder-collision 570 bamboo grain structure 255 bandwidth light emission 708 Beilby layer 161 BET (Brunauer, Emmett, Teller) nitrogen adsorption 201 bimetal composition 201, 369 bimetallic clusters 369 bimodal pore distribution 192, 194 binary alloys 64, 126, 159 binary Fe-based alloys 489 biomedical applications 458–61 Bohr model 680 Boltzmann factor 63, 295, 296, 298, 441 bond strength 91, 93–5, 103, 107, 108, 113–15, 369 bonds and coordinations of Si nanostructure 688–91 boundary mobility 197, 199, 200, 307, 344 Bragg peaks 130, 558, 559, 559 brightfield electron micrograph 606 brittle ceramics 120 broad exothermic reaction 135

Index Brownian motion 6, 7, 63, 65, 456 bulk glass formation key parameter 592 I-phase Al-based alloys 584–6, 599, 620, 632, 633, 633, 634 metallic glasses 569, 580, 600–1, 607, 641, 664 crystallization 580 multiphase material 599 bulk alloys 591, 601–7, 633, 647 Al-based 634 glass forming 566, 568 phase separation in 579–80 nanocrystalline 631–2 nanostructured 566 bulk ferromagnetism 441–2 bulk materials extreme deformation 143 nanocrystalline 52, 157, 158, 176, 214 nanostructured 78, 159, 566, 568, 580, 601, 603, 610, 622, 626 bulk metallic glasses Mg-based nanostructured alloys from 600–1 bulk nanocrystalline alloys 584–6 bulk nanostructured samples by extrusion 590–1 bulk samples 129, 140, 161, 321, 340, 580, 592, 593, 599, 600, 601, 618, 634, 645, 659, 660 amorphous 637, 640 extruded 590–1, 650 residual porosity 630, 660 slowly cooled 617, 625, 626, 635 bulk specimens amorphous Mg-based 636 as-quenched 579 composite 659 consolidated from amorphous powders 650–1 dense 659, 663

723 buta-1,3-diene hydrogenation 378 C C3H8 401 Ca- (CSZ) oxygen diffusivities 354 cadmium acetate hydrate 17 Cahn’s theory spinodal decomposition 579 calorimetric investigations 126 Canadian CANDU 246 capacitance 384, 677, 685–7, 704, 707, 708, 712 devices 705, 706 in nanoparticle 685–7 carbide particles 97 carrier concentration variation 387 cassiterite 395 cast material 634, 635, 641 specimens devitrification 601 cast alloys microstructure 634 cast bulk composites 657 glasses thermal stability data 601 samples partially crystallized 641, 650 specimens deformation behavior 654 casting processes limitations 592 catalysis 9, 265, 365, 372, 374 and electrocatalysis 366–7 impact of structure on 367–70 catalyst 4, 20, 23, 27, 31, 32, 176, 279, 309, 365–71, 374–83, 409, 411, 413, 415, 417 catalytic activity 23, 367–70, 374–6, 382 CO oxidation study 376 dopants 390 hydrogenation 373 reaction 23, 367, 369, 383 selectivity 376

724 cation exchange 19 cation self-diffusion 353, 355, 356 cationic dopants 352 centrifugation method 188 Ceracon consolidation 216 ceramics 4, 16–23, 32–4, 48, 51, 57, 60, 78, 92, 94, 112, 135, 161, 173, 175–7, 182, 191–3, 197–200, 210–15, 309, 315, 337, 355–8 compaction 187 composites sintering behavior 205 densification HIP 211 dopants 199 fully dense 203 nanocomposites 21, 120, 122 nanocrystalline 331, 332, 345, 352, 355, 356 nanoparticles 16, 26, 180, 182, 214 sintering 176 nanopowders 186, 187, 194, 198, 202, 203 sinterability 180, 185, 203 particles 54, 120, 569, 617 phase mixtures 120 powder 128 sintering 202–5 cerium oxide 370, 371 chamber size and growth distance 75–6 changes in transport properties 384 charge transfer 235, 237, 296 interaction 410 chemical diffusion 343 chemical potential measurements of 351 chemical precipitation 371, 375 chemical reaction 57, 570 chemical synthesis material diffusion 3, 14 of nanostructured particles 3–30 chemical vapor deposition (CVD) 57, 235, 392, 696, 707, 708 plasma-enhanced 396 chemisorption 387, 389, 393, 404 chill methods 568

Index chill-mold cavities 568 classical scaling rule 183 closed-loop processing 573, 590 cluster-compacted metals 344, 351 clusters 21, 23, 51, 54, 55, 58–60, 62, 64, 65, 67, 70, 71, 73, 74, 76, 78–83, 122, 294, 299–302, 304, 318, 319, 321, 341, 369, 371, 375, 381, 439, 449, 451–3, 548, 558, 688–91, 695, 712 deposition 344 luminescence from 694–5 magnetic 442–3 model 343 of oxygen 703 size and distribution 367 size control mechanisms 694 Co 238 catalytic hydrogenation 373 effect of 615–16 peaks 126 selectivity 391 variations in sensitivity 396, 397 Co-Fe-Hf-O films 528 typical application items and characteristics 531 Co-W 238 CO2 hydrogenation 374 coagulation 7, 63, 65, 70 coalescence 51, 54, 62, 63, 66, 71, 271, 275, 309, 310, 315, 320, 324, 394 coarse-grain 21 coarsening 173, 174, 179, 185, 186, 193–9, 203, 206, 208, 212, 215, 216, 262, 281, 320, 509, 551, 572–4, 598, 618, 633, 663 coatings 29–34, 60, 91–9, 103–5, 107–12, 114–16, 187, 236, 252, 264, 276, 279, 457, 632 applications 91 characteristics 92

Index coatings (cont’d) porosity 94 quality 193 cobalt acetate tetrahydrate 11 nickel alloys 12 cobalt phthalocyanine (CoPc) 369 pyrolysis 377 Coble creep 254, 540 cold compaction 140, 174, 186–91 of metals 188–90 of nanocrystalline ceramics 191–3 cold isostatic pressing 189, 190–2 cold rolling 125, 136, 143, 159–61, 256 cold weld 120, 408 colloidal particles reductive synthesis 376 colloids 4, 6, 14, 19, 25, 26, 28, 31, 375 common synthesis routes 567 compaction 52–4, 53, 61, 62, 79, 80, 81, 82, 122, 129, 140, 174, 186–92, 188, 189 techniques 572 compatibility with electrolyte 237 compositional metastabilities 176–7 compounds 69 compression testing 556 computer modeling of nanostructured materials 292–324 methods 295–9 computer simulation 125, 142, 179, 309, 335, 537, 550, 552 of mechanical behavior of nanocrystalline materials 538–47 molecular dynamics 543–7 condensation 5, 19, 20, 22, 23, 27, 34, 47–51, 53, 54, 55, 57, 60, 62, 63, 68, 72, 73, 73, 75, 76, 80–3, 82, 122, 126, 176, 178, 185, 186, 189, 190, 201, 235, 241, 248–51, 254, 269, 332, 339, 341, 342, 344, 352, 354, 370, 372, 378, 396, 410, 548, 549

725 conduction electron 389, 454, 471, 472 consolidation of bulk specimens from amorphous powders 650–1 elevated temperatures 598 methods 173–217 of powders 572–4 techniques 573, 618 constraint condition effects 700 contamination degree of 185 continuous pulse operation 692 conventional casting techniques 598 conventional deformation 136 conventional processing 136 conventional sintering 181, 202–6 conventional solidification and devitrification of bulk samples 598–616 coordinations and bonds of Si nanostructure 688–91 copper Fermi level 269 foil 278 mold casting 568, 598 corrosion 93, 95, 96, 98, 102, 103, 105, 108, 113, 114, 236, 246, 247, 265–8, 266, 280, 331, 334, 450 properties 265–8 resistance 103, 120, 265, 267, 268, 278, 279, 569, 587, 615 and wear resistance 276 Cottrell theory 538 coulomb blockade 700, 706, 711 interaction 683, 685–7 staircase 706 coumarin 238 Cr, effect of 615–16 Cr3C2—NiCr cermet 97 crack formation 106, 144, 257, 640, 660 generation 187, 631 propagation 108, 110, 315

726 creep 182, 185, 208, 209, 254, 256, 334, 540–2, 555, 556, 644–6 rate 645 cryomilled powders 157 crystal (cluster) size 58, 59, 65, 74, 78, 80, 300–2, 367, 443, 694 fcc-crystal 15, 143, 596 growth 18, 574, 696, 700 and nucleation 574–5 rate 579 lattice 47, 135, 379 defects 120, 566 nucleation 574–5, 580, 592 structure 6, 123, 176, 177, 354, 442, 492 and morphology 65–7 volume diffusion 352–6 bcc-crystal structures 66, 67, 123, 377, 487, 489 crystal-to-amorphous transition 571 crystal-to-quasicrystal transition 571 crystal/glass interfacial energy decrease 575, 579, 631 crystalline Al alloys 630, 632–4 alloys 261, 340, 349, 350, 351, 402, 492, 509, 514, 584, 587, 617, 619, 630, 638 cores 689–91 layer 162, 700 metals or alloys mechanical strength 566, 581, 587, 641, 642, 660 phases 23, 126, 316, 378, 408, 489, 518, 566, 567, 571, 572, 576, 577, 580, 581, 586, 592, 594, 596, 600, 601, 602, 611, 618, 620, 621, 623, 626, 631, 632, 635, 663, 693, 694 structure 145, 151, 152, 155, 161, 163, 205, 214, 215, 266, 306, 332, 333, 345, 346, 367, 378, 487–9, 497, 502, 577, 582, 629, 631

Index volume fraction 242, 254, 256, 269, 642 crystalline metal electrodeposits 237, 239–46, 270 crystallite 22, 28, 31, 34, 48, 60, 61, 63, 82, 144, 159, 236, 300, 306, 315, 334, 340–53, 367, 368, 399, 605, 606, 606, 643, 645, 682, 684, 688, 700, 702, 704 condensation and compaction 338, 339, 341, 342, 344, 354 diameter 333, 388, 389 edges or corners 366 growth 331, 333, 338, 348, 349 and diffusion, correlation between 343–5 interfaces 331, 333, 342, 343, 346, 347, 351, 352, 354 size 16, 64, 66, 67, 132, 201, 338, 343, 344, 346, 354, 355, 366–9, 373, 374, 376, 377, 379, 384, 388, 389, 392, 395–7, 397, 405, 409, 412, 417, 596, 597 volume fraction 605, 606 crystallization 5, 16, 22, 25, 33, 126, 130, 136, 142, 159, 190, 237–9, 239, 243, 250, 270, 304, 306, 331, 332, 338, 340, 346–8, 346, 351, 372, 378, 381, 394, 412, 414, 487–99, 504, 514, 526, 526, 533, 571, 572, 574–9, 583, 588, 590, 592–6, 598, 600, 604–7, 617, 619, 621, 623–8, 631, 638, 640, 647, 655, 656, 660, 663, 700 behavior 500, 525, 580, 593, 602, 651 eutectic 576–7 -induced nanostructure 488–92 metastable phases oxygen-triggered formation 626 new studies 702 of amorphous alloys 348, 575 partial 637, 641, 650, 662 polymorphous 576

