Physical Metallurgy Handbook (Mcgraw-Hill Handbooks) [1 ed.] 0070579865, 9780070579866

Citation preview

PHYSICAL METALLURGY HANDBOOK Anil Kumar Sinha

McGRAW-HILL New York Chicago San Francisco Lisbon London Madrid Mexico City Milan New Delhi San Juan Seoul Singapore Sydney Toronto

Cataloging-in-Publication Data is on file with the Library of Congress

Copyright © 2003 by The McGraw-Hill Companies, Inc. All rights reserved. Printed in the United States of America. Except as permitted under the United States Copyright Act of 1976, no part of this publication may be reproduced or distributed in any form or by any means, or stored in a data base or retrieval system, without the prior written permission of the publisher. 1 2 3 4 5 6 7 8 9 0

DOC/DOC 0 9 8 7 6 5 4 3 2

ISBN 0-07-057986-5

The sponsoring editor for this book was Kenneth P. McCombs, the editing supervisor was David E. Fogarty, and the production supervisor was Pamela A. Pelton. It was set in Times Roman by SNP Best-set Typesetter Ltd., Hong Kong. Printed and bound by RR Donnelley.

This book is printed on acid-free paper. McGraw-Hill books are available at special quantity discounts to use as premiums and sales promotions, or for use in corporate training programs. For more information, please write to the Director of Special Sales, Professional Publishing, McGraw-Hill, Two Penn Plaza, New York, NY 10121-2298. Or contact your local bookstore.

Information contained in this work has been obtained by The McGraw-Hill Companies, Inc. (“McGraw-Hill”) from sources believed to be reliable. However, neither McGraw-Hill nor its authors guarantee the accuracy or completeness of any information published herein, and neither McGraw-Hill nor its authors shall be responsible for any errors, omissions, or damages arising out of use of this information. This work is published with the understanding that McGraw-Hill and its authors are supplying information but are not attempting to render engineering or other professional services. If such services are required, the assistance of an appropriate professional should be sought.

PREFACE

Physical Metallurgy Handbook is an enlarged edition of my earlier Ferrous Physical Metallurgy (Butterworths, 1989). Four new chapters of increased significance, namely, diffusion in metals and alloys, solidification, surface modification and thin film deposition, and thermal spray coatings have been added in addition to the complete revision of all 15 chapters. As before this Handbook focuses on both the theoretical elements, such as those dealing with diffusion, solidification, deformation, annealing phenomena, nucleation in solids, phase transformation in solids, kinetics of phase transformations, structure-property relationships, and the processing elements such as dealing with heat treating operations, surface modification and thin film deposition, and thermal spray coatings. This Handbook covers mostly ferrous and nonferrous alloys. Chapter 1 is devoted to iron-carbon alloys and their relevant phase diagrams, effects of alloying elements, and classification of steels and cast irons. Chapter 2 is on diffusion in metals and alloys, and it deals with the diffusion equation, diffusion mechanisms, the effect of key variables on diffusivity, self-diffusion, diffusion in substitutional and concentrated alloys, electro- and thermomigration; diffusion along short circuits, diffusion in ionic solids and semiconductors, and radiation effects and diffusion. Chapter 3 describes heat transfer in solidification; nucleation and growth; plane front solidification of alloys; cellular and dendritic growth; eutectics, monotectics, and peritectics; segregation; solidification processes and cast structures; single crystal growth; and grain refinement and eutectic modification. Chapter 4 treats tensile properties, yielding phenomena, flow stress, cold working, slip and twinning, strain aging, and deformation texture in engineering materials. Chapter 5 focuses on release of stored energy, recovery, recrystallization, laws of recrystallization, recrystallization texture, and grain growth. Chapter 6 deals with classical nucleation theories, nonclassical nucleation (i.e., spinodal decomposition), precipitation hardening in a variety of ferrous and nonferrous alloys, and strength of precipitation hardened alloys. Chapter 7 explains both the mechanisms and kinetics of transformation of austenite into pearlite, interphase precipitation, fibrous carbide, proeutectoid phases, ferrite morphology, proeutectoid cementite, and ferritepearlite and pearlitic steels. Chapter 8 introduces general and typical characteristics of martensitic transformation, ferrous and nonferrous martensites, nucleation and growth in martensitic transformation, thermoelastic martensitic transformation, strengthening mechanisms, toughness of martensite, and omega transformation. Chapter 9 emphasizes three definitions of bainite based on microstructure, surface relief, and kinetics; overall mechanisms; acicular ferrite; and bainitic steels. Chapter 10 elaborates on the formation of austenite and retained austenite, austenitic grain size, austenitic and superaustenitic stainless steels, duplex and superduplex stainless steels, and physical properties and machinability of stainless steels. Chapter 11 details isothermal and continuous cooling transformation diagrams. Chapter 12 discusses the definition, importance, selection, and classification of heat treatment; annealing, normalizing, decarburization, and graphitization of steels; annealing and ix

x

PREFACE

normalizing of cast irons; engineering properties and applications of cast irons; and structure-property relations in gray iron. Chapter 13 expounds quenching, quenchhardening, and inverse-quench hardening of steels; direct quenching; intense quenching and martempering of steels; austempering of steels and ductile iron; quench cracking; hardenability and hardenabilty steels; and alloy steel selection based on hardeanbility. Chapter 14 offers a detailed treatment on tempering, the structural and mechanical property changes associated with tempering of hardened steels, secondary hardening of steels, the tempering parameter, tempering methods, strengthening mechanisms of tempered martensite and bainite, various types of embrittlement phenomena occurring in low alloy quenched and tempered steels, and maraging steels. Chapter 15 gives an account of ferrous and nonferrous thermomechanical treatments, superplasticity largely found in ferrous and nonferrous alloys, and potential applications in aerospace industries. Chapter 16 thoroughly covers various surface hardening heat treatments and their advantages, disadvantages, and applications as well as newer processes such as supercarburizing, boriding, and the thermoreactive deposition/diffusion (TRD) processes. Chapter 17 concentrates on overheating and burning of low alloy steels, residual stresses, distortion in heat treatment, and the importance of correct design to lessen distortion and the danger of cracking. Chapters 18 dwells on, in detail, ion beam processes, physical vapor deposition, molecular beam epitaxy, and chemical vapor deposition. The final chapter describes the advantages, disadvantages, important processes, recent developments, coating characteristics, and applications of thermal spray techniques. The purpose of this Handbook is to present to the readers the latest information on fundamental principles, alloy design, and technologically useful microstructures, properties, forms, and applications of ferrous and nonferrous materials. This Handbook, which describes physical metallurgy with a novel approach and a comprehensive treatment, will serve as a valuable tool in understanding the interplay between microstructure, properties, and performance of a variety of engineering materials; in the selection of materials, treatments, and processes for specific applications; in solving heat treatment, surface modification, and other processing problems; in the tradeoff decisions that are often made in the automotive, aerospace, and other metalworking industries; and in the design, product development, and materials engineering of components that must operate reliably under service conditions. The intended audience of this Handbook includes practicing materials scientists; practicing manufacturing, mechanical, metallurgical, and product engineers; design engineers; researchers; heat treaters; sophisticated coaters; senior undergraduate students; and beginning graduate students. Academic courses for which the book might be useful as a text or for collateral readings are physical metallurgy, ferrous physical metallurgy, phase transformations, heat treatment of ferrous and nonferrous alloys, surface modification and thin film deposition, and thermal spray coatings. The exhaustive lists of references provided at the end of each chapter will enable readers to pursue the subject in still greater detail. The abundance of figures and tables provided in the text will be useful for better comprehension of the concepts of physical metallurgy. Anil Kumar Sinha

ACKNOWLEDGMENTS

I acknowledge with gratitude the helpful comments and valuable advice on various sections of the Handbook provided by Professor W. C. Leslie and C. R. Brooks (Chapter 1); Professors R. W. Balluffi, P. G. Shewmon, U. Gösele, and T. Y. Tan and Drs. M. C. Petri and E. P. Simonen (Chapter 2); Professors M. C. Flemings, R. K. Trivedi, J. D. Verhoeven, W. Kurz, and A. Hellawell and Dr. K. P. Young (Chapter 3); Professor F. B. Pickering and Drs. M. A. Imam, C. S. Pande, H. Jones, Z. Zimerman (Chapter 4); Professors R. W. Cahn, R. D. Doherty, T. Gladman, and C. L. Briant and Drs. B. B. Rath, and B. P. Bewlay (Chapter 5); Professors T. H. Sanders,W. A. Soffa, and A. J. Ardell and Drs. J. F. Grubb and Terry Tebold (Chapter 6); Professors G. J. Shiflet, F. B. Pickering, Paul Clayton, M. R. Notis, and T. Gladman and Drs. Bruce L. Bramfitt and Roger K. Steele (Chapter 7); Professor G. B. Olson and F. B. Pickering and Dr. L. McD. Schetky (Chapter 8); Professors H. I. Aaronson, H. K. D. H. Bhadeshia, and F. B. Pickering and Drs. Bruce L. Bramfitt and Roger K. Steele (Chapter 9); Drs. Riad Asfahani, and J. F. Grubb (Chapter 10); Dr. R. Vishwanathan (Chapter 12); Professor J. S. Kirkaldy, Drs. B. M. Kapadia and R. W. Foreman, and Messrs. R. R. Blackwood, R. Keogh, and Rick Houghton (Chapter 13); Professors R. A. Oriani and C. J. McMahon, Jr. and Drs. A. M. Sherman, K. A. Taylor, Michael L. Schmidt, Terry Tebold, James M. Dahl, and J. H. Bulloch (Chapter 14); Professors C. M. Sellars, T. Gladman, D. C. Dunand, F. B. Pickering, and J. C. Pilling and E. M. Taleef and Drs. Jeffrey Wadsworth, Steve Madeiro, Noshir M. Bhathena, and Rolf G. Sundberg (Chapter 15); Drs. Tohru Arai, V. S. Nemkov, V. I. Rudnev, C. A. Stickels, R. W. Foreman, David Pye, H.-J. Hunger, R. Bakish, Joarchim Bosslet, and W. K. Liliental and Messrs. R. C. Goldstein, Tom Sterner, Steven Verhoff, Mike Ives, Rick Houghton, M. M. Stirrine, Joseph Greene, and J. R. Easterday (Chapter 16); Messrs. G. Parrish and W. T. Cook (Chapter 17); Professors R. L. Boxman, Markus Pessa, Deepak G. Bhat, and William Rees, Jr., and Drs. Gary Tompa, Robert Aharonov, D. M. Mattox, Bruce Sartwell, Dennis Teer, A. J. Armini, and Angel Sanjurjo (Chapter 18); and Professor Lech Pawlowski and Drs. R. C. Tucker, Jr. and Richard Knight (Chapter 19). I also acknowledge many societies and publishers for their generous permission to use figures, photographs, and tables in this Handbook. Thanks are due to the management and staff of McGraw Hill for their editorial and administrative contributions to the production of this book. The author would like to express his appreciation to Messrs. Mark J. Eriksen and Roy Smith of Winona State University Library for their dedicated help in getting articles, books, and a majority of reference materials. Finally, I wish to pay tribute to my wife Priti, son Manish, daughterin-law Rashmi, and daughter Shruti for their understanding, love, moral support, and sacrifice, without which this book would not have been completed.

xi

ABOUT THE AUTHOR

Anil Kumar Sinha, Ph.D., M. Tech, is a former professor of metallurgical engineering at Notre Dame University, Cornell University, the University of Wisconsin, and Ranchi University. He also worked at Peerless Chain Company as Staff Metallurgist, Thompson Steel Company, Inc. as Manager of Metallurgy and Quality Control, Bohn Engine & Foundry as Senior Metallurgist and Chief Metallurgist, and National Metallurgical Laboratory as Senior Research Fellow. Currently president of Computer Wire EDM Corporation and a consultant, he is the author of an earlier version of this book, Ferrous Physical Metallurgy, as well as another book, Powder Metallurgy, and sixteen research papers. Dr. Sinha earned his Ph.D. at the University of Minnesota and his Master of Technology in Physical Metallurgy from the Indian Institute of Technology.

xii

CONTENTS

Preface ix Acknowledgments About the Author

xi xii

Chapter 1. Iron-Carbon Alloys

1.1

1.1. Introduction 1.1 1.2. Crystal Structures of Iron and Iron-Carbon Alloys 1.1 1.3. Phase Diagram 1.3 1.4. Critical Temperatures 1.7 1.5. Slowly Cooled Plain-Carbon Steels 1.9 1.6. Solubility of Carbon and Nitrogen in Ferrite and Austenite 1.7. Effects of Alloying Elements 1.14 1.8. Steel Classifications 1.26 1.9. Designations for Steels 1.51 1.10. Cast Iron Classifications 1.53 1.11. Alloying Elements in Gray Cast Iron 1.66 1.12. Alloying Elements in Ductile Iron 1.68 References 1.69

1.12

Chapter 2. Diffusion in Metals and Alloys 2.1. Introduction 2.1 2.2. Diffusion Equation 2.2 2.3. Atomistic Diffusion Mechanisms 2.26 2.4. Random Walk Theory of Microscopic Diffusion 2.32 2.5. Vacancy Diffusion 2.33 2.6. Effect of Key Variables on Diffusivity 2.40 2.7. Types of Diffusion Coefficients 2.50 2.8. Self-Diffusion in Pure Metals 2.53 2.9. Diffusion in Dilute Substitutional Alloys 2.56 2.10. Diffusion in Concentrated Alloys 2.58 2.11. Electro- and Thermomigration 2.69 2.12. Diffusion along Short Circuits 2.76 2.13. Application of Thin Films to Diffusion Study 2.92 2.14. Diffusion in Ionic Solids 2.93 2.15. Diffusion and Diffusion-Induced Defects in Semiconductors 2.16. Radiation Effects and Diffusion 2.103 References 2.112

iii

2.1

2.96

iv

CONTENTS

Chapter 3. Solidification

3.1

3.1. Introduction 3.1 3.2. Heat Transfer in Solidification 3.1 3.3. Nucleation during solidification 3.7 3.4. Growth from Melt 3.14 3.5. Interface Growth 3.24 3.6. Plane Front Solidification of Alloys 3.26 3.7. Cellular and Dendritic Growth 3.41 3.8. Solidification of Eutectics, Monotectics, and Peritectics 3.54 3.9. Segregation 3.78 3.10. Solidification Processes and Cast Structures 3.93 3.11. Single-Crystal Growth, Grain Refinement, and Eutectic Modification 3.12. New Solidification Processes 3.141 3.13. Directional Solidification Processing 3.163 References 3.165

Chapter 4. Plastic Deformation

3.115

4.1

4.1. Introduction 4.1 4.2. Tensile Properties 4.1 4.3. Yielding and Plastic Flow 4.12 4.4. Flow Stress 4.18 4.5. Cold Working 4.27 4.6. Strain Aging 4.28 4.7. Deformation Twinning 4.33 4.8. Deformation Texture 4.44 References 4.49

Chapter 5. Recovery, Recrystallization, and Grain Growth 5.1. Introduction 5.1 5.2. Release of Stored Energy 5.1 5.3. Recovery 5.3 5.4. Recrystallization 5.12 5.5. Recrystallization of Two-Phase Alloys 5.6. Recrystallization Temperature 5.29 5.7. Laws of Recrystallization 5.32 5.8. Recrystallization Texture 5.33 5.9. Grain Growth 5.39 References 5.55

5.1

5.27

Chapter 6. Nucleation in Solids 6.1. Introduction 6.1 6.2. Classical Homogeneous Nucleation 6.2 6.3. Classical Homogeneous Nucleation Rate in Solids 6.4. Heterogeneous Nucleation 6.8 6.5. Mechanism of Loss of Coherency 6.15 6.6. Spinodal Decomposition 6.15 6.7. Precipitation Hardening 6.28 6.8. Strength of Precipitation-Hardened Alloys 6.57 References 6.70

6.1

6.7

CONTENTS

v

Chapter 7. Pearlite and Proeutectoid Phases 7.1. Introduction 7.1 7.2. Pearlite 7.1 7.3. Interphase Precipitation 7.15 7.4. Fibrous Carbides 7.20 7.5. Proeutectoid Phases 7.21 7.6. Ferrite Morphology 7.21 7.7. Proeutectoid Cementite 7.41 7.8. Ferrite-Pearlite and Pearlitic Steels References 7.63

7.1

7.43

Chapter 8. Martensite

8.1

8.1. Introduction 8.1 8.2. General Characteristics of Martensitic Transformation 8.3. Typical Characteristics of Martensitic Transformation 8.4. Ferrous Martensite 8.24 8.5. Nonferrous Martensite 8.36 8.6. Nucleation and Growth in Martensitic Transformation 8.7. Thermoelastic Martensitic Transformation 8.45 8.8. Strengthening Mechanisms 8.75 8.9. Toughness of Martensite 8.81 8.10. Omega Transformation 8.81 References 8.83

Chapter 9. Bainite 9.1. Introduction 9.1 9.2. Definitions of Bainite 9.2 9.3. Overall Mechanisms 9.21 9.4. Acicular Ferrite 9.24 9.5. Bainitic Steels 9.29 9.6. Weldability of Bainitic Steels References 9.45