Index crystallization (cont’d) primary 577 significant incubation times 579 stepwise behavior 593, 596, 602, 607, 651 see also devitrification CsCl compounds 124 Cu alloys 466, 502, 507–10, 515, 516, 518, 519, 531, 622, 633 arrays 307 peaks 126 Cu—Bi alloys 64 Cu—Fe 125 Cu-free quaternary Fe—Zr—Nb—B alloys soft magnetic properties and structure 503–7 Cu—Ta 125 Cu—V 125 Cu—W 125 CuO/Ca mixtures 128 Curie temperature 32, 441, 449, 451, 454, 455, 499, 500 effect on soft magnetic properties 513–20 current density 30, 239, 244, 265, 268, 277, 283, 377, 471 CVD see chemical vapor deposition cyclic deformation 143 cytotoxicity of nanoparticles 28–30 D data storage media 462–7 DBRs see distributed Bragg reflectors DC operation 692 De Broglie waves 703 deagglomerated powders 8 deagglomeration 8, 25, 195 Debye screening length 705 decarburization process 94 defect formation 316, 318 deform 120, 134, 215, 308 deformation behavior of amorphous alloys 635–6 at elevated temperatures 643–6 of nanostructured alloys 638–9 curves 645

727 mechanisms 558–60 mode 144, 315, 540, 541 process 119, 122, 123, 135, 137, 143, 154, 163, 165, 260, 537, 543, 544, 547, 553 deformation-induced heating 650 dense bulk compacts 591 densification 21, 33, 98, 173–5, 178–82, 184, 186, 188, 189, 193, 195, 199–201, 205–17, 343, 354, 356, 572, 573 behavior, effects of pore size on 193–6 full 200–16 heating rates 179, 203 measurements 200–1 mechanism 208, 210 nanocomposite 205–6 of nanoparticles 179, 180 of nanopowders 200–16 process 98, 174, 178, 186, 203 rate 178, 196, 198, 206, 208 stress-assisted 208 density improvements 80, 207, 573 measurements 200, 201, 242, 548 reproducibility values 201 density-pressure plot 181 deposit structure changes 237 basis oriented and reproduction type (BR) 237 field oriented-type (FT) 237 twin transition types (TT) 237 unoriented dispersion type (UD) 237 deposition techniques highly controlled 677 design, nanostructure 370–83 device lifetime modest improvements 692 devitrification 49, 568, 574, 581, 582, 584, 593, 598, 602, 626 of bulk samples 598–616 of metallic glasses 574–80, 663 partial or complete 580, 590, 601, 622, 630, 663 stepwise behavior 602 see also crystallization

728 die casting 568, 598 high pressure 600, 636 die method 568 dielectric constant of nanoscale silicon 678–80 discontinuity 680, 681, 685 function 679 mismatch 682, 684, 685, 687 properties 200 spacer layer 706 dielectric constant 214, 387, 390, 677–80, 682, 683, 687 several values 681, 681, 683, 683 size dependent 679–81, 680 differential scanning calorimetry (DSC) 135, 141, 142, 143, 156, 157, 159, 271, 271, 276, 526, 585, 588, 590, 594, 599, 602, 603, 604, 605, 619, 623, 623, 624, 627, 640, 658 curves 157, 605 exothermic peak 593, 603 heating experiment 140 measurements 142, 618, 627 scan 272, 274, 591, 593, 598, 599, 603, 624, 628, 640, 659, 660 thermal stability data 591, 599, 601, 618, 627, 629 traces 588, 590, 594, 602–4, 603, 604, 619, 623, 623, 624, 627, 628 diffusion 3, 6, 14, 20, 24, 27, 51, 52, 63, 68, 70, 72, 76, 79, 124–6, 143, 178–82, 184, 186, 193, 194, 197–9, 203, 205, 206, 208, 209, 212, 237, 238, 239, 256, 272, 298, 300, 302, 307, 309, 310, 314, 317, 320, 323, 333–54, 356, 395, 399, 404, 409, 412, 414, 461, 468, 471, 541, 544, 572, 575, 577, 696, 699 activation energy 355 annealing 338, 342

Index atomic 331, 332, 341 chemical 343 and chemical redistribution 605 coefficients 4, 217, 332, 334, 339, 350, 353, 353, 354 and crystallite growth, correlation between 343–5 59 Fe 341, 342, 349, 350 59 Fe-tracer 346, 346 Fe-tracer 346, 351 grain boundary 178, 182, 184, 197, 198, 208, 272, 307, 320, 341, 343, 349, 356, 541 of metals 334–5 of hydrogen 404 in nanocrystalline metals 350–1 interface 331, 342, 343, 348, 349, 352–4 modeling 332–4 mechanisms 184, 256 nanocrystalline ceramics 352–6 nanocrystalline materials 331–56 nanocrystalline metals 332, 335–51 oxygen 352, 353, 354, 355, 699 tracer 339, 340, 346, 348, 349, 351, 353 volume 182, 184, 193, 197, 333, 352–6 diffusion studies 332, 335, 336, 339–43, 346, 349, 352 Fe-tracer 340 nanocrystalline metals 332 nanocrystalline ZrO2 352, 354 diffusion-controlled processes 332, 342 diffusivities 178, 334, 335, 337, 339–43, 346, 348, 349, 351, 353–6 digital circuitry 706 dipole stress fringes 180 direct coal liquefaction 368 direct condensation reactions 22 direct current plating 239, 267 discrete quantum states 700 dislocation densities 136, 143, 314 motion 180, 181, 186, 216, 308, 311, 312, 630, 632, 661

Index dislocations 124, 130, 131, 133, 134, 138, 141–4, 147–9, 151, 153–5, 180, 261, 307, 310–15, 317, 318, 323, 405, 537–9, 542–4,545, 547, 557, 558, 560, 571, 584, 631, 632, 642, 696 dispersed compound particles density 584 dispersed nanocrystalline metal particles 27 dispersion and agglomeration 6–9 strengthening 629, 632 distributed Bragg reflectors (DBRs) 692, 708–10 DLVO theory 6 DMC method 215 donor binding energy 681, 682 dopants 194, 199–202, 352, 371, 390, 391, 396, 413, 414, 711 binding energy 679–82 solid solubility limit 712 doping 33, 200, 345, 390, 399–401, 413, 415, 677–82, 711, 712 of nanoparticle 680–2 quantum size effect 680 dry compaction 188, 188, 191 DSC see differential scanning calorimetry ductility 95, 96, 130, 135, 174, 244, 247, 253, 255–9, 264, 278, 331, 553, 556, 557, 560, 565, 566, 571, 583, 585, 630, 632–8, 641–3, 647, 648, 650, 651, 653, 654, 663, 664 good bending 581, 635, 636 of nanocrystalline metals 556–8 E early sintering stages orientation 174 early transition metal (ETM) 581, 583, 630 edge dislocations 143 EDS see epilayer doping superlattice EDX see energy-dispersive x-ray EELS 496, 496

729 Einstein equation 655, 656 Einstein relation 658 EL see electroluminescent elastic anisotropy 514, 539 behavior 553, 642 constants 321 moduli 548, 553, 654 properties 252, 281, 298, 553, 555 shear strains 147, 573 stresses inhomogeneous 144 electrical arc discharge 59 electrical conductivity 30, 31, 244, 379, 468 electrical resistivity 247, 253, 255, 269–70, 280, 399, 488, 518, 519, 521, 527 electrical transport properties effect of grain size 269 electrocatalysis 365, 372, 377 and catalysis 366–7 impact of structure on 367–70 electrocatalytic activity 372, 373, 379 oxygen reduction reaction 368 reaction 372 electrochemical etching 681, 689 process 377 electrocrystallization 237–9, 239 electrodeposited Cu 258, 269, 279, 558 electrodeposited metals 270 electrodeposited nanocrystalline materials 235–7, 246, 250, 253, 260–2, 264, 278 electrodeposited nanocrystalline metals, alloys, and composites 235–83 applications 276–83 properties 246–76 structure 239–46 electrodeposited nanocrystalline Pd large free volumes 241 electrodeposited nanocrystals 240, 241, 251, 257, 279 porosity-free 242

730 electrodeposited Ni 241, 250–2, 254, 257, 269, 270, 273, 549, 556–8 electrodeposition 30–2, 48, 49, 235–40, 243–5, 249–50, 254, 260, 262, 269, 276, 277, 281, 282, 283, 338, 345, 351, 377, 401, 464, 465, 466, 548, 553, 557 parameters 237, 240, 243 synthesis of nanostructured materials by 237–9 technique 557 electrodeposits 236–9, 241–3, 245, 247, 248, 250–2, 254, 256, 257, 260, 261, 263–77, 281, 283 structure of 239–46 electroforming 277, 281 electroluminescent (EL) 681, 690–2, 694, 695, 697, 699, 699 devices 697 diodes 681, 697 spectrum 691, 694, 699 electro-magnetic interference (EMI) 530 electromagnetic waves 532 electron beam evaporation 53, 56, 694 lithography 383, 463, 464, 705 electron microscope 52, 130, 540 electron microscopy 61, 66, 130, 175, 257, 275, 457, 491, 512, 513, 516, 548, 585, 599, 607, 612 high-resolution 161, 175, 241, 257, 300, 450, 495, 496, 523, 688 technique 201 electron mobility 389, 455, 718 electron scattering 269, 270, 280 electron spin resonance (ESR) 387 electron-hole polarization 683 electron-hole wave functions 683 electron-phonon interaction 680 electronic applications 48, 236, 276, 682, 697, 702, 712 of Si/O superlattices 702–3

Index electronic devices 278, 520, 678 electronic structure calculations 249 electrons 31, 59, 235, 296, 385, 387, 408, 454, 467, 468, 468, 469, 470, 471, 472, 678, 680, 685–7, 700, 703, 704, 708, 711, 712 tunneling 454, 684, 706, 707 electroplating 30, 31, 240, 245, 246, 275, 277, 283 ElectrosleeveTM 267, 279 technology 246, 261, 273, 278 electrostatic repulsion 6, 8, 25 electrostatic stabilization 8 electrostatic terms 683 elemental particles 10 elemental powder mixtures 571, 621 mechanical alloying 617, 621 elemental sheets 159 elevated temperatures deformation behavior 643–6 superplastic flow 629 elliptical crystallites 606 embedded atom method potentials 251 EMI see electro-magnetic interference energy input 571 energy-dispersive x-ray (EDX) 275, 495, 495, 496, 496, 521, 523, 618, 710 analysis 130, 652 enhanced catalytic activity 565 enhanced magnetization 32 enthalpy release 140, 141, 157, 271 entropy 8, 125, 456 epilayer doping superlattice (EDS) 107, 609, 610, 678 epitaxial growth 696, 697 epitaxial structure 696, 697 epitaxy 696, 697, 698, 702 equiaxed grain structure 66 equilibrium compounds 594, 599, 604, 607 configuration 140, 297 intermetallic phase mixture 128 phase 568, 575, 576, 581, 596 solubility limit 120, 124, 569, 570 thermodynamics 125

Index ESR see electron spin resonance etching 11, 105, 278, 280, 281, 465, 681, 689, 692, 705 ethyl phthalate 201 ETM see early transition metal eutectic crystallization 576 eutectic nanocrystallization 577 evaporation effect of source position 76 rate 74–5 techniques 56–7 evaporation-condensation method 178 EXAFS see extended x-ray absorption fine structure excess resistivity 269 exchange interactions between nanoparticles 453–5 excitation photoluminescence (PLE) 688, 690 exciton binding and recombination energies 682–4 excitonic energy 682 excitons 682 recombination and binding energies 682 exothermic crystallization 159 event 588, 606 peak 593, 627 exothermic event 602, 603, 607 exothermic peak 489, 593, 603 exothermic reaction 135, 489 extended grain boundary 241 extended plasticity 644 extended solid solutions 126 extended x-ray absorption fine structure (EXAFS) 688–91, 690 analyses 694 characterization 694 luminescence 689 extrinsic doping 679 extrusion 26, 136, 207, 210, 212, 573, 583, 626, 633, 639, 640 bulk alloys 631 bulk nanostructured samples 590–1 conditions 650