8.3 8.9 8.39

9.1

9.44

Chapter 10. Austenite 10.1. Introduction 10.1 10.2. Austenite Formation 10.1 10.3. Austenite Grain Size 10.7 10.4. Retained Austenite 10.14 10.5. Austenitic Stainless Steels 10.19 10.6. Superaustenitic Stainless Steels 10.86 10.7. Duplex Stainless Steels 10.88 10.8. Superduplex Stainless Steels 10.98 10.9. Physical Properties of Stainless Steels 10.99 10.10. Machinability of Stainless Steels 10.104 References 10.107

10.1

CONTENTS

vi

Chapter 11. Isothermal and Continuous Transformation Diagrams 11.1. Introduction 11.1 11.2. Isothermal Transformation Diagrams 11.1 11.3. Continuous Cooling Transformation Diagrams 11.4. Conclusion 11.19 References 11.21

11.1

11.10

Chapter 12. Basic Heat Treatment

12.1

12.1. Introduction 12.1 12.2. Importance, Selection, and Classification of Heat Treatment 12.1 12.3. Annealing of Steels 12.3 12.4. Decarburization 12.20 12.5. (Secondary) Graphitization of Steels 12.21 12.6. Normalizing 12.23 12.7. Anncaling and Normalizing of Cast Irons 12.27 12.8. Engineering Properties and Applications of Cast Irons 12.36 12.9. Structure-Property Relations in Gray Iron 12.67 References 12.67

Chapter 13. Hardening and Hardenability

13.1

13.1. Introduction 13.1 13.2. Quenching 13.1 13.3. Quench-Hardening 13.65 13.4. Inverse Quench-Hardening of Steel 13.68 13.5. Direct Quenching 13.68 13.6. Intense Quenching of Steel 13.69 13.7. Martempering of Steel 13.72 13.8. Martempering of Gray Cast Iron 13.74 13.9. Austempering of Steel 13.77 13.10. Austempering of Ductile Iron 13.86 13.11. Quench Cracking 13.94 13.12. Effect of Carbon on Hardness in Hardened Steel 13.96 13.13. Hardenability 13.98 13.14. Hardenability (or H-) Steels 13.133 13.15. Alloy Steel Selection Guide Based on Hardenability 13.140 References 13.147

Chapter 14. Tempering 14.1. 14.2. 14.3. 14.4. 14.5. 14.6. 14.7. 14.8. 14.9. 14.10.

Introduction 14.1 Structural Changes on Tempering 14.1 Mechanical Property Changes on Tempering 14.20 Effects of Alloying Elements on Tempering 14.24 Secondary Hardening 14.26 Secondary Hardening Steels 14.28 Nucleation and Growth of Alloy Carbides 14.45 Effect of Time and Temperature on Tempering (or Tempering Parameter) Methods of Tempering 14.50 Strengthening Mechanisms of Tempered Martensite and Bainite 14.59

14.1

14.45

CONTENTS

14.11. Thermally Induced Embrittlement Phenomena 14.12. Hydrogen Damage of Steels 14.84 14.13. Metal-Induced Embrittlement 14.102 14.14. Maraging Steels 14.108 References 14.119

vii 14.65

Chapter 15. Thermomechanical Treatment

15.1

15.1. Introduction 15.1 15.2. Ferrous TMT 15.1 15.3. High-Strength Low-Alloy Steels 15.31 15.4. TRIP or Multiphase Steels 15.39 15.5. Dual-Phase Steels 15.45 15.6. Ultrahigh-Strength Low-Alloy Steels 15.52 15.7. Ultrahigh-Carbon (UHC) Steels 15.53 15.8. Ductile Iron 15.56 15.9. Nonferrous TMT 15.56 15.10. Superplasticity 15.78 15.11. Superplastic Sheet Forming Process 15.99 References 15.110

Chapter 16. Surface Hardening Treatments

16.1

16.1. Introduction 16.1 16.2. Thermal Surface Hardening 16.3 16.3. Austenitic (or High-Temperature) Thermochemical Surface Hardening Treatment 16.42 16.4. Ferritic Thermochemical Surface Hardening Treatments 16.81 16.5. Supercarburizing 16.111 16.6. Boriding (or Boronizing) 16.114 16.7. Thermoreactive Deposition/Diffusion (TRD) Process 16.138 References 16.144

Chapter 17. Defects and Distortion in Heat-Treated Parts 17.1. Introduction 17.1 17.2. Overheating and Burning of Low-Alloy Steels 17.3. Residual Stresses 17.9 17.4. Quench Cracking 17.28 17.5. Distortion in Heat Treatment 17.34 17.6. Importance of Design 17.52 References 17.53

17.2

Chapter 18. Surface Modification and Thin-Film Deposition 18.1. Introduction 18.1 18.2. Ion Beam Processes 18.1 18.3. Physical Vapor Deposition 18.4. Chemical Vapor Deposition References 18.112

18.21 18.57

17.1

18.1

CONTENTS

viii

Chapter 19. Thermal Spray Coatings 19.1. Introduction 19.1 19.2. Advantages and Disadvantages 19.3. Processes 19.6 19.4. Post-Spray Treatment 19.24 19.5. Coating Characteristics 19.33 19.6. Applications 19.36 References 19.51

19.1

19.2

Appendix A. Conversion Table for Units, Constants, and Factors in Common Use

A.1

Appendix B. Temperature Conversions

B.1

Index follows Appendix B

CHAPTER 1

IRON-CARBON ALLOYS

1.1 INTRODUCTION Steels are the most complex and widely used engineering materials because of the abundance of iron in the earth’s crust, high melting temperature of iron (1534°C), and wide range of mechanical properties and associated microstructures produced by solid-state phase transformations by varying the cooling rate from the austenitic condition. The iron-cementite phase diagram is the very useful foundation on which analysis of all steel heat treating processes depends, whereas both iron-cementite and iron-graphite diagrams are useful for the heat treatment of cast iron. The phase diagram is a map showing structures or phases and phase boundaries present as the temperature and overall composition of the alloy are varied under constant pressure (usually 1 atm). This chapter deals with structures of iron and iron-carbon alloys, iron-cementite and iron-graphite phase diagrams, critical temperatures, solubilities of carbon and nitrogen in ferrite and austenite, effects of alloying elements, and classification of steels and cast irons.

1.2 CRYSTAL STRUCTURES OF IRON AND IRONCARBON ALLOYS 1.2.1 Alpha-Iron Pure iron, the carbon steels, and other metals such as V, Cr, Mo, and W have the body-centered cubic (bcc) structure at room temperature, which is characterized by the unit cell shown in Fig. 1.1.1 The bcc structure of pure iron at room temperature, called either a-iron or ferrite, has one atom at the center of the cube and an atom at each corner of the unit cell and constitutes 1 + (8 ¥ 1/8) = 2 atoms per unit cell. The atomic packing factor for this structure is 0.68 and represents the volume fraction of the unit cell occupied by two atoms. The lattice parameter of a-iron at room temperature is 2.86 Å. The ferrite is a more open or less dense structure than the other structural modification of iron, called either gamma-iron or austenite. The difference in atomic packing of a- and g-iron is responsible for volume contraction that takes place on heating low-density ferrite to higher-density austenite. Austenite is properly used for g-iron with carbon in solid solution. Both ferrite and austenite are quite soft and ductile. Their average properties include tensile strength, 40,000 psi; elongation, 40%; hardness, 150 BHN, RC-0, or less than RB-90. Ferrite 1.1

1.2

CHAPTER ONE

FIGURE 1.1 Body-centered cubic structure of a metal showing (a) the atomic-site unit cell, (b) the isolated unit cell,1 (c) model made from hard balls. [(a) and (b) reprinted by permission of McGraw-Hill, New York; (c) reprinted with permission from B. A. Rogers, The Nature of Metals, 2d edition, ASM, Metals Park, Ohio.]

is ferromagnetic below 768°C (1414°F) and paramagnetic in the temperature range of 768 to 910°C (1414 to 1670°F). The temperature at which this magnetic transformation takes place is called the Curie temperature.

1.2.2 Gamma-Iron Austenite as well as other metals such as Al, Ni, Cu, Ag, Pt, and Au have the closepacked face-centered cubic (fcc) structure. Its unit cell is shown in Fig. 1.21 and has an atom at each corner and an atom at the center of each face. Each corner atom is shared by eight unit cells that come together at the corner, while each face atom is shared by two adjacent unit cells. Thus there are (8 ¥ 1/8) + (6 ¥ 1/2) = 1 + 3 = 4 atoms per unit cell. The atomic packing factor for this structure is 0.74. The lattice parameter of austenite is 3.57 Å, which is larger than that of ferrite. Gamma-iron is the stable form of pure iron in the temperature range of 910 to 1393°C (1670 to 2540°F). Unlike ferrite, austenite is paramagnetic. It is also soft and ductile.

1.2.3 Delta-Iron The third phase that occurs in pure iron is d-iron or ferrite with a bcc structure which is crystallographically similar to alpha-iron. Delta-iron is stable at temperature between 1393 and 1534°C (2540 and 2793°F). Its lattice parameter is 2.89 Å; it is also soft and ductile, and its hardness and elongation are similar to those of ferrite and austenite in their stable forms. The d-ferrite is of no significance in heat treating practice used for plain carbon and low-alloy steels.

IRON-CARBON ALLOYS

1.3

FIGURE 1.2 Face-centered cubic structure of a metal showing (a) the atomic-site unit cell, (b) the isolated unit cell,1 (c) model made from hard balls. [(a) and (b) reprinted by permission of McGraw-Hill, New York; (c) reprinted with permission from B. A. Rogers,The Nature of Metals, 2d edition, ASM, Metals Park, Ohio.]

1.2.4 Cementite Cementite, represented by the formula Fe3C, is a metastable Fe-C intermetallic compound. It has a negligible solubility limit in a-iron and contains 6.67 wt % (29 at %) carbon. It is ferromagnetic with a Curie temperature of 215°C (419°F). In sharp contrast to ferrite and austenite, cementite is hard (BHN over 700, VPH 1300) and brittle (0% elongation). This is an interstitial compound of low tensile strength (~5 ksi) but high compressive strength. It is the hardest structure that appears on the Fe-Fe3C phase diagram. It plays an important role in the hardening of many commercial steels. It has an orthorhombic crystal structure with lattice limit: a = 4.52 Å, b = 5.09 Å, and c = 6.74 Å and 12 iron and 4 carbon atoms per unit cell.

1.3 PHASE DIAGRAM A phase is a portion of a system whose properties, composition, and crystal structure are uniform or homogeneous and which is separated from the remainder by distinct bounding surfaces. By the word system we mean an isolated and homogeneous portion of matter, and the components of a system are the metallic elements* that constitute or form the system. A one-component system is a single metallic element (e.g., pure iron); a two-component system is a mixture of two metallic elements, called binary alloys; and three-component systems are mixtures of three metallic elements, called ternary alloys.2 In a phase diagram (also called equilibrium diagram or constitutional diagram), temperature is plotted vertically and composition horizontally at constant (atmos* The elements need not be metallic; however, this term has been used to emphasize the metallic system that is of immediate significance to us.

CHAPTER ONE

1.4

pheric) pressure. Figure 1.3a and b shows the conventional and modified versions, respectively, of the phase diagram, where each part shows both the metastable Fe-Fe3C and stable or equilibrium Fe-graphite diagram; the former is indicated by full lines, and the latter is indicated by dashed lines.3,4 The phases present in Fe-Fe3C and Fe-C diagrams are molten alloy, austenite, a-ferrite, d-ferrite, cementite, and graphite. These phases are alternatively called constituents. However, not all constituents are single phases, but rather are a mixture of two phases—ferrite and cementite like pearlite and bainite. There is a slight difference between the two sets of diagrams in both temperatures and compositions corresponding to the critical points and the reaction curves. Readers may use either version. However, the conventional version of these diagrams will be used hereafter in the entire text because of its wider use and greater familiarity. The iron-iron carbide diagram is not a true equilibrium diagram but rather a metastable equilibrium diagram because cementite is a metastable phase. Given a very long period, cementite will decompose into more stable, or equilibrium, phases of graphite and iron. However, once cementite is formed, it is very stable and may be treated for all practical purposes as an equilibrium phase.2 A study of Fe-Fe3C and Fe-C diagrams is valuable in understanding the heat treatment, accurate determination of phase compositions, control of their properties and solid-state reactions in general, and the basic differences among iron alloys in particular. The liquidus curve ABCD or ABC¢D¢ shown in Fig. 1.3a represents the boundary between the two-phase regions and the liquid. The liquidus curve AB is practically a straight line joining the melting temperature of iron and the endpoint of the first isothermal reaction (point B). This horizontal line, HJB, at 1493°C represents the peritectic temperature. The peritectic reaction, occurring during solidification of carbon steels, may be expressed in the form: liquid (0.51 wt % C, point B) + solid (d-ferrite phase, point H) cooling

∫ heating austenite phase (0.16 wt % C, point J) Point J is called a peritectic point. The liquidus curve BC of the austenite phase ends at the eutectic horizontal: cooling liquid (4.3 wt % C, point C) ∫ heating g (2.06 wt % C, point E) + Fe3C (6.67 wt % C, point F). Thus the horizontal line ECF at 1147°C represents the eutectic temperature, and point C represents the eutectic point. The eutectic mixture of g and Fe3C is called ledeburite. The solidus curve AHJECF represents the boundary between the two-phase region and the solid. The horizontal line PSK at 723°C corresponds to the euteccooling toid reaction: g (0.8 wt % C, point S) ∫ heating a-ferrite (0.02 wt % C, point P) + Fe3C (6.67 wt % C, point K). This eutectoid mixture is a lamellar structure comprising alternate lamellae of ferrite and cementite to which the name pearlite has been given. The eutectoid point is represented by 0.8 wt % C content, point S. It is thus clear that the portion of the diagram that lies between the liquidus and solidus lines (ABCD and AHJECF, respectively) represents the solidification of the liquid solution, whereas the areas between GSECF and PSK as well as below PSK represent the decomposition of austenite on slow cooling. Although the Fe-Fe3C diagram extends from a temperature of 1925°C (3500°F) down to room temperature, the left-hand bottom portion of the diagram (Fig. 1.3a) which lies below 1035°C (1900°F) is commonly used for the heat treatment of steel because the steel heat treating practice seldom involves temperatures beyond this value.5 The large phase field of austenite shows solubility of carbon in this structure ranging from 0 to 2.06 wt % through 0.8 wt % carbon at 723°C (Fig. 1.3a). This

FIGURE 1.3 The Fe-Fe3C and Fe-C phase diagrams: (a) conventional3 and (b) modified.4 [(a) reprinted by permission of McGraw-Hill, New York; (b) reprinted by permission of ASM International, Materials Park, Ohio.]

1.6

CHAPTER ONE

maximum solubility of carbon content of 2.06 wt % corresponds to the boundary between steels and cast irons. Fe-C alloys with carbon content up to 2.06 wt % are arbitrarily classed as steels, and those beyond this amount are called cast irons. Actually, only in rare instances is steel used with more than 1.1 wt % carbon.2 The solubility of carbon in the a-iron phase field is very low, with a maximum of about 0.02 wt % carbon at the eutectoid temperature, 723°C, and it decreases with decreasing temperature until it is about 0.008 wt % at 0°C. This large difference in solubility of carbon in the two phases of iron is of great significance in the heat treatment of steel.6 It is customary to subdivide the steel range into hypoeutectoid and hypereutectoid depending on whether composition lies below or above the eutectoid composition. Likewise, the cast iron range may be subdivided into hypoeutectic and hypereutectic, if the carbon content in cast iron is below or above the eutectic composition, respectively. Steels and white cast irons obey the metastable Fe-Fe3C phase diagram, whereas other cast irons (consisting of graphite precipitates in a solid metal matrix, similar to steel) obey both the equilibrium Fe-graphite and metastable FeFe3C phase diagrams. Usually the carbon content varies from 2.2 to 4.5 wt % for the cast irons. The microstructures of the cast irons, which are dependent on the carbon content and cooling rate, may be deduced from both of the phase diagrams. However, it was realized later that the analysis of structures of cast irons is much more complex than that of steel and is much more sensitive to the processing conditions employed in their manufacture. It is noted here that commercial cast irons are not simple alloys of iron and carbon but, instead, contain appreciable amounts of other elements which have important and powerful effects on the structure of cast iron. The most important additional element is silicon ranging from 1.0 to 3.0%. Thus we may treat cast irons as the ternary alloy of Fe, C, and Si. However, cast irons usually contain minor additions of S, Mn, P, and trace elements such as Al, Sn, Sb, and Bi as well as gaseous elements H, N, and O. Cast irons may also differ in many respects from steels. For example, cast irons have low melting temperature, low ductility, and poor impact properties, which may restrict their applications. The molten cast iron is more fluid, is less reactive with air and molding materials, and is of relatively low cost when compared to steels or other common alloys. In addition, it can be readily machined. Thus it is an excellent engineering material.7,8 Another difference between cast irons and steels is the fact that cast iron properties are determined by four factors, i.e., chemical composition, inoculation, solidification rate, and cooling rate, while steel properties are controlled primarily by the chemical composition. The addition of Si promotes graphitization in cast iron. In other words, Si is the catalytic agent that permits free carbon (flakes, nodules, etc.) to appear in the microstructure. High temperature and the presence of silicon greater than 1% speed up the dissociation reaction of Fe3C, which may be written as follows:9 Fe3C æ æÆ 3Fe (austenite) + C (graphite)

(1.1)

As a consequence, cast irons may contain carbon in free form as graphite and in combined form as cementite. It is clear that cast iron can solidify according to either the stable Fe-graphite system (gray iron) or to metastable Fe-Fe3C system (white iron). As a result, the eutectic may be g-graphite or g-Fe3C (ledeburite). This also differentiates cast irons from steels because the latter possess only combined carbon as cementite. The formation of lower-density graphite during the solidification of

IRON-CARBON ALLOYS

1.7

iron castings causes the reduced or negligible volume change of the metal from liquid to solid. This permits the formation of very complex castings such as onepiece water-jacketed internal combustion engine blocks without any shrinkage voids in the metal.8 The shape and distribution of free graphite, rather than variations in composition, are commonly used to classify cast irons.