731 higher temperatures 584, 590, 591, 638, 650 F fabrication technique 702 fatigue response 261 Fe concentration 487, 488, 489, 492 diffusivity 348 effect of 615–16 α-Fe 163 phase 489 α-Fe2O3/Ti mixtures 128 bcc-Fe 489 59 Fe diffusion 341 59 Fe-tracer diffusion 346 powder samples 131, 138 ball-milled 131 nanocrystalline 138 tracer diffusion 340, 351 Fe(CO)5 thermal decomposition 15 Fe—(Hf,Zr,Re)—O films soft magnetic properties 527 Fe-based alloys 340 amorphous 586 α-Fe—C-powder 163 Fe—Co alloys 70, 238, 262 Fe-contamination 121 Fe—Cu nanopowders 208 powder mixtures mechanical attrition 125 Fe—Hf—O films two-stage crystallization behavior 525 Fe—M—B alloys and crystallization-induced nanostructure 488–92 soft magnetic properties 492–9 Fe—M—O films typical application items and characteristics 531–2, 532 Fe—Nb—B alloys 489 Fe–Nb–B–P–Cu alloys produced in air soft magnetic properties effect of grain-size distribution 513–20 and structure 507–13

732 Fe-surfaces nanocrystalline 161 Fe—Zr—B alloys 492 Fe/Ag nano-multilayers 159 Fe : Ti metallic ratio 407 Fe2O3-type crystals lattice parameter 395 Fe3Al powders green density 190 Fe73.5Si13.5B9Nb3Cu1 soft-magnetic alloy 345–9 bcc-Fe84−xNb7B9Cux function of Cu content 502 bcc-Fe90−xZr7B3Cux function of Cu content 502 Fe90Hf10 amorphous alloy crystallized structure 489 Fe90Nd10 amorphous alloy crystallized structure 489 Fe90Zr7B3 soft-magnetic alloy 345–9 Fe90Zr10 amorphous alloy crystallized structure 489 Fermi level 269, 385, 390 ferrofluids 455–8 ferromagnetic materials 444 ideal monodomain particles made from 444–9 ferromagnetism, bulk 441–2 FET see field effect transistor FeTi Pd-doped storage properties 408 powder 405 fiber reinforcement 630 field activated sintering technique (FAST) 214 field-assisted sintering 214–15 field effect transistor (FET) 389, 391 field emission SEM (scanning electron micrograph) 110, 111 field ion microscopy 497 field sintering 214–15 film growth 470 films and coatings 30–2 final grain size 133, 188 fine-grained microstructure 574 FINEMET 332, 344, 345–7, 346, 347, 348

Index alloy 332, 346, 347 nanocomposite 345, 348 finite element method 298, 661 Finnis-Sinclair many-body potentials 296 flocculation 7 flow behavior better insight 654 of supercooled liquid and consolidation 655–62 foreign nuclei 6 forging see sinter forging formation of new nuclei high overpotential and low diffusion rates 238 fossil fuel combustion 371 Fourier-transformation method 131 fracture 120 and mechanical deformation 311–16 morphology 643 strain 648 stress 642 fracturing 144 free electron momentum 679 free energy 5, 62, 125, 186, 196, 199, 575, 576, 576, 578, 578, 579 free surface 308, 505, 508, 509, 510, 511 free volume model 137, 657 friction 97, 119, 154, 161, 162, 187, 215, 253, 255, 263, 351, 537, 566, 632, 650 full densification of nanopowders methods for 200–16 fully amorphous specimens rapidly quenched or cast 640 fully nanocrystalline bulk alloys 632–3 G gas atomized powders 563, 635, 663 condensation 47–84, 126, 176, 185, 186, 189, 190, 235, 241, 248–51, 254, 269, 352, 410, 548, 549

Index gas (cont’d) condensed powder 251 detection 383, 387, 396, 399 detection materials 396 exchange technique 353, 354 mixtures 55, 379, 693 phase condensation 50, 332, 370, 372, 396, 718 phase evaporation 370, 392 reactive applications 365–418 sensing materials 392, 393, 417 sensing process 387, 388 sensors 34, 365, 383–5, 392, 399 Gaussian peak shapes 123 getters 417 Gibbs-Thomson effect 177, 196 glass formation 126, 509, 568, 587, 592, 600, 611, 621, 624 critical cooling rate 592 different mechanisms 626 glass forming alloys bulk metallic 566, 579–80, 607 systems 583, 592 glass transition phenomenon 583 glassy and partially crystallized cast material 641 glassy matrix alloy 656, 661 thermal stability 623 glove box technique 14 glycerol monooleate 26 good ductility 581, 583, 632, 634–6, 641, 642, 663, 664 good magnetic softness 527 grain boundary 133, 135–8, 141–4, 147, 149, 150, 153, 165, 176, 178–82, 184, 186, 193, 196–200, 210, 212, 241, 243, 249, 250, 254, 256, 257, 269–72, 278, 299, 304, 306, 310–12, 314, 315, 317, 318, 320–3, 333–8, 340, 343–5, 354–6, 389, 389, 396, 465, 538–41, 545, 547, 551, 554, 560, 565, 566, 571, 629, 705 characteristics 345 energy 142, 180, 196, 312, 320, 345 engineering 237, 242

733 free volume 135, 137, 251, 544 lattice diffusion 184, 197, 307 low-energy 312 melting 180, 216 of metals, diffusion in 334–5 migration 193 mobility 197, 199, 200, 307, 344 regime 138 segregation 333 self diffusion 344 slip 180, 216 structural disorder 249 structure 270, 275, 294 volume fractions 237, 242, 252 grain boundary diffusion 178, 182, 184, 186, 197, 198, 208, 272, 307, 320, 334, 341, 343, 349, 356 creep term 541 grain boundary sliding 135, 180, 181, 210, 256, 257, 311, 312, 318, 321, 541, 571, 629 grain coarsening 173, 174, 179, 185, 186, 193, 195, 197 activation energies 197 influence of pores 174, 198 tendency 196 grain growth 3, 4, 22, 30, 63, 78, 83, 129, 140–3, 158, 174, 179, 185, 190, 193, 194, 194, 196–200, 202, 203, 205, 206, 210–12, 214, 216, 239, 246, 257, 259–61, 270–4, 304, 307, 310, 320, 321, 332, 335, 339, 342, 345, 348, 355, 356, 376, 379, 385, 396, 460, 489, 502, 525, 544, 551, 552, 572–4, 647 activation energies 197 dynamics and sintering 307–11 early stages 275 heating rates 179 low overpotential and high surface diffusion rates 238 grain refinement 133, 144, 145, 238, 261, 537, 571, 632 softening 546 strengthening effects 539

734 grain rotation 150, 180, 186, 210, 216, 256, 309–11, 320, 324, 551 grain shape modifications 239 grain size 3, 32, 47, 49, 51, 55, 60, 66, 80, 83, 84, 91, 96, 119, 122–5, 126, 128–30, 132–8, 141, 144–7, 150–2, 156–8, 160–5, 173–8, 180, 186, 188, 190, 193, 194, 196, 198–206, 208, 210–16, 239–75, 276, 278–81, 304, 310–12, 314–16, 318, 320, 323, 331, 352–5, 367, 369, 376, 387, 392–9, 407, 409, 413, 462, 489, 492, 494, 495, 499–506, 508, 509, 511, 513–15, 516, 519, 527, 537–48, 550–8, 560, 565, 566, 573, 574, 584, 591, 594, 604, 606, 614, 620, 623, 628, 629, 631, 639, 643, 664, 694 concomitant increase 140 control 237, 278 decrease 163, 253 dependence 135, 137, 178, 198, 247, 251, 253, 254, 283, 543 strong 253–76 weak 247–53 dispersion 541, 544, 556 distributions 83, 544, 551, 552 effect on soft magnetic properties 513–20 log-normal 66, 304, 305, 310, 514, 515, 516, 540, 541 measurements 201–2 morphological metastability 176 reduction 135, 213 refinement 133, 151, 266, 503, 629 steady-state 125 stress-strain curves 542, 542, 546 yield stress 135, 188, 539 grain structure 66, 255, 281, 315, 551 grain switching 331 green body 186 green ceramics 193

Index green compact 186 green density 180, 208, 216 high value 187, 191, 192 lower values 190 of nanopowders 186–93 Green’s function method 685 ground state wave functions 686 growth forms 237 lower activation energy 596 process 574, 578 gun technique 567 Gutmanas’ review 190 H H2 variations in sensitivity 396 Hall effect 387 Hall-Petch behavior 253, 254, 256, 264, 311, 312, 314, 320, 528, 539–41, 546, 556 dependence 188 equation 537, 566 limit of applicability 538 plot 537, 541, 555, 556, 557 relation(ship) 134, 139, 256, 259, 265, 311, 539, 628 slope 539, 543 negative 543, 556 hard magnetic properties 487 hardness 253, 555 hcp crystal structures 123 hcp Mg compounds nanostructured phase mixtures 591 grains 638 particles 591 Hcp particles 637, 638 Hcp-Zr 126 heat capacity 126, 127, 137, 247, 248, 250, 251, 318 and thermal capacity 250–1 heat cycle mixing and grinding steps 3 heat transfer 53 HER see hydrogen evolution reaction

Index Herring’s law 184, 185 scaling 183 Hertzian stresses 180 HES see hetero-epilattice superlattice hetero-epilattice superlattice (HES) 678, 695, 698, 700 EL device 699 hetero-epitaxial systems 708 heterogeneous catalysts 365, 371 heterogeneous crystal formation contamination-induced 621 heterogeneous crystallization 617 heterogeneous nucleation 5 heterojunction 678, 695 band-edge alignment 678 Hf alloys 129 diffusion 355 high compaction stresses 188 high density 192 high energy milling 120 high energy mills 569 high-frequency permeability 488, 520, 534 characteristics 521 decrease in sample thickness 521 electrical resistivity 521 improvement by dissolution of oxygen 521–5 high pressure consolidation 189 die casting 568 technique 210 high resolution electron microscopy (HREM) 161, 175, 241, 257 high resolution transmission electron microscopy (HRTEM) images 28, 492, 698 high-strength Al alloys development of starting point 630 high supersaturation levels 6 high temperatures excessive structural coarsening 573 high-velocity oxy-fuel (HVOF) spraying 93, 94 high wear resistance 641

735 high yield stress 636 HIP see hot isostatic pressing homogeneous flow 644 homogeneous nucleation 5 host-derived hybrid materials 23–5 hot isostatic pressing (HIP) 140, 207, 208, 210–12, 215, 216, 572, 573 densification 211 hydrostatic stress 208 hot pressing 206, 207, 210–11, 213, 357, 626, 640, 659 HREM see high resolution electron microscopy HRTEM see high resolution transmission electron microscopy HVOF see high-velocity oxy-fuel spraying hybrid coatings 33 hybrid materials host-derived 23–5 organic-inorganic 22 hydride three types 402 hydride materials activation procedure 405 capacity 403 complex 412–14 kinetics of the H2-exchange 403–4 multi-cycle stability 405 sensitivity to impurity gases 404 weight 403 hydrocarbons 368 hydrogen absorption 336, 403, 404 activation 407 absorption-desorption 410 kinetics 410 desorption 404 diffusion 332 in nanocrystalline metals 350–1 diffusivity 351 higher volume density 402 jumps 351 local atomic jumps 351 sorption behavior 407