1.4 CRITICAL TEMPERATURES There are three transformation temperatures, often referred to as critical temperatures, which are of interest in heat treatment of steels. The temperature A1 is the eutectoid temperature of 723°C in the binary phase diagram which is the boundary between ferrite-cementite field and the austenite-ferrite or austenite-cementite field. Temperature A3 is the temperature at which a-iron transforms to g-iron, which, for pure iron, occurs at 910°C. The A3 line represents the boundary between the ferrite-austenite and austenite fields. Similarly, the Acm line is the boundary between the cementite-austenite and the austenite fields. The temperature difference between A1 and A3 is called the critical (temperature) range. Sometimes A1, A3, and Acm are written as Ae1, Ae3, and Aecm, respectively, denoting equilibrium conditions. These critical temperatures are detected by thermal analysis or dilatometry during heating or cooling cycles, and some thermal hysteresis (lag) is observed. The thermal hysteresis that occurs on heating is indicated by the letter c, representing the French word chauffage, meaning heating. Similarly, thermal hysteresis on cooling is indicated by the letter r, representing the French word refroidissement, meaning cooling. Thus there are two sets of critical temperatures: Ac1, Ac3, and Accm for heating and Ar1, Ar3, and Arcm for cooling. These sets of critical temperatures are shown in Fig. 1.4.9 The faster the rate of heating, the higher the Ac point: the faster the rate of cooling, the lower the Ar point. Thus the faster the heating and cooling rates, the larger the difference between the Ac and Ar points of the reversible equilibrium point A. Usually the critical temperatures which are necessary for the heat treatment of carbon and alloy steels can be known experimentally.10 However, empirical formulas that show the effects of alloying elements on the critical temperatures have been developed by regression analysis of large amounts of experimental data by Andrews,11 Grange,12 Kunitake et al.13,14 and Miyoshi et al.15 for calculating the practical Ac1 and Ac3 temperatures. For 0.08 ~ 1.4% C steel:11 Ac1 (∞C) = 723 - 10.7Mn - 16.9 Ni + 29.1Si + 16.9 Cr + 290 As + 6.38 W (1.2a) For 0.3 ~ 0.6% C low-alloy steel:12 Ac1 (∞C) = 723 - 13.9 Mn - 14.4 Ni + 22.2 Si + 23.3 Cr For 0.25 ~ 0.45% C low-alloy steel:

Ac1 (∞C) = 755 - 32.3 C - 17.8 Mn + 23.3 Si + 17.1 Cr + 4.5 Mo + 15.6 V For 0.10 ~ 0.55% C low-alloy steel:

(1.2b)

13

(1.2c)

14

Ac1 (∞C) = 751 - 16.3 C - 27.5Mn - 5.5 Cu - 15.9 Ni + 34.9Si + 12.7 Cr + 3.4Mo (1.2d)

CHAPTER ONE

1.8

940 E

1700 G

Ac1 and Ar1

900

Other data 1600

820

1500

Ar3 1400

Ac3

Accm

A3 Ac1

S

780

Arcm Acm Ac1

740 A1

A1 Ar1

Ar1

1300

0

Temperature, C

Temperature, F

860

0.2

0.8 0.6 Carbon, %

0.4

1.0

700 1.2

FIGURE 1.4 A portion of Fe-Fe3C diagram showing two sets of critical cooling temperatures: Ac1, Ac3, and Accm for heating and Ar1, Ar3, and Arcm for cooling. Rate of heating and cooling at 0.125°C/min.9 (Reprinted by permission of ASM International, Materials Park, Ohio.)

For 0.07 ~ 0.22% C low-alloy steel:15 Ac1 (∞ C) = 751 - 26.6C - 11.1Mn - 22.9Cu - 23.0 Ni + 17.6 Si + 24.1Cr + 22.5Mo - 39.7 V + 223 Nb - 169 Al - 895B

(1.2e)

For 0.08 ~ 1.4% C steel:11 Ac3 (∞C) = 910 - 203 C - 30 Mn - 20Cu - 15.2 Ni - 11Cr - 700 P + 44.7 Si + 31.5Mo + 104 V + 460 Al + 13.1W + 120 As

(1.3a)

For 0.3 ~ 0.6% C low-alloy steel:12 Ac 3 (∞C) = 854 - 179 C - 13.9Mn - 17.8 Cu - 1.7 Ni + 44.4Si For 0.25 ~ 0.45% C-Si-Cr-Mo-V low-alloy steel:

Ac 3 (∞C) = 930 - 395C - 14.4Mn + 55Si + 5.8 Ni + 24.5 Cr + 83.4Mo For 0.10 ~ 0.55% C low-alloy steel:

(1.3b)

13

(1.3c)

14

Ac 3 (∞C) = 881 - 206 C - 15 Mn - 26.5 Cu - 20.1 Ni - 0.7 Cr + 53.1Si + 41.1 V (1.3d)

IRON-CARBON ALLOYS

1.9

For 0.07 ~ 0.22% C low-alloy steel:15 Ac3 (∞C) = 937 - 476C - 19.7 Mn - 16.3Cu - 26.6 Ni - 4.9Cr (1.3e)

+ 56 Si + 38.1Mo + 12.5V - 19 Nb + 198 Al + 3315B

Among these empirical relations, the ones formulated by Andrews are widely adopted. There is good agreement between the calculated and observed temperatures.

1.5 SLOWLY COOLED PLAIN-CARBON STEELS 1.5.1 Eutectoid Steel Figure 1.5 shows the enlarged section of the Fe-Fe3C diagram. When a eutectoid steel is heated to the austenitizing temperature and held there for a sufficient time, its structure will become homogeneous austenite. On very slow cooling under conditions approaching equilibrium, the structure will remain that of austenite until just above the eutectoid temperature. At the eutectoid temperature or just below it, the entire structure of austenite will transform into pearlite. The ferrite and cementite that are incorporated in the pearlite are called eutectoid ferrite and eutectoid cementite, respectively. Figure 1.6 shows a light micrograph of pearlite advancing into the unstable austenite.16

1000 1800

910°C 1600

g 800 a+g

1400 a

723°C

Temperature, °F

Temperature, °C

900

0.8

700 0.02

a + carbide 600

0

0.2

0.4

0.6

0.8

Fe3C 1.0

1.2

1200

6.67

Weight percent carbon FIGURE 1.5 The eutectoid portion of the Fe-Fe3C diagram. (Reprinted by permission of Addison-Wesley Publishing Co., Reading, Massachusetts.)

CHAPTER ONE

1.10

FIGURE 1.6 Light micrograph of pearlite colony advancing into an austenite grain.16 (Courtesy of the Metallurgical Society, Warrendale, Pennsylvania; after J. R. Vilella.)

The amount of phases present in a two-phase field of a binary phase diagram can be determined by applying the lever rule. The alloy composition represents the fulcrum of a lever with the horizontal line, called a tie line, touching the two-phase field representing its length. The ends of the tie line fix the compositions of the coexisting phases, and the relative amounts of the phases are directly proportional to the length of the “opposite lever arm.”Thus, the amount of phases present in the pearlite formed just below 723°C can be easily computed by using the lever rule: wt % ferrite =

6.67 - 0.8 5.87 ¥ 100 = ¥ 100 = 88.27 ª 88 6.67 - 0.02 6.65

wt % cementite =

0.8 - 0.02 0.78 ¥ 100 = ¥ 100 = 11.73 ª 12 6.67 - 0.02 6.65

(1.4a) (1.4b)

Since the densities of ferrite and cementite, being 7.87 and 7.70 g/cm3, respectively, are very close, the lamellae of ferrite and cementite have respective widths of about 7.5 to 1.