736 hydrogen (cont’d) spectroscopy 351 storage 351 hydrogen evolution reaction (HER) 265, 372, 379 electrocatalytic activity 372, 373 hydrogen gas storage compounds 403 parameters 403 hydrogen permeation measurements of 351 hydrogen propulsion test vehicles metal hydride storage 402 hydrogen storage 351, 401–3 properties of compounds 403–5 hydrogen sulfide major impact 397 hydrolysis 10, 13, 16, 19, 20, 22, 26 hydrothermal techniques 18, 19, 23 I I-phase volume fraction 586 IAG see interface adsorbed gas superlattice icosahedral configuration 175 icosahedral phases 585, 586 metastable 620 IGC see inert gas condensation imaging with magnetic nanoparticles 451–2 impact fracture energy 633 improved ductility 565 impurities 185–6 in-flight chemical reaction 63 inclusions micromechanics 541 increase in elongation 638 increased resistivity 397 incubation 579, 604 Indium Tin Oxide 692 see also ITO induced polarization charges 681 industrial application 246 industrial infrastructure 277 inelastic neutron scattering measurements of 351 inert-gas atmosphere 570

Index condensation (IGC) 47–51, 53, 55, 56, 60, 61, 65, 68–71, 73, 74–6, 78–83, 185, 186, 189, 190, 235, 241, 248, 250, 251, 254, 269, 352, 410, 548, 549, 550, 553, 555, 557, 558 advantages of 78–9 drawbacks of 79–80 early studies 50–1 principle 51–6 recent developments 80–3 flow rate 73–4 glove box 121 phase condensation 396 pressure 68–71 temperature 71–2 type 72–3 initial grain size 573 initial powder metastability 177 instrumental broadening 123 interaction of semiconductors 385 interaction with support 367 interatomic potentials 304 interatomic spacing 367 intercrystalline volume fraction 242 interface adsorbed gas (IAG)-superlattice 693 interface diffusion 331, 332, 342, 343, 348, 349, 352 modeling 332–4 interface and surface effects in nanoparticles 449–50 interface migration 333, 334, 342 interface regions bulk 694 intergranular amorphous phase 513–20 intergranular cracking processes 246, 265, 267 intermetallic compounds 78, 120, 135, 144, 145, 150, 151, 181, 197, 331, 343, 346, 375, 405, 569, 582–4, 590, 593, 594, 631, 638, 639 minimum grain or domain size 124

Index intermetallic compounds (cont’d) nanostructured phase mixtures 591 precipitated brittle nature 643 intermetallic nanocrystallites 346 intermetallic phases 584, 586, 619, 626 intermetallic precipitates 630 intermetallics 122, 135, 136, 151 internal friction measurements of 351 internal refining process 119 interparticle bonding 190, 215, 573 intrinsic sintering pressure 207 Invar effect 493, 526 ionic compounds 9 ionic species 385 IR drop 701 iron carbides catalytic properties 374 ISDN 531 isoelectric point 8 isothermal annealing 593 isotropic effective mass 680 isotropy 175 ITO 692 see also Indium tin oxide J Japanese Institute of Steels (JIS) handbook 633 K kinetic approaches 5, 174 kinetic effects 175–8 kinetic energy scales 686 Kissinger analysis 272 L laboratory scale phenomena 4, 568 lamellae 92 lamellar structure 48, 92 Langmuir-Blodgett films 26 techniques 34 lanthanide metal 581, 586 see also Ln

737 Laplace equation 175 larger supersaturation levels 6 laser vaporization 57–8, 378 late transition metal (LTM) 581 lattice defects 119, 120, 130, 164, 366, 566, 570 diffusion 184, 197, 307, 352 dislocations 142 point defects 141 LED see light emitting diodes levels of detection ppm and ppb 384 light emitting diodes (LED) 677, 691, 691 performance micro-cavities 692 PSi 691, 691, 692 simple approach 711 line broadening 123 liquefaction direct coal 368 liquid phases 235 liquid quenched specimens 617 liquid quenching 159, 581 lixiviation technique 411 Ln 581, 583, 584, 586, 587 see also lanthanide metal localized defects resonant tunneling 678 log-normal distribution 63, 304, 310, 514 logic gate devices 691, 692, 704 low Fe concentration 487 low supersaturation levels 5–6 low-friction 632 LTM see late transition metal luminescence from clusters 694–5 lyophilic group 7 lyophobic group 7 M M(OR)x 19 see also metal alkoxides M-O atomic pair 524 MA see mechanical attrition magnesium alloys superplastic 639

738 magnetic after-effect measurements of 351 magnetic anisotropy angle dispersion 528 magnetic clusters 442–3 magnetic coupling degree of 501 magnetic nanoparticles and applications 439–74 fundamental physics 441–55 imaging with 451–2 magnetic properties of nanocrystalline materials 487–534 magnetic resonance imaging (MRI) 440, 450, 458, 461, 462 magnetic scattering 131 contrast 133 magnetism, molecular 443–4 magnetization saturation 246 see also saturation magnetization magnetocrystalline anisotropy 15, 442, 452, 462, 467, 505, 514, 527 magnetoelastic coupling 133 magnetoresistance 159 magnetoresistive devices 467–73 magnetostatic interactions between nanoparticles 451–3 magnets, applications of monodomain 455–73 main sintering mechanism 184 major energy contribution 136 martensitic transformation 155, 156 massive nucleation 239 material synthesis and characterization 418 chemical reactions 3 materials chemistry 3 performance 3 structure–property relations 565 Materials Research Society (MRS) 679 MBE see molecular beam epitaxy MD simulations 180 mean activation energy 351 mean grain size 145–7, 151, 152, 314

Index mechanical alloying 48, 49, 119, 120, 124–6, 128–30, 151, 153, 161, 164, 338, 340, 371–4, 395, 408–12, 553, 617–21, 623, 624, 626 glass formation 621 high-energy 371, 395, 408, 411 mechanical alloying/milling 49 mechanical attrition (MA) 119–25, 128, 129, 133, 134, 136–9, 141, 143, 144, 159, 161–5, 186, 235, 262, 337, 338, 555, 564, 569–72, 574, 619, 626, 633, 641 alloy formation 569, 570 crystal-to-glass transition 128 deformation mechanisms 143 high-energy 121 microstructural changes 123 nanostructure formation 571 nanostructured particles 165 process 570 mechanical behavior 138, 173, 260, 294, 565–7, 570, 629, 641, 643, 660 of nanocrystalline metals 537–60 models and computer simulations 538–47 of two-phase nanostructed materials 565–664 mechanical deformation 119, 122, 135, 165, 302, 306, 417, 372, 571, 572, 664 and fracture 311–16 mechanical measurements 538 mechanical milling 126, 128, 129, 371, 415, 553 high-energy 374, 405 mechanical properties 129, 138–40, 164, 185, 191, 237, 246, 247, 260, 261, 281, 294, 299, 305, 308, 316, 321, 439, 543, 547, 548, 550, 553, 566, 582, 598, 615, 616, 628, 634, 635, 641, 650–3, 660, 661, 663, 664

Index mechanical properties (cont’d) at room and elevated temperatures 628–9 structural applications 566 mechanical working 581 mechanically alloyed powders 190, 210, 617, 624–6, 630, 634, 639–41, 655, 659 Mg-based 621 viscosity data 655 mechanically attrited composites 654–62 mechanically attrited material 640 mechanically attrited powders 617–28, 660, 663 compaction 580 mechanically milled Mg hydrogen absorption characteristics 410 mechanism of deformation 138 mechanochemical reactions 128 mechanochemistry high-energy 375 melt spinning 338, 448, 449, 507, 583, 586–91, 593, 598, 618, 636, 649 limitations 618 method 488 rapid solidification 568 technique 488, 507 melt-extraction 583, 630 melt-spun Al—ETM—LTM 583, 630 Al—Ln—TM 583, 630 Fe—Hf—B 489 Fe—Nb—B 489, 492 Fe—Zr—B 489, 492 state 492, 521 melt-spun ribbons 583, 586, 591, 593, 594, 599, 617–19, 622, 625, 632 single-phase amorphous 91 melting 83 of nanoparticles 181, 300 membrane structures 25 mesoporous materials 381

739 mesoscopic scale 310 mesoscopic sliding 331 metal (hydrous) oxide particles 16 metal alkoxides 19 metal colloids 26 metal microparticles electrodeposition 377 metal nanoparticles 212 sintering 186 metal nanopowders 194, 202 green density 188, 190 in-situ consolidation 185 oxidation 190 metal particles hydrolysis 13 metal powders chemical synthesis 10 metal–ceramic composite nanocrystalline 129 metal-matrix composites 236 metal-metal systems 587, 588 metal-metalloid systems 583, 587, 588 metal-organic chemical vapor deposition (MOCVD) 708 metal-organic vapor phase epitaxy (MOVPE) 708 metal/ceramic nanocomposites 21, 120 metallic additives 390 metallic alloys amorphization 587 metallic glass amorphous phase separation 577 composites 621, 627, 661 mechanically attrited 655 viscosity changes 657 crystallization 574 devitrification of 574–80 easy deformation 650 Mg-based 621–4 thermal stability 595 metallic glass/ceramic composite 130 metallic hydrides 402, 405–12 metallic matrix 98 metallic nanoclusters net effect 390

740 metallized tubules 32 metals 4, 9–15, 21, 30–2, 48, 50, 51, 54, 56–9, 68, 69, 78, 81, 92, 94, 103, 112, 119, 120, 122–4, 133, 135–7, 142, 144–7, 151, 155, 161, 173, 175, 177, 180, 183, 187–9, 191, 194, 197, 200, 201, 205, 210, 212, 213, 215, 235–8, 249, 251, 265, 269, 270, 277, 278, 280, 281, 283, 296, 298, 307, 310, 311, 314–17, 319, 332–7, 339–45, 347–51, 354, 356, 370, 373, 402, 454, 499, 521, 526, 537–41, 543–7, 550–3, 555–60, 566, 569–71, 629 and alloys cold working 136 production of nanocrystalline 537 cold compaction 188–90 sinter forging 212 sintering 205 metastability 440, 489, 579 metastable crystallization 572, 577 equilibria 570, 576, 577 fcc phase 594 fcc-type phase increasing Ti content 604 metastable phase 570, 572, 575 directly synthesizing 574 formation 569, 571, 575, 577, 583, 592, 610, 617, 618, 626 icosahedral 599, 604 mixture DSC trace 585 stability regions 594 methylene iodide 201 Mg-based alloys 403, 410, 587–92, 600, 621, 635, 636, 638, 660 amorphous and nanostructured 635–41 glass transition and crystallization 588–90

Index good bending ductility 635 high glass forming ability 600, 636 high hardness values 660 nanostructured 409 from bulk metallic glasses 600–1 two-phase nanostructured high-strength 635 Mg-based metallic glasses and metallic glass deposits 621–4 Mg-based powders 590 Mg-based two-phase materials superplasticity 639 Mg—Ln—TM amorphous alloys 588 Mg-rich alloys 588 Mg—Y—Cu alloys mechanically alloyed 580 Mg2Ni-H system properties 410 Mg55Y15Cu30-based composites 660 micelles 26, 27 microcracks 96, 553 microcrystalline powders mechanically grinding 405 microelectronic devices application 678 microemulsions 26 microhardness 99 microplasticity 641, 642 microscopic process 161 mechanism 181 microstructural coarsening 598, 663 microstructural evolution of electrodeposits key factor 237 microstructure modeling 302–7 microstructure–property relationship energetic 137 microstructures 22, 23, 25, 120, 122, 125, 147, 161, 237, 244, 273, 274, 281, 304, 567, 571, 572, 580, 581, 582, 584, 595, 598, 607, 609, 611, 615, 617, 621, 635, 654 microswitching converters 531 dc-dc converters 531

Index microsystem components nanocrystalline metals in 280–3 microwave sintering 213–14 milled powders high-energy 374 milling 8, 14, 25, 48, 49, 120–4, 126, 128–33, 135–7, 139, 140, 151, 157, 163, 250, 371–5, 395, 396, 405, 408–16, 553, 555–8, 560, 570–2, 618–23, 625, 627, 640 conditions 571 contamination devices 121–3 fluid 129 high-energy 375, 395, 405 high-energy forces 121 induced phase transitions 570–1 kinetic energy 570 media 570 of materials 120 operation 120 parameters 570–1 process 120, 123, 126 time 121 miscibility gap 578, 579 mixed phase alloys 583 mixed phase materials 630 Mo metal powders 129 MOCVD see metal-organic chemical vapor deposition model deformation 537 electrochemical etching 689 modeling 64, 176 interface diffusion 332–4 methods 295–9 microstructure 302–7 prospects for future 322–4 models of mechanical behavior of nanocrystalline materials 538–47 models of deformation 538–43 multi-component concept 539 molecular beam epitaxy (MBE) 708 molecular dynamics 176, 180, 213, 249, 251, 254 computer simulations 537, 543–7

741 and Monte Carlo modeling 294–6 simulations 176, 180, 213, 249, 251, 295–8, 300, 303, 304, 306, 307, 310, 311, 314–20, 322, 323, 546 molecular magnetism 443–4 molten droplets 59 molybdenum carbide catalytic activity 375 momentum conservation 684 monoclinic structure 128 monocrystalline thin films studies 387 monocrystals 389 monodomain magnets applications 455–73 monodomain particles, ideal made from ferromagnetic materials 444–9 monolayer films 25, 706 Monte Carlo 305 method 452 model 294, 296–8 and molecular dynamics 294–6 simulations 65, 125, 294, 295, 321, 322 MOS FET 705 Mössbauer-studies 126 motion of electrons 680 MOVPE see metal-organic vapor phase epitaxy MRI see magnetic resonance imaging MRS see Materials Research Society MTP 27 multicomponent alloys 603, 611, 619, 624, 629 development 664 multicomponent particles 5, 21 multicycle stability 405 multidisciplinary techniques 5 multilamellar vesicles 26 multiphase alloys 144, 584 multiphase polycrystals 565 multiscale modeling 297–9 N nano-indentation methods 138 nano-iron sintering 186

742 nanoalloys 319–22 nanobeam electron diffraction 492 patterns 495, 510 nanoceramics 57, 191, 205, 212 nanoclusters 55, 56, 299, 304, 306, 308, 372 simulations of 304 transition metal 377 nanocomposite catalyst study 370 nanocomposite densification 205–6 nanocomposite powders Fe-based 372 N-containing 372 nanocomposites densification 205 densification rate 206 sintering 206 nanoconfined proteins and enzymes 22 nanocrystal formation 237, 579 major rate determining steps 237 nucleation and growth characteristics 596 second important factor 238 nanocrystalline agglomerates 354 alumina 21 Au particles 27 ceramic 129 copper grain size 250 electrodeposit measurements 250 Fe—M—B alloys 488 Gd 249 grain boundaries 135 iron 142, 154, 162, 215, 248 microstructure 185 nitride 129 palladium 241, 250 particles 10, 248, 611, 656, 661 soft magnets 279 solids 318, 331 nanocrystalline alloys 261, 340 Al-based 584 bulk 584–6, 631–2 crystallization-prepared 351 room temperature ductility 630

Index nanocrystalline ceramics 331, 332, 345, 352 cation self-diffusion 353, 355 cold compaction 191–3 diffusion 352–6 nanocrystalline copper (Cu) 254, 270, 335, 345, 371, 549 elastic moduli 554 foil 278 for printing wiring boards 278–80 stress-strain curves 546, 556 nanocrystalline Cu-Ni alloys elastic moduli 553 nanocrystalline Fe 554 elastic moduli 553 grain growth 143 nanocrystalline Fe50Ti50 406 surface chemistry 407 nanocrystalline FexCu100-x powders 126 nanocrystalline hard-magnetic Nd2Fe14B compounds 349–50 nanocrystalline materials 51, 52, 121, 135, 138, 165, 176, 177, 187, 196, 200, 202, 208, 209, 211, 235, 242, 243, 262, 264, 269, 277, 323, 331–7, 339, 349, 370, 405, 407, 409, 487, 538, 540, 571–3, 664 author’s research 217 characterization 537 chemical and physical microstructure 247 corrosion resistance 265 diffusion 331–56 electrodeposition 236 emerging applications 247 magnetic properties of 487–534 mechanical properties 260 stability 331 strain rate of measurements 555 synthesis 236, 237 thermal stability 197, 270, 273 nanocrystalline metal electrodeposits structure of 239–46

Index nanocrystalline metals 155, 189, 235, 236, 238, 249, 251, 269, 277, 278, 280, 283, 314, 316, 332, 333, 338–45, 349, 350, 357–9, 541, 543–5, 547, 550–3, 555–8, 560, 629 characterization of 547–8 density, pores and microcracks 548–52 controlled stabilization 345 diffusion 332, 335 of hydrogen in 350–1 dispersion in grain size 540 ductility 556–8 elastic properties 553–5 electrodeposited 557 Hall-Petch plots 537, 541, 555 hardness, yield and ultimate strengths 555–6 high hardness values 555, 560 limited ductility 556 mechanical behavior 537–60 in microsystem components 280–3 pure 252, 259 relaxation effects 554–5 room temperature ductility 566 strength 537 voids 548 nanocrystalline Ni 239–41, 242, 244, 246–54, 256, 257, 259–61, 263–7, 269–76, 315, 334, 340, 374, 548, 555, 557, 559, 560 alloys 241, 273 annealed 273 as-plated 273 corrosion behavior 265 deposits 258 electrodeposited 557 electrodeposits 241, 248 heat capacity 251 mechanical properties 281 powder 241 pure 252 specimen 239

743 nanocrystalline Ni—P 241 nanocrystalline Pd 241, 340–2, 346, 549, 552 and Cu 254 nanocrystalline powder consolidation methods 173–217 nanocrystalline powders 173, 185, 207, 208, 214, 395, 408, 409, 549, 553, 572 nanocrystalline pure metals 340–3 nanocrystalline refractory metals 177 nanocrystalline soft-magnetic alloys Fe73.5Si13.5B9Nb3Cu1 and Fe90Zr7B3 345–9 nanocrystalline-Si (nc-Si) 594, 693 in oxide matrix 700–2 nanocrystalline specimens 157, 335 nanocrystalline structures 145, 161, 205, 214, 306, 577 deformation behavior 629 nanocrystalline ternary alloys soft magnetic properties and structural analyses 492–9 nanocrystalline thin films 339 nanocrystalline/crystalline phases 663 nanocrystallites 159, 306, 315, 334, 340, 346–9, 645, 682, 684, 688, 702, 704 Fe-diffusion 346 intermetallic 346 nucleation and growth 700 thermal-vacancy concentration 348 nanocrystallization 488, 533 nanocrystallized material 412, 641 nanocrystals 11, 15, 17, 18, 20, 31, 120, 175–7, 187, 197, 200, 240–2, 251, 257, 279, 301, 310, 333, 366, 371–3, 378, 381, 392, 407, 408, 415, 450, 505, 580, 590, 596, 597, 601, 606, 607, 618, 642–5, 658, 661, 683, 684, 704, 705 Au 376 Cu 279 electrodeposition 240 formation of 606

744 nanocrystals (cont’d) isothermal grain growth 197 Pb 279 nanograined WC-Co powders 205 nanograins 137, 312, 315, 320, 411 nanogranular two-phase alloys 586–7 nanoindentation technique 553 nanoindenter 551 nanomaterials 34, 56, 83, 99, 173–5, 189, 191, 197, 199, 200, 236, 237, 246, 247, 252–4, 260, 262–4, 277, 366, 416 diffusion data 198 fully dense 277 properties 216 nanometal densification 191 nanometals green density 189 nanometer dimensions 48, 140 nanometer scale 18, 83, 121, 125, 293 device fabrications 677 lithographic fabrications 677 model catalysts 382 nanometer-sized crystals 660 nanoparticle sintering 176 nanoparticles 4, 9–20, 23–6, 28, 29, 33, 48–51, 55–63, 65–7, 72, 75, 78, 81–4, 98, 174–6, 179–82, 184–7, 190, 191, 194, 200, 202, 203, 212, 214, 215, 245, 302, 306, 308, 309, 311, 345, 376, 369, 376–8, 381–3, 396, 411, 439–41, 443–7, 449–63, 465–7, 469, 470, 471, 473, 474, 642, 661, 681, 692, 694, 701, 712 capacitance in 685–7 copper 300 cytotoxicity 28–30 dispersed 377 dopants 679 doping of 680–2 exchange interactions between 453–5 iron and copper 182

Index magnetic 439–74 magnetostatic interactions between 451–3 measurement techniques 200 principal differences 187 sintering 181, 182 surface and interface effects in 449–50 nanophase materials 84 mechanical properties 119 structures 565 nanopores 201, 241, 345 nanoporous TiO2 thin films 34 nanopowder consolidation pressure effects 206–10 process 174 success 196 nanopowder densification 200, 210, 213, 214 pressure effects 206 thermodynamic and kinetic aspects 174 nanopowder processing 173 nanopowder sintering 173, 178 process 216 nanopowder synthesis method major improvements 174 nanopowders 75–9, 159, 173–6, 178–80, 182–91, 194, 195, 198, 202, 203, 205–8, 215, 216, 246, 374 attrition-milled 186 compaction 187 densification 173, 183 densification methods 200 full densification 174, 178 green density 186–93 intrinsic curvature stress 207 pressure-assisted sintering 174 processing 173 sintering 216 surface oxide 176 nanoscale bcc alloys 487 bcc structure 487 composites 628 compound formation 596–7 compound particles 593

Index nanoscale (cont’d) crystalline particles 626 electronics 687 fcc Al particles 581 I-phase 585 icosahedral (I) quasicrystalline particle 582 materials 54 microstructures 120, 571, 572, 581 mixed phase alloys 583 particles 217, 366, 367, 637, 638, 641, 661, 682, 684, 700, 701 phases 662 precipitates 588 quasicrystalline alloys 584, 619 particles 619 silicon dielectric constant of 678–80 particles 677 structure 566 nanoscale phase mixture 571–2, 631 after solidification 631 nanoscaled memory devices 704 nanosintering 174 pore effects 174 nanosize grains 203, 215 metals 173, 197 particles 177 powders 173, 174 precipitates 606 nanostructure crystallization-induced 487, 488–92 design 370–83 formation 567, 571, 580, 598 influence of oxygen on 625–6 on mechanical attrition 571 phenomenology of 580–628 materials 48 nanostructure-controlled materials future progress 534 nanostructured alloys 319, 488, 581, 600, 629–31, 635 deformation behavior 638–9 Mg-based 600–1