1.5.2 Hypoeutectoid Steel When a hypoeutectoid steel is allowed to cool slowly after heating into the austenite phase field, that is, above the GS line (or A3 line) (Fig. 1.3), the primary or proeu-

IRON-CARBON ALLOYS

1.11

FIGURE 1.7 Structures of slowly cooled steels, 500¥: (a) hypoeutectoid steel, 0.45% C, showing ferrite (white areas) and pearlite (resolved and unresolved); (b) hypereutectoid steel, 0.9% C. The white network is cementite. [(b) reprinted by permission of Wadsworth, Inc., Belmont, California. Source: D. S. Clark and W. R. Varney, Physical Metallurgy for Engineers, Van Nostrand, New York, 1962.]

tectoid ferrite will begin to precipitate at the austenite grain boundaries at a temperature indicated by a point on the (a + g)/g phase boundary for the alloy composition concerned. As the temperature decreases further (proeutectoid) ferrite is formed and the carbon content of austenite coexisting in the two-phase (a + g) region continuously increases (due to rejection of excess carbon at the austenite/ferrite interface from the ferrite formed) until the eutectoid composition is reached at 723°C. On passing through a temperature of 723°C, the austenite will transform to pearlite. The final structure of slowly cooled hypoeutectoid steel will consist of primary or proeutectoid ferrite and pearlite (Fig. 1.7a); the proportion of the latter increases with the carbon content until, at 0.8% C, the structure will be completely pearlitic. Figure 1.5 shows that the proeutectoid ferrite in slowly cooled 0.4% plain carbon steel begins to form at about 800°C. It forms at the austenite grain boundaries. As the alloy is continuously cooled to a temperature just above the eutectoid temperature (say, 724°C), the weight percent proeutectoid ferrite and weight percent austenite can be determined by using the lever rule: wt % proeutectoid ferrite = wt % austenite =

0.80 - 0.40 ¥ 100 ª 50 0.80 - 0.02

0.40 - 0.02 ¥ 100 ª 50 0.80 - 0.02

Since all the remaining austenite will transform to pearlite at the eutectoid temperature, the weight percent of pearlite just below 723°C (say, 722°C) will be 50%, if conditions approaching equilibrium exist.

1.12

CHAPTER ONE

Since the solubility of carbon in ferrite decreases from 723°C to room temperature, further slow cooling to room temperature results in the precipitation of carbide from ferrite; however, its amount is small, and therefore it increases the overall hardness of the steel only to a slight extent which cannot be easily measured. Also, the slightly increased amount of carbide is difficult to detect in the microstructure. Hence, the amount of ferrite and carbide analyzed and calculated at just below 723°C can be considered to be valid at room temperature.17

1.5.3 Hypereutectoid Steel Consider the slow cooling of a hypereutectoid steel (say, 1.2% plain-carbon steel) from an austenitizing temperature of 920°C. The separation of proeutectoid cementite occurs at the grain boundaries of austenite beginning at a temperature of about 880°C and is completed at a temperature of 723°C. An increasing amount of primary cementite precipitates with decreasing temperature until a temperature of 723°C is reached, where the remaining austenite, depleted in carbon content, reaches its minimum level of 0.80%. The microstructure of slowly cooled hypereutectoid steel will consist of primary cementite and pearlite at any temperature below 723°C. At a temperature slightly above the eutectoid temperature, the weight percentage of proeutectoid cementite is [(1.20 - 0.80)/(6.67 - 0.80)] ¥ 100 = 6.8%, and that of the remainder austenite is [(6.67 - 1.20)/(6.67 - 0.80)] ¥ 100 = 93.2%. This austenite transforms to pearlite at or below 723°C. Thus the whole structure of a hypereutectoid steel will consist of primary cementite and pearlite at any temperature below 723°C. Figure 1.7b shows the microstructure of a slowly cooled hypereutectoid steel containing 0.9% C.

1.6 SOLUBILITY OF CARBON AND NITROGEN IN FERRITE AND AUSTENITE The atomic radii of carbon (0.8 Å) and nitrogen (0.7 Å) are much smaller than that of iron (1.28 Å), which allows these solute elements to enter into the interstices or “holes” of the a-iron and g-iron crystal lattices. On a hard sphere model for fcc austenite, the largest hole at an octahedral site is 0.52 Å in radius which is surrounded by six atoms located at the corners of a regular octahedron (Fig. 1.8a); the next largest hole is 0.28 Å in radius at a tetrahedral site which is surrounded by a tetrahedron of four atoms (Fig. 1.8b). In spite of the fact that the bcc ferrite is not densely packed, the octahedral site is only 0.19 Å in radius and is surrounded by six atoms at the corners of a slightly compressed octahedron (Fig. 1.9a) while the tetrahedral site is 0.36 Å in radius (Fig. 1.9b).18 Evidently, octahedral holes in a-iron are much smaller than those in g-iron. Since C and N are larger than the available interstices in either lattice, it is apparent that some distortion must ensue when they occupy the interstices of iron lattices. However, these interstitial atoms reside in octahedral sites, with an expansion caused by the displacement of two nearestneighbor iron atoms rather than four, as in tetrahedral sites. It is thus clear that the maximum solubility of carbon (2.0 wt %) and nitrogen (2.8 wt %) in g-iron is much greater than in a-iron as a result of much larger octahedral holes in austenite. This large difference in solubilities is of great significance in the heat treatment of steels and is fully exploited to improve strength. Nitrogen

IRON-CARBON ALLOYS

1.13

FIGURE 1.8 (a) Octahedral and (b) tetrahedral interstitial voids in fcc structure.18 (Courtesy C. S. Barrett and T. B. Massalski.)

FIGURE 1.9 (a) Octahedral and (b) tetrahedral interstitial voids in bcc structure.18 (Courtesy C. S. Barrett and T. B. Massalski.)

1.14

CHAPTER ONE

has more solubility than carbon in ferrite. That is why the nitriding heat treatment is usually performed at a temperature lower than the eutectoid temperature.

1.7 EFFECTS OF ALLOYING ELEMENTS 1.7.1 The g- and a-Phase Fields Steels contain alloying elements and impurities that must be associated with austenite, ferrite, and cementite. It has been pointed out that we can divide the alloying elements into two groups based on their influence on the phase diagram.19,20 1. By widening (or opening) the g-phase field and promoting the formation of austenite over larger compositional range. These elements are called austenite stabilizers or formers. 2. By shrinking (or closing) the g-phase field and promoting the formation of ferrite over a larger compositional range. These elements are called ferrite stabilizers or formers. Based on the alloying elements, the iron binary phase diagrams can be divided into four categories, as follows.19,21 Type 1: Open g-phase field. In this group the alloying elements, for example, Mn, Ni, Co, Ru, Rh, Pd, Os, Ir, and Pt, expand the temperature range for stable austenite by lowering the two-phase (a + g) region toward room temperature and raising the two-phase (d + g) zone to the melting range (Fig. 1.10a); that is, both Ae1 and Ae3 are depressed. It is also easier to obtain metastable g by quenching from the g region to room temperature. Consequently Mn and Ni are useful elements in the formulation of stainless steels. Type 2: Expanded g-phase field. This is the same as type 1 above except its range is shortened by iron-rich compound formation (Fig. 1.10b). Examples are C, N, Cu, Zn, and Au. Thus the presence of C and N expands the g-field to the extent that the solid solubility of C and N increases to 2.0 and 2.8 wt %, respectively, in the austenite. This effect underlines the whole of the heat treatment of steels. Type 3: Closed g-phase field. In this group the alloying elements restrict the temperature range for stable austenite, with the results that a smaller area of g-phase (called the gamma loop) and the continuous and wider d- and a-phase fields are obtained (Fig. 1.10c). This means that the alloying elements in this category promote the formation of ferrite, which include Si, Al, Be, and P, together with the strong carbide-forming elements Cr, Ti, V, Mo, and W. Type 4: Contracted g-phase field. In this group the a- and g-phase fields are bounded by a miscibility gap. That is, a- and g-solid solutions are in equilibrium with an intermetallic compound or solid solution (Fig. 1.10d). Boron is the most significant element in this group together with the carbide formers Ta, Zr, and Nb. These phenomena are associated with the crystal structure of the alloying elements since no fcc alloying element stabilizes ferrite and likewise no bcc element stabilizes austenite. Thus it appears from these phase diagrams that the crystal structure of solid solutions of iron at room temperature is the important basis for classifying steels. If austenite is predominant at room temperature because of the addition of sufficiently large amounts of Ni and Mn, it is called an austenitic steel. Examples are Hadfield steel containing 13% Mn, 1.2% Cr, and 1% C; 18% Cr-8% Ni austenitic stainless steel; and precipitation-hardening stainless steels with fine dis-