745 nanostructured AlN powders 20 nanostructured alumina–titania coating 98–9 nanostructured catalysts 375 nanostructured ceramic oxide films 32 nanostructured coatings 92 nanostructured composites 159, 580, 610 flow behavior 650 new 595 nanostructured compounds 389 nanostructured design of sensing materials 392–9 nanostructured electrodeposits equiaxed 239 nanostructured electronics and optoelectronic materials 677–712 applications 691–711 nanostructured feedstock powders 92 nanostructured films chemical synthesis of 30–5 nanostructured grains 30, 365 nanostructured magnesium-alloys hydrogen storage capacity 410 nanostructured materials 3, 4, 13, 35, 47–9, 51, 53, 78, 83, 91, 99, 119, 161, 165, 236, 237, 239, 246, 247, 293, 294, 299–321, 324, 365–7, 383–5, 387, 388, 394, 402, 403, 412, 414, 416, 417, 469, 565, 566, 568, 569, 580, 601, 607, 619, 629, 630, 647, 651, 654, 663, 664, 677, 682, 694 definition 417 development 565 for gas reactive applications 365–418 interface-to-volume ratio 566 magnetic structure 247 mechanical properties 439 metallic 567 physics 677–91 prepared by solid state processing 119–65

746 nanostructured materials (cont’d) surface defect structure 417 synthesis 235 synthesizing 49, 51 technological importance 365 thermal stability 270 three gas reactive applications 365 nanostructured metal coatings 30, 31 nanostructured Mg85Zn12La3 alloy two-phase 637 nanostructured Ni electrodeposited 265 nanostructured particles 3, 7, 34, 75 chemical synthesis 3–35 nanostructured powders 52. 55. 108, 165, 573 nanostructured quasicrystalline alloys 619–21 nanostructured systems, novel 399–401 nanostructured two-phase alloys 630 nanostructured two-phase materials 567 nanostructures 4, 11, 23–5, 48, 55, 119, 164, 176, 237, 245, 250, 277, 302–7, 320, 321, 323, 381, 382, 392, 414, 415, 417, 465, 566, 617, 663, 677, 688, 692 electronic and optical properties 677 fully dense 277 nanovoids 241 nanowires 382, 401, 465, 684, 688 narrow size distribution 6, 13, 14, 17, 19, 21, 50, 195, 573 nature of defects impact on sensing properties 417 Nb, effect of 613–15 nc-Si see nanocrystalline-Si nc-Si/SiO2 superlattices photoluminescence 692–4 Nd2Fe14B hard-magnetic compounds 349–50 NDC 707 near edge x-ray absorption fine structure (NEXAFS) 688, 690

Index neck formation 64, 178–81, 213 growth 76, 179, 181, 186 negative enthalpy of mixing 126, 577 neutron scattering 130, 147 method 133 Newtonian flow behavior 650, 654 viscous 646, 654 NEXAFS see near edge x-ray absorption fine structure Ni atoms 249 fcc 251 grain size 205, 249–50 peaks 126 triple line softening effects 255 Ni-metal hydride batteries 402 Ni—Mo 279, 373 Ni—P electrodeposited 240, 241, 254 electrodeposits 256 bct-Ni3P 251 nickel electrodeposits 251, 256, 261 nickel nanocrystals pulse plated 240 nitride formation 21 NMR see nuclear magnetic resonance noise attenuation characteristics 531 non-agglomerated powders 192 non-aqueous electroless deposition 32 non-conventional sintering methods 213–16 non-crystalline structures deformation behavior 629 non-equilibrium alloys 581 non-equilibrium phases 581 non-equilibrium processing 240, 569 non-equilibrium structures 582 non-equilibrium vacancies 136 non-hydrolytic sol-gel methods 22–3 nonpolar chains lyophilic 8 nonstoichiometric oxides 373 nonaqueous solvents 5, 8 NOR-gate function 693 nozzle diameter 75

Index nuclear magnetic resonance (NMR) 339 nuclear scattering 131 contrast 133 nuclear steam generator rehabilitation 246 nucleation 4–6, 10, 13, 15–18, 30, 32, 50, 51, 53, 55, 60, 62, 70, 130, 142, 239, 239, 243, 257, 305, 312, 314, 317, 318, 323, 450, 466, 497, 502, 545, 549, 577–80, 583, 592, 596, 599–601, 607, 626, 628, 700 activation energy 575, 696 barrier 575 and crystal growth 574–5 and growth 4, 5–6, 18, 50, 62–4, 71, 142, 545, 574, 578, 579, 592, 596, 599–601, 607, 626 process 578 rate 9, 71, 203, 213, 574 theory 63–5 limits of 64–5 nucleophilic addition 20 nucleophilic substitution 20 O 18 O diffusivities 337 18 O-diffusion studies 353 O-ring 457 OEE see ozone-enhanced evaporation one-electron cases 685 open pores pinning effect 198 optical absorption 368 gap 682 optical applications 678 optical-logic gate 691 optimum optical device 708 optoelectronic applications 682, 697 nanostructured materials 682 optoelectronic devices 678 optoelectronic materials and nanostructured electronics 677–712

747 organic fluids ball milling 129 organic materials 22 organometallic complex thermal decomposition 377 organometallic compound advantages 14 organometallic methods 14–15, 16–18 organometallic precursors sonochemical decomposition 375 thermal decomposition 16 Orowan looping 543 overpotential 237 oxidation 4, 10, 13, 21, 32, 60, 61, 71, 76, 78, 94, 98, 105, 128, 158, 187, 190, 236, 302, 303, 352, 368–73, 376–9, 381, 390, 391, 394, 396, 398, 403, 407, 410, 450, 463, 507 oxidation-reduction process 373 oxidation/reduction treatment 372 oxide formation 249 layer 6, 34, 72, 77 matrix 21, 22, 677, 685, 700 particles 16, 21, 27, 157, 158, 622, 623, 527, 628, 655, 660, 661 phases 98, 371 oxygen contamination 573 detection 396 diffusion 352–4, 699 diffusivities 354 influence on thermal stability and nanostructure formation 625–6 peaks 693 oxygen-induced precipitation of nanoscale compound particles upon crystallization 593–5 oxygen-triggered crystallization 625–6 ozone-enhanced evaporation (OEE) 398

748 P P in Ni room temperature solid solubility 240 PA/ABS 129 PA/PE 129 packing density slight changes 657 parallel plate rheometry 646, 654 partially crystallized alloys 583–4 particle(s) 5–30 behavior 71 bonding improvements 573 collection 61–2 growth 5 morphology 5 narrow size distribution 6 Ostwald ripening 6 rotation 180 size control 369 distribution 702 influence of process variables on 67–8 reduction 13 steady-state 144 variation 701 temperature 66, 67 transport 60–1 wide size distribution 6 Pauli exclusion principle 685 Pd 342 103 Pd 342 n-Pd 342, 345 nanocrystalline 342 submicrocrystalline 341, 342 Zr-doped 345 PE/Cu 129 peak broadening additional defects 124 peak luminescence energy 682 peak recombination energy 682 peak stress value 645 Penn model 679, 683 periodic boundary conditions 304 permeation 351 bcc-phase 489, 492, 493, 495, 521, 522, 524, 525, 613, 652 heterogeneous nucleation 497, 502

Index phase bi-stability 706 phase changes 71, 299, 492, 575 phase equilibria metastable 570 stable 570 phase separation 577, 578, 601 in bulk glass-forming alloys 579–80 phase-separated domains 580 phases amorphous 315 equilibrium studies 19 formation 568, 571 stability 299 transformations 573 transitions 455, 571 phenomena of precipitation 5 photoacoustic infrared spectroscopy 185 photodetector 691, 692 photoluminescence (PL) 680, 697, 699 blue 690, 691, 694 efficiencies 708 efficiency 691 in nc-Si/SiO2 superlattices 692–4 peak 688, 689, 690, 693, 699 stable blue 694 unstable UV 694 physical vapor deposition 392 physics of nanostructured materials 677–91 piston cylinder method 209 PL see photoluminescence planar inductor 532 plasma heating 60 spraying 93 synthesis 49 plastic deformation 121, 122, 124, 135, 143, 144, 147, 151, 152–4, 159, 161, 178, 180, 209, 210, 213, 255, 260, 307, 308, 311, 312, 315, 317, 322, 332, 338, 339, 341–3, 344, 541, 558, 560, 570–2, 630, 634, 638, 639, 644, 647, 652, 653

Index plastic (cont’d) yielding 187 plateau pressure 408 plating bath saccharin content 275 platinum array catalytic activity 382 PLE see excitation photoluminescence PMMA see polymethylmethacrylate point defects intermetallics 136 polar compounds 9 polar organic media 8 polaritons 680 polarization 681 interaction 683, 685 poly(vinylpyrrolidone) (PVP) 11 polyamide, PA 128 polycrystalline copper grain size 250, 251 counterpart 135, 246, 251, 266 Gd 249 iron 248 Ni 241, 249, 252, 253, 255 nickel 244, 252, 261, 263, 265, 268, 269, 272 palladium 250 Si films 694 wires 255 polycrystalline material 124, 197, 241, 247, 251, 257, 334, 387, 412 fine-grained 261 polycrystallinity effect of 387 polycrystals 331 polyethylene (PE) 128 polyethylene glycol 458 polymer blends 120 polymer chains milling process 128 polymer mixtures mechanical alloying 129 polymer science principles 129 polymeric materials mechanical alloying 128 polymerization 20

749 polymers with ceramic mechanical alloying 129 polymers with metal powder 129 polymethylmethacrylate (PMMA) 11, 14, 400, 465, 466, 706, 707 polymorphous crystallization 576, 577 polymorphous nanocrystallization 577 polyol method 10–13 polypropylene, PP 129 polysilicon floating gate 705 polystyrene, PS 128 pore breakaway 193 closure 178 growth 193 shrinkage 193 structure 216 pore size 174, 193 distribution 186, 192 effects on densification behavior 193–6 pore-boundary separation 193 pore-grain boundary breakaway 193 pore-to-grain size ratio 193 porosimetry technique 192 porosity 33, 34, 79, 80, 94, 96, 98, 111, 115, 178, 180, 193, 194, 194, 198, 201, 242, 249, 251, 257, 270, 277, 278, 301, 308, 334, 335, 339–42, 345, 348, 354, 396, 547, 548, 553, 554, 572, 573, 630, 660, 692 porous silicon (Psi) 399, 677, 679–82, 682, 684, 688–93 application 690, 691–2 bonds 688 cluster-like 689 color dimensionality and size 689, 690 coordination 688 electrochemically etched 679, 680, 688 encapsulating 692 improved morphology 692 luminescence 689

750 porous silicon (Psi) (cont’d) nanostructures 414–16 oxidizing 692 photoluminescence 684 PL and EL spectra 692 red, yellow, and green 689–90 and Si clusters 688–91 structural robustness 692 structure 688, 692 porous silicon-based diodes 691 position annihilation spectroscopy 241 positive enthalpy of mixing 578 positron annihilation spectroscopy 201 positron lifetime 345 high-temperature studies 346, 349 spectroscopy 351 positron spectroscopy 548 post-oxidation 352 potential crack nucleation site 549 potentials application 350 embedded atom method 251 Finnis-Sinclair many-body 296 interatomic 304 Stillinger-Weber 296 potentiodynamic polarizations 268 powder(s) components 570 consolidation 572–4 contamination 174 high reactivity 622 metallurgical pathway 130 milling experiments 164 particles 570, 571 thermal stability 620 powder material crystallinity 128 powder metallurgy 569, 580 contamination effects 625 powder mixtures ceramic 128 Fe2O3/Cr2O3 128 mechanical alloying 124 Zr60Al10Ni9Cu18Co3 126 ZrO2/Y2O3 128 powder particles 119, 121 cold work 135, 140