IRON-CARBON ALLOYS

d+M

MP d A4

d+g

Melt

MP d A4

1.15

Melt M+C

g+M

g+M g

g

a

g+C

g+a

A3

A3

a

a+g

a+C

Type 2

Type 1

(b)

(a)

MP A4

Melt d

A4 d+M

Melt

MP d

d+M

d+g

g

g a+g

g + a(d) A3

A3

a

Type 3

M+ C

g+M

a

g+C a+C

Type 4 (c)

(d)

FIGURE 1.10 Classification of equilibrium diagram for iron alloys: (a) open g-phase field; (b) expanded g-phase field; (c) closed g-phase field; (d) contracted g-phase field.6,21 (Reprinted by permission of ASM International, Materials Park, Ohio.)

persion of stable or metastable coherent, ordered fcc g ¢ (Ni3Al, Ti) phase in the fcc iron-rich matrix. This type of mixed microstructure is also observed in agehardenable nickel-base superalloys. On the other hand, if the room-temperature structure consists mostly of a-iron solid solution that is made possible by the ferrite-forming elements (e.g., Cr, Si, Mo, W, and Al), it is called ferritic steel. Examples of ferritic steels are Fe-Cr alloys containing more than 13% Cr and low-carbon transformer steel containing about 3% Si. 1.7.2 Purpose of Alloying Elements Alloy additions are made to fulfill the following functions: (1) increase the hardenability (or strength in large sections), (2) reduce distortion due to heat treatment, (3) provide improved toughness at a particular hardness level, (4) increase the abrasion resistance at a given hardness level, and (5) increase the elevated temperature

CHAPTER ONE

1.16 TABLE 1.1

The Effect of Alloying Elements on Some Specific Properties22

Property Hardenability Minimum distortion Toughness Wear resistance Hot hardness

Elements (in order of decreasing effectiveness) Mn, Mo, Cr, Si, Ni, V Mo (with Cr), Cr, Mn Ni (produces general toughness), V, W, Mo, Mn, Cr V, W, Mo, Cr, Mn W, Mo, Co, V, Cr, Mn

Source: After G. A. Roberts, J. C. Hamaker, and A. R. Johnson, Tool Steels, ASM, Materials Park, Ohio, 1971.

FIGURE 1.11 Solid-solution hardening effects of various alloying elements dissolved in a-iron.20 (Reprinted by permission of ASM International, Materials Park, Ohio.)

strength and hardness. Table 1.1 shows the effects of various alloying elements on some specific properties.22,23 This approach has some drawbacks in complex alloys because of the mutual interactions between two or more elements. 1.7.3 Distribution of Alloying Elements If only steels in which g transforms to ferrite and carbide on slow cooling are considered, the alloying elements can be divided into three groups: (1) elements entering only in the ferrite phase, (2) elements forming stable carbides and entering the ferrite phase, and (3) elements entering only the carbide phase. In the first group belongs Cu, Ni, P, and Si, elements that are usually found in solid solution in the ferrite phase. Their solid solubility in cementite or alloy carbides is negligible. The relative strengthening effects of some substitutional solutes in solid solution in a-iron are shown in Fig. 1.11.20 The carbide formers Cr, W, V, and Mo appear to be relatively ineffective.

IRON-CARBON ALLOYS

1.17

The majority of alloying elements belong to the second group. They are carbide formers as well as ferrite formers with respect to iron. At higher concentrations most will form thermodynamically more stable alloy carbides than cementite. At low concentrations they go into solid solution in cementite and also form solid solutions in ferrite. Typical examples are Mn, Cr, Mo, V, Ti, W, and Nb. Manganese does not form a separate carbide in steel; rather Mn can dissolve readily in Fe3C. The carbide-forming elements are usually present in greater amounts than required in the carbide phase, which are determined mainly by the carbon content of the steel. The remainder enter into solid solution in the ferrite with the non-carbide-forming elements Ni and Si. Some of these elements, particularly Ti, W, and Mo, produce considerable solid solution hardening of ferrite. In the third group are those elements which enter directly into the carbide phase. The affinity of elements for carbon increases from left to right: Mn, Cr, W, Mo, V, Ti, Nb, Ta, to Zr. All carbide formers are also nitride formers. Nitrogen is the most significant element, and it forms carbonitrides with iron and many alloying elements. However, in the presence of certain very strong nitride-forming elements, such as Ti and Al, a separate alloy nitride phase can form. The affinity of elements for nitrogen decreases from left to right: Al, Ti, Mo, Cr, V, and Ni.

1.7.4 Alloy Carbides At the carbon-rich side we find hard metastable cementite (Fe3C) or M3C with a complex orthorhombic crystal structure, where M stands for a metal atom or some combination of metal atoms. Now M3C is the predominant carbide type in lowcarbon low-alloy steels in the absence of strong carbide-forming elements. Thus, M3C occurs in annealed steels containing low W, V, and Mo contents. Any substitution of Fe in the carbide is done mainly by Mn or Cr.23 There are three other metastable iron carbides with iron-carbon ratios £3 : 1 which may be produced by low-temperature carburization of iron, iron oxide, or iron nitrides. These are (1) Hägg carbide, c-carbide or iron percarbide, Fe5C2 or M5C2 phase; (2) Fe7C3; and (3) e-carbide [formed during low-temperature (175 to 250°C) carburization of iron or tempering of certain high-carbon ferrous martensites].24 M5C2 Carbides. In steels M5C2 has been observed to precipitate as rodlike carbides during tempering of martensitic carbon steels, in the AISI 4340 steels used for gun barrels, and in the 1 Cr–0.5 Mo steels. It is a monoclinic phase and appears to nucleate preferentially on intraferrite M2C carbides and replace the needlelike M2C carbides after prolonged service. This appears to be a more stable phase thermodynamically than M2C under typical service conditions.25 In iron-based alloys or stainless steels, the predominant carbide is M23C6. In CrMo-V steels, precipitation of several types of carbides such as MC, M2C, M7C3, M23C6, and M6C has been reported. Several researchers have observed carbide transformation types MC Æ M2C, M3C Æ M7C3, M2C Æ M6C, and M23C6 Æ M6C. In low and more highly alloyed steels, transitional carbides may form during aging before the stable occurs (see Chap. 8 for more details).24,26 Generally, the sequence of carbide precipitation in all steels during martensite aging may be written as M3C Æ MC + M2C + M7C3 Æ M23C6 Æ M6C.27 It is further suggested that prolonged service exposure gives rise to precipitation of one or more carbides M2C, M7C3, M6C, and M23C6. In Si-containing steel, the sequence of formed carbides is (e-carbide) Æ M3C Æ M2C Æ M7C3 Æ M23C6.28 M23C6 Carbide. M23C6 represents single alloy carbide as well as double and complex carbides containing iron and carbide-forming elements. Examples are