Index deformation processes 122 intermetallic AlRu 133 large frictional forces 187 mechanical alloying 119 mechanical attrition 119 single-crystalline 119, 164 powder samples mechanical attrition 162 x-ray diffraction 140 powder synthesis 569–70 power transformer efficiency 530 PP/Al 129 PP/SiC 129 precipitates volume fraction 584, 590 precipitation 4, 5, 12, 15, 19 chemical method 16 reaction 4 precipitation and transformation details 596 precipitation hardening 261–2, 629 precomposite gels synthesized by pyrolysis 22 precursor 4 compounds 10 methods 14 powder material 237 powders 20 pressure-assisted consolidation methods 210–13 pressure-assisted densification 216 pressure-assisted sintering 174, 208, 216 pressure effects in nanopowder consolidation 206–10 pressureless sintering 21, 188, 202 primary crystallization 575, 577 primary phases 587 nanocrystalline 577, 580 printed wiring boards nanocrystalline copper for 278–80 process variables influence on particle size 67–8 processing application of methods 385 techniques 567 propylene gas 93

Index PS/Sn 129 PSi see porous silicon pulse echo technique 553 pulse electro-discharge consolidation 214 pulse plating 239 PVD crystallite size 395 PVP see poly(vinylpyrrolidone) pyrolysis 14, 22, 48, 369, 377, 392, 397 Q QD see quantum dot QD vertical-cavity surface-emitting laser (VCSEL) 708 QD-FETs see quantum field effect transistors quantum computing 702 quantum confinement 679, 680, 682, 684, 685, 688, 693, 700, 709 effects of charge accumulation 685 superlattices and quantum wells 677–8 quantum dot (QD) 680–5, 687, 704, 707–9, 711, 712 active region 708, 709 devices challenges in 711–12 wide variety 704 EL 690 electron tunneling 684 exciton binding energy 684 laser 708–11 three dimensionally confined 677 quantum effects 678 quantum field effect transistors (QD-FETs) 677 quantum interference 678 size effect 677, 680, 694 wells 677, 680, 696, 708, 711 wire 679, 690 quantum-mechanical calculation 685 quantum system man-made 678 quantum wells and superlattices 677–8

751 quasicrystal formation 619 mechanical attrition 619 quasicrystal grain size 620 quasicrystal-to-amorphous transition 571 quasicrystalline alloys 617–18 quasicrystalline particles 598, 603, 619 Al-based alloys containing 598–9 icosahedral 584 quasicrystalline phases 567, 581, 586, 621 alloys with 632–4 metastable 572, 594, 620 nanoscale 596, 602 quasicrystalline polycrystals micrometer-sized 599 quasielastic neutron scattering 351 quasiperiodic phases 581 R radial distribution functions (RDF) 131, 132 radiative decay time 692 radiative recombination 684, 710 Raman phonon frequency 682 scattering 694 shift 688 rapid crystallization 604 rapid grain growth 332 rapid heating treatment 502 rapid quenching 587, 630, 639 fabrication routes 580 from melt 580, 663 techniques 569, 574, 618 rapid sintering 180 rapid solidification 46, 96, 235, 240, 488, 583, 598, 600, 617 constitutional changes 567 of materials 581–97 microstructural effects 567 techniques 567–9 rapid thermal oxidized (RTO) 692 rapidly quenched and cast materials 635 rapidly quenched powders 580

752 rapidly quenched ribbons 592, 626, 635, 663 consolidation 598 rare earth carbide nanocrystals 187 Rayleigh scattering 679 RDF see radial distribution functions reactants 4 reactive ball milling 129 reactive gases 55, 76, 78, 81 atmosphere 129 effect of 76–8 reactive ion etching 705 reactive milling 129, 373 reactive sputtering 392 one advantage 393 recombination energies and exciton binding 682–4 and scattering centers 700 recovery and recrystallization processes high-angle grain boundaries 571 recrystallization 5, 142, 190, 571, 647 redox 4 reactions 370 reduced grain growth 206 reduction/dissociation transformation 407 reduction/oxidation reaction 4 reflection high energy electron diffraction (RHEED) 697, 698, 710 refractory metals milling 122 relative mass density 342 residual amorphous matrix 577, 590 residual amorphous phase 489 residual porosity 79, 251, 270, 573, 630, 660 artifacts 278 resonant conductance peak 701 resonant tunneling 701, 707, localized defects 678 reversed micelles 26 rhapidosomes 31 RHEED see reflection high energy electron diffraction

Index room temperature ductility 630 mechanical properties 642–4 rotating-beam fatigue strength 632 RTO see rapid thermal oxidized Rutherford backscattering 338 S saccharin 238 SANS see small angle neutron scattering SAS see semiconductor/atomic/ superlattice saturation magnetization 32, 246–50, 442 large reduction 247 see also magnetization: saturation SAXS see small-angle x-ray scattering scaling laws 174, 182–5 scanning electron microscopy (SEM) 107, 111, 146, 599, 607, 612, 621 grain sizes 201 micrograph 78, 106, 110, 383, 392, 393, 550, 600, 608, 613, 614, 616, 617, 643 scanning tunneling microscopy (STM) 382, 454 three dimensional 201 scattering intensity 131, 133 Scherrer analysis 124, 201 Scherrer formula 124 Schlenck line technique 14 Schottky junctions 371 Schottky-barrier height 389 second phase particles 259, 261, 270 secondary ion mass spectroscopy (SIMS) 338, 339, 352 segregation 106, 153, 156, 176, 199, 263, 266, 275, 320, 321, 323, 345 selected-area electron diffraction pattern 145, 489 self-assembled phospholipid hollow tubules 31 self-polarization 683

Index SEM see scanning electron microscopy semiconductor/atomic/superlattice (SAS) 695–9 semiconductors 16, 28, 695 depletion depth 390 gas sensors 365 physical principles 385–92 temperature 389 sensing materials nanostructured design 392–9 sensors gas 383–5 semiconductor physical principles 385–92 separation, amorphous phase 577–9 sequential transformations 594 SET see single electron transistor severe plastic deformation consolidation (SPDC) 210, 213 shear bands 131, 133, 134, 137, 147, 257, 314, 315, 571, 607, 631, 643, 647, 651, 662 inhomogeneous deformation 642 shear conditions ductility 135 shear deformation 144, 145, 147, 156, 180, 573 shear forces 162 shear modulus see elastic moduli shearing processes 125 shock loading 316–18 shockwave consolidation 215–16 Si-clustered samples 707 Si nanostructure bonds and coordinations 688–91 structure 688 Si—O 691 Si/O superlattice 695–9 electronic applications of 702–3 Si photodiode UV-enhanced 692 Si quantum dots 707 Si—Si bond length 691 shells 689 Si—Si—Si 691

753 α-Si/SiO2 700 Si3N4 powders adsorbates 179 SiC particles uniform distribution 130 SiC/PSi/Si pn junction 682 silicon 19, 58, 81, 106, 179, 306, 309, 315, 316, 382, 395, 399, 401, 520, 677–85, 688, 689, 691–7, 699, 700, 702, 704–7 barrier system 696 clusters samples 694 dips 693 effective barrier 678 electrochemical etching 681 epitaxial growth 696 epitaxy 696 extrinsically doped 689 layer 692 nanocrystallites memory device 704 nanocrystals 705 broad size distribution 705 nanoparticles grain size 694 recombination 682 silicon quantum dots 533 excitons recombination and binding energies 682 siloxene derivatives 688 SIMS experiments 352 simulations atomic level 294 atomic scale 125 atomistic 293, 335 hybrid techniques 298 molecular dynamics 176, 180, 183, 213, 249, 251, 294–8, 300, 303, 304–7, 309–11, 314–20, 322, 323, 537, 543, 560 Monte Carlo 65, 125, 295–8, 305, 321, 322, 453 and molecular dynamics 294

754 single-electron charging effects 704 devices nanocrystal-based 706 MOS memory 706 nano-capacitance devices 705 transistor (SET) 704–8 tunnel junctions 706 tunneling events 707 oscillations 706 single-phase amorphous alloy 565, 583 polycrystals 493 sinter forging 182, 207, 211–12 technique 212 two-stage 212 sinterability 180, 185, 188, 203 sintering 21, 47, 63, 65, 67, 76, 79, 119, 173–90, 192–208, 210–17, 299, 304, 307–9, 311, 315, 352–6, 370, 381, 395, 396, 409, 466, 549, 572 ceramics 202–5 conventional 181, 202–6, 211, 213–16 driving force 176 early stages 181 field-assisted 214–15 final results 186 final stages 193 and grain growth dynamics 307–11 high-temperature 396 influence of contamination 186 intermediate stage 198, 212 kinetics 307 late stages 182, 194 lower temperatures 178, 184 mechanisms 174, 178–85 metals 205 microwave 213–14 nanoparticles 180 nickel particles 184 nonconventional methods 213–16 plasticity-driven 206 pressure-assisted 175 process(es) 174, 175, 178, 181, 185, 196, 216

Index rates 307 second stage 193 surface controlled process 185 temperature decrease 174 temperature-pressure trade-off 206 ultrahigh-pressure 213 SiO2 matrix 694 SiO2/Si interface 696 SiO4 23 sliding wear 99, 122 slip and twinning mechanism 133 slip bands 143 slow cooling fabrication routes 580 methods 598 slow evaporation slow process 33 slowly cooled bulk composites Ti-based nanostructured 610–12 Zr-based 607–10 containing ductile phases 647–50 small-angle neutron scattering (SANS) 130–3, 450, 549, 579, 580 small-angle x-ray scattering (SAXS) 61, 79, 580, 601 Smc-Pd see Pd: submicrocrystalline Sn effect of 615 laser ablation 394 SnO2 Ag-doped 398 critical size 393 crystallite size 395 disadvantage 391 Fermi level 390 laser ablation 394 sensing properties 391 SnO2films 376, 387, 392, 394, 396 Pd-doped 396 pure 396, 397 soft magnetic Fe—Zr—Nb—B—Cu alloys 488 films high permeability 529 low loss 530 materials 488

Index soft magnetic (cont’d) properties 487–9, 492–4, 497, 499, 501–7, 509, 513–16, 518, 520, 521, 527 improvement 499–503 mechanism 499 and structure of Cu-free quaternary Fe—Zr—Nb—B alloys 503–7 and structure of Fe—Nb—B— P—Cu alloys produced in air 507–20 soft x-ray absorption (XAS) 688 sol-gel 23, 24, 26, 376 film 34 methods 19–22 non-hydrolytic 22–3 process 16, 19–22, 32–4, 235 encapsulation of biomolecules 22 thin films and coatings 34 techniques 19, 32 titania 34, 203 solid particle bonding 178 solid solution hardening 261–2 solid solution strengthening 629 solid-state alloying 120, 569 solid-state densification three stages 178 solid-state devices pn-junctions 682 solid-state gas sensors important problems 392 solid-state processing 119–65, 617 solid state reaction 632, 663 solid-state sensors 391 solid-state synthetic approach 3 solid state transformations 162 solid/solid interfacial energy 631 solidification behavior 587 conventional, of bulk samples 598–616 high cooling rates 567 sols 6 solubility of alloys 125 ranges 240 solute elements to improve soft magnetic properties 499–503

755 redistribution 631 segregation effect 567 solute drag 270 effect 199 solvothermal method 18–19 sonochemical methods 13–140 spark erosion 49 spark plasma sintering 214 SPDC see severe plastic deformation consolidation specific heat 247 SPEX model 8000 121 spherical Bessel functions 679, 685 spherical morphology 586 spinodal behavior 125 spinodal decomposition 578 spray methods 569 spray pyrolysis 392 sprayed coating 93 spraying parameters 108 sputtered Fe-Hf-O electrical resistivity 488 sputtered Fe-Zr-O electrical resistivity 488 sputtered nanocrystalline powder inert gas condensation 410 sputtering 49, 53–7, 58–9, 62, 72, 78, 81–3, 270, 354, 371, 392–4, 399, 410, 411, 470, 521, 527, 690, 694 squeeze casting 598, 599, 634 stable crystalline compound 576 stable equilibrium phases 596 stable nuclei 5 static compression 640 steam generator 273 tubes 246 stepwise transformation 593, 596 steric stabilization 8 Stillinger-Weber potentials 296 STM see scanning tunneling microscopy stored enthalpy 136 strain broadening 123 distribution 124 hardening 138 strain-layer barrier 696

Index

756 strain rate–stress relation 646 Stranski-Krastanow (SK) mode 708 strength data 537 influence of grain size 553 measurements 539 stress corrosion cracking 246 stress–strain curves 257–9, 259, 542, 542, 546, 546, 555, 556, 559, 641, 642, 647 stress-strain plot 555 strong carrier confinement 707 structural coarsening 572 structure of Si nanostructure 688 structure formation of two-phase nanostructed materials 565–664 structure-microproperty relationships 135 submicrocrystalline Pd 342 see also Pd: submicrocrystalline substitutional atom diffusion 331 substrate microstructure 237 suction casting 568 high pressure 636 sulfated zirconia 24 supercooled liquid stability 600 supercooled liquid/amorphous phase 631 supercritical drying fast process 33 superlattices 5, 236, 677, 678, 680, 692, 694, 702 and quantum wells 677–8 strain-layer 696 superplastic deformation 260, 638, 639 strain rate 639 superplasticity 173, 180 high strain rate 632 supersaturated solutions 5 solid 571, 576 surface activation 375 adsorption characteristics 190, 237 defects 375 energies 307 free energy 62, 196

impurities 185 and interface effects in nanoparticles 449–50 lithography 382 morphology 266, 372 oxide layers 207 states 688 stresses 302 structures 366 surface diffusion 76, 178–80, 182, 186, 203, 205, 237–9, 298, 307, 309 of adions 237 mechanism 178 surface-related nanostructures 414–16 surfactant 7 membrane-mediated synthesis 25–8 membrane structures 25 molecules 9 polar head group 8 self-assembled structures 25 synthesis of nanostructured materials 48–50 by electrodeposition 237–9 by inert gas condensation methods 47–84 of nanostructured particles and films 3–35 of particles 4 T Ta effect of 612–13 metal powders 129 technological applications 572 TEM see transmission electron microscopy temperature stability 238 temperature-pressure requirements 572 temperatures elevated deformation behavior at 643–7 room mechanical properties at 628–9 tensile ductility 253–6

Index tensile fracture strengths 581, 587, 630, 635 tensile strength 257 tensile stress 144 tensile yield strength 638 ternary alloys 492 terrace-ledge-kink mechanism 179 tetrahydrofuran (THF) 409 tetrakaidecahedral grains 242 theoretical density values measurement techniques 200 theoretical fits luminescence 689 thermal annealing 377, 395, 396, 593, 603 thermal barrier coatings 632 thermal conductivity 205 thermal desorption measurements of 351 thermal expansion 97, 107, 247, 245, 250, 251, 278, 279 effect of grain size 251 and heat capacity 250–1 volumetric 250 thermal phonons 684 thermal shock resistance 283 thermal spray processes 92, 96–8 types of 93–5 technology 92–8 thermal sprayed nanostructured coatings 93 alumina–titania 98–9 applications and developments 91–116 current understanding 103 thermal spraying 91, 100, 106, 108, 109 thermal stabilization 384 thermal stability 270–6 influence of oxygen on 625–6 thermochemical oxide reduction 174 thermodynamic and kinetic effects 175–8 thermodynamic approaches 198 thermodynamic driving force increase 579

757 thermodynamic equilibrium 5, 120 thermodynamic properties 318 thermomechanical analysis (TMA) 640 THF see tetrahydrofuran thin coatings 252 thin epitaxial layer 696 thin film 339, 376, 387, 392 transformers 532 thin oxide layers 622 thin ribbons rapidly quenched 635 three-dimensional bulk specimens direct preparation 568 Ti alloys 262 metal powders 129 self-diffusion 355 Ti-based alloys 651–4 Ti-based bulk nanostructure composites, slowly cooled 610–12 tight binding electronic structure calculations 249 time-temperature-transformation curves 592 TiN nanopowders green density 191 Tin oxide 385 TiNi—C 125 TiO2 nanopowders green compact 188 titania 203 titania-based coating development and application 102–16 background 102–3 case study 106–16 failure analyses 103–7 TMA see thermo mechanical analysis topological metastabilities 176–7 torsion 143 toughness 553 TPD experiments 374 TPL see tunneling phase logic tracer atoms 332 tracer diffusion 339

758 transformation eutectic crystallization 576 mechanism 574 polymorphous crystallization 576 primary crystallization 577 types of 575–7 transitions crystal-to-amorphous 571 crystal-to-quasicrystal 571 quasicrystal-to-amorphous 571 transmission electron microscopic observation 489 transmission electron microscopy (TEM) 12, 18, 61, 74, 79, 133, 134, 146–8, 150, 152, 156–60, 162, 163, 175, 179–81, 186, 201, 202, 275, 276, 377, 378, 392, 393, 450, 457, 489, 491, 492, 494–6, 510, 512, 513, 513, 515, 516, 521, 523, 544, 549–51, 580, 585, 586, 591, 593, 601–3, 605, 606, 606, 611, 612–15, 618, 619, 621–3, 627, 627, 642, 645, 659, 661, 688, 689, 693, 693–5, 703, 710 analysis 159 brightfield image 606 grain size 146, 158, 162 image 491 investigations 585 measurements 202, 551, 593 micrographs 133, 392, 393 microstructure 591 results 162 sintering studies 175 studies 175, 179, 180, 642 studies of alumina and zirconia 179 tribological performance 283 triple junction 131, 237, 240, 242, 243, 250, 252–7, 266, 269, 278, 312, 314, 318, 366, 544, 545, 547, 548 defects 242 increased density 547 volume fraction 237, 240, 242, 252

Index tubules 31 tungsten carbide contamination 395 tunneling oxide 704 tunneling phase logic (TPL) 707 two-component concept 539 models problems 538 system 554 two-electron ground state energies 686 two-electron cases 685, 686 four terms 685 two-phase alloys 586 amorphous 630 nanogranular 586–7 nanostructured 629–30 with amorphous matrix 630–1 two-phase deformation models 540 two-phase nanostructure 663 two-phase nanostructured materials 565, 607 structure formation and mechanical behavior 565–664 Zr-based 596 U US Navy thermal spray applications 99–102 US Pressurized Water Reactors 246 ultimate tensile strength 257, 555 ultrafine Ni particles saturation magnetization 249 ultrafine particles 248 ultrafine structures 236 ultrahigh hardness 318, 565 ultrahigh-pressure sintering 213 ultrahigh strength 565 ultrasonic irradiation 13 ultrasonic shot peening 161 unagglomerated particles 7 uniaxial anisotropy 514 uniaxial hot pressing 572 uniform pore size and distribution 192 unpolymerized vesicles 27 HRTEM micrograph 28

Index V V metal powders 129 vacancy diffusion 194 vacuum plasma spraying (VPS) 94 van der Waals force 6 vapor condensation 57 deposition 57 transport 307 vaporization, laser 57–8, 378 VCSEL see vertical-cavity surfaceemitting laser vertical-cavity surface-emitting laser (VCSEL) 708–10 vesicle-mediated approach 27 vesicles 27 VFT see Vogel-Fulcher-Tammann vibrating sample magnetometry (VSM) 446 vibrational properties 318–19 Vickers hardness 253, 630, 635, 637, 640, 662 Vickers microhardness 551 viscosity data 655 measurements 646, 654, 655 viscous flow 178, 180, 181 behavior 640 viscous matrices sintering 205 visible light emission 699 Vogel-Fulcher-Tammann (VFT) 657, 658 equation 657 relation 658 temperature 657 void formation 257 volatile precursors 13 volume diffusivity 333 self-diffusion 331 volume diffusion 182, 184, 193, 197, 333, 352–6 vVolume fractions intercrystalline 242, 254 β-VOPO4 378 γ-VOPO4 378

759 VPO powders 377 VPS see vacuum plasma spraying VSM see vibrating sample magnetometry W W metal powders 129 warm compaction 190–1, 205 Warren-Averbach analysis 124 methods 201 water-in-oil microemulsions 26 Watts bath direct current plating 239–40 wear debris Fe-based 121 wear properties 263–5 wear resistance 130, 164 wear situations 122 wet compaction 188 wetting 188 Williamson and Hall methods 124 wire drawing 143 wires electrochemical thinning 255 work hardening 161, 189, 257 X x-ray amorphous structure 126 analysis 124 results 202 spectrum 159 x-ray diffraction (XRD) 123, 130, 140, 146, 152, 160, 162, 186, 201, 202, 241, 241, 354, 409, 489, 508, 508–11, 515, 521, 522, 550, 585, 585, 593, 598, 599, 599, 604, 606, 612, 613, 615, 616, 618–23, 627, 645, 653, 659, 659 analysis 585, 616 grazing angle 696 measurements 550, 593 methods 123, 124, 146 pattern 123, 159, 162, 239, 495 scan 598

760 x-ray diffraction (XRD) (cont’d) powder 241 scans 241 techniques 201 x-ray line broadening 123, 125 x-ray photoelectron spectroscopy (XPS) 371 analysis 390, 407 x-ray scattering 61, 580 x-ray spectra 126, 127 wide angle 131 XAS see soft x-ray absorption XPS see x-ray photoelectron spectroscopy XRD see x-ray diffraction xylose hydrogenation of 375 Y y-stabilized zirconia (YSZ) oxygen diffusivities 177, 193, 352–4, 356 yield strength 247, 253 experimental measurements 538 yield stress 555 and grain size 539 yielding conventional description 566 Young’s modulus 139, 247, 248, 251–3, 278, 281–3, 553, 554, 584, 615, 631–3, 635, 637, 641, 642, 649–52 see also elastic moduli YSZ see y-stabilized zirconia oxygen diffusivities yttria-stabilized zirconia see y-stabilized zirconia Z Zener drag 199, 273 process 574 zeolites 23 zirconia 179

Index Zn—Ni 240 Zr α-Zr 126 alloys 159 metal powders 129 Zr-based alloys 591–6, 598, 601, 602, 627, 641, 651 amorphous 595 bulk glass forming 602, 641 with nanoscale precipitates 601–7 mechanical behavior 641 glass forming ability 601 glassy and partially crystallized cast material 641 good ductility 651 high hardness values 660 high strengths 647 multicomponent 593, 594, 601, 624–8, 641 nanoscale compound formation 596–7 oxygen-induced precipitation of nanoscale compound particles upon crystallization 593–5 Zr-based bulk composites slowly cooled 607–10 containing ductile phases 647–50 Zr-based composites 607 W-containing 656 Zr-based metallic glass matrix composites 626–8 Zr-doped Pd 345 Zr—Ti—Cu—Ni metallic glasses 580 ZrC particles 129 micrometer-sized 656, 657 ZrO2 cationic self-diffusion 356 n-ZrO2 undoped 353, 354 yttria stabilized 197