1.18

CHAPTER ONE

Cr23C6, Mn23C6, (CrFe)23C6, (Fe21Mo2)C6, (Fe21W2)C6, and (FeMnVNbMoW)23C6. M23C6 carbides readily form in alloys with moderate to high Cr content. Their formation, along with that of g ¢ (gamma prime), occurs at lower aging temperature according to Eq. (10.15). M6C Carbides. M6C represents double carbides containing iron and carbideforming elements such as Mo and W. Examples are Fe4Mo2C and Fe4(MoW)2C, Fe4W2C, and Fe3(MoW)3C. This carbide can also dissolve moderate quantities of Cr, V, and Co. They form when Mo and/or W content is >6 to 8 at %, typically in the range of 815 to 980°C (1500 to 1800°F). Thus it is the main carbide in high-speed steels and is resistant to solution during austenitizing, leaving undissolved abrasion-resistant particles which are also growth-resistant during tempering.23 It is also observed in stainless steels containing >6% Mo or 0.8 to 2% Nb. In types 316 and 316L stainless steels, M6C appears to form from M23C6 after a prolonged aging time (>1500 hr) in a limited temperature range around 650°C according to Eq. (10.16). Also, M6C and M23C6 interact, forming one from the other: M 6C + M ¢ Æ M 23C6 + M ¢¢ or Mo3 ( NiCo)3 C + Cr ∫ Cr21Mo2C6 + NiMoCo (1.5) Because M6C carbides are stable at higher temperatures than are M23C6 carbides, M6C is more commercially important as a grain boundary precipitate for controlling grain size during the processing of wrought alloys. This also forms in neutronirradiated type 316 stainless steel. M7C3 Carbides. Chromium-rich M7C3, a hexagonal structure, probably forms in Fe-Cr-C or Fe-Cr-Ni-C alloys where carbon concentrations are considerably larger than those specified for the 300 series. M2C, like M6C, is W- or Mo-rich but has hexagonal crystal structure, for example, W2C and Mo2C. It dissolves in Cr but not Fe, and is mainly associated with secondary hardening. Tempering causes its transformation into either M6C or M23C6 and is not commonly present in annealed steels.23 MC Carbides. MC represents iron-free VC, V4C3, NbC, TiC, (Ti, Nb)C, TaC, or ZrC carbide, which is coherent with the g-iron or nickel-base matrix and has a fcc NaCl-type structure. All these elements are very strong carbide formers. Thus MC always occurs if V is present. For the formation of MC carbides, usually sufficient amounts of Nb and Ti are added to exceed the stoichiometric (i.e., atomic weight) ratios of 4 : 1 and 8 : 1, respectively.19 The high yield strength of microalloyed steels (i.e., high-strength low-alloy steels) is attributed to the ultrafine dispersion of these carbides in a-iron solid solution. MC carbides are a major source of carbon for subsequent phase reactions during heat treatment and service. MC formation is favored more by the presence of Mo than by W, and after heat treatment undissolved MC particles are very abrasion-resistant and play a significant role in improving wear resistance.23 In V-bearing steels, fine precipitates of MC (or VC) mostly contribute to the secondary hardening and their associated high tempering resistance. (See also Chap. 12 for more details.) The preferred order of formation (in order of decreasing stability) in superalloys for these carbides is HfC, TaC, NbC, and TiC. 1.7.5 Effect on the Eutectoid Composition and Temperature Austenite and ferrite stabilizers widen the respective phase fields. If alloying elements are added to the iron-carbon alloy (steel), the position of A1, A3, and Acm

IRON-CARBON ALLOYS

1.19

1300 Molybdenum

1200 Titanium

1100

2200 2000

Tungsten

1000

1800

900

Silicon

1600

800 Chromiun

1400

700 Manganese 1200

600 1000

Nickel

500 0

Eutectoid transformation temperature, F

Eutectoid transformation temperature, C

boundaries as well as the eutectoid composition are changed. Classical diagrams, introduced by Bain,5 illustrate the influence of increasing the amount of a selected number of alloying elements on eutectoid carbon content and A1 (eutectoid temperature) (Fig. 1.12a and b). Figure 1.12a shows the influence of alloying addition on eutectoid temperature, and Fig. 1.12b shows the related influence on eutectoid carbon content.9,20 The austenite stabilizers lower the eutectoid temperature, thereby widening the temperature range over which austenite is stable. Similarly, the ferrite formers raise the eutectoid temperature, thereby restricting the g-phase field. The effects of alloying elements, particularly Ti and Cr, on the g-phase field in the Fe-Ti-C and Fe-Cr-C systems, respectively, are shown in Fig. 1.13,29 from which it is evident that just over 1% Ti is required to eliminate the g-loop, whereas 20%

4 6 8 10 12 14 16 18 Alloying element, wt %

2

(a)

Eutectoid carbon content, wt %

0.80 0.70 0.60

Nickel

0.50 Chromiun

0.40 Silicon

0.30

Manganese Tungsten

0.20 Titanium

0.10

Molybdenum

0 0

2

4

8 10 12 14 16 18 6 Alloying element, wt % (b)

FIGURE 1.12 Effects of alloying elements on (a) the eutectoid reaction temperature and (b) the eutectoid carbon content.9,20 (Reprinted by permission of ASM International, Materials Park, Ohio.)

1.20

CHAPTER ONE

FIGURE 1.13 Effect of alloying additions on the g-phase field: (a) titanium, (b) chromium.20,29 (Reprinted by permission of ASM International, Materials Park, Ohio.)

Cr is necessary to achieve this result. Associated changes in eutectoid temperature and composition are also illustrated. The effect on the bainitic and martensitic transformation and tempering processes will be discussed in later chapters.

1.7.6 Steelmaking Practices and Characteristics Steels contain alloying elements and impurities that must be associated with austenite, ferrite, and cementite. The combined effect of alloying elements and heat treatment produces an enormous variety of microstructures and properties. In this section, the effects of various alloying elements, residual (or impurity) elements (such as P, S, As, Sb, Sn, N, H, and O), and deoxidants (such as Al, Si, and Ca) commonly found in steels are summarized. It should be noted that the effects of a single alloying element on either steelmaking practice or steel characteristics are modified by the presence of other elements. Such interactive effects are complex; these interreactions must be considered when evaluating a change in the composition of a steel. Carbon. The amount of C required in the finished steel limits the type of steel that can be made. As the C content of rimmed steels* increases, surface quality deteriorates. In contrast, killed steels† in the approximate range of 0.15 to 0.30% C may * Rimmed steels are cast into ingots without deoxidation by Al or Si. They have less shrinkage pipes than killed steel ingots, some shrinkage porosity, and lower C, S, and P near the surface region than the average composition of the ingot. † Killed steels are produced by adding deoxidizing elements, mostly Al and Si, to the ladle before pouring. They have a much larger shrinkage pipe in the center of the ingot, but usually have uniform chemical composition and mechanical properties throughout the ingot.

IRON-CARBON ALLOYS

1.21

have poorer surface quality and require special processing to attain surface quality comparable to steels with higher or lower carbon contents. Carbon has a moderate tendency to segregate, and C segregation is often more significant than any other alloying elements. Carbon is the main hardening element in all steels, except the austenitic PH stainless steels and maraging steels. Tensile strength (in the as-rolled conditions), hardness, and hardenability increase as the C content increases up to about 0.85%. However, toughness, ductility, and weldability decrease with the increasing C content.30 Manganese. Managanese is present in virtually all steels in amounts of 0.3% or more.31 Manganese is essentially a deoxidizer and desulfurizer.32 It has a lesser tendency for macrosegregation than any of the common elements. Steels above 0.60% Mn cannot be readily rimmed. Manganese is beneficial to surface quality in all carbon ranges (except extremely low-carbon, rimming steels) and is especially beneficial in resulfurized and freecutting steels due to a reduction in the risk of red or hot shortness* by forming dispersed MnS inclusions. Manganese addition contributes to the strength and hardness of steel, but to a lesser extent than carbon and, in addition, favorably affects forgeability and weldability. Manganese is a solid-solution strengthener in steel and is very effective in increasing the hardenability, but contributes to temper embrittlement (see also Chap. 14 for details). However, large quantities (>2%) result in increased tendency toward cracking and distortion during quenching.33 Phosphorus. Phosphorus segregates, but to a lesser extent than C and S. A small amount of P dissolves in the ferrite and slightly increases the strength and hardness of steel. A large quantity decreases the ductility and (notch) impact toughness in the as-rolled condition, imparting cold shortness (or brittleness under impact) to the steel, particularly in quenched and tempered higher-carbon steels. Higher P content is often specified in low-carbon (free-machining) steels to improve machinability. In low-alloy structural steels containing ~0.1% C, P increases strength and atmospheric corrosion resistance (rust-resistant steels). In austenitic Cr-Ni steels, P addition can cause precipitation effects and increase in yield point.33 Sulfur. Increased amount of S has a detrimental effect on transverse ductility, notch-impact toughness, weldability, and surface quality (particularly in low-carbon and low-manganese steels), but has only a slight effect on longitudinal tensile properties. It can cause reduction in hot working properties (i.e., increased red/hot shortness) due to the low-melting sulfide eutectics surrounding the grains in a network fashion.33,34 Higher S grades (>0.05%) are more susceptible to quench cracking than the low-S grades. The reasons are as follows: (1) S, mainly present in the form of sulfide inclusions, has a greater segregation tendency than any other common elements. Obviously, greater frequency of sulfide inclusions can be expected in the resulfurized grades. (2) The surface of the hot-rolled high-sulfur-containing steel has a greater tendency to form seams, which act as stress raisers during quenching. (3) They are usually coarse-grained for better machinability, which increases brittleness and, therefore, promotes quench cracking.35 Hence, only a low S content (7% produce austenitic structure to chemically resistant steels down to well below room temperature.33 Nickel alloy steels also have superior low-temperature strength and toughness.37 The good ductility, toughness, and flexible heat treatment of low-carbon nickel steels make them good case hardening materials.37 In combination with Cr, Ni produces alloy steels with increased hardenability, impact strength, and fatigue resistance than are possible with carbon steels. Molybdenum. Molybdenum is a pronounced carbide former. Molybdenum addition produces fine-grained steels, increases the hardenability, and improves the fatigue strength. It is added in constructional steels, usually in the range of 0.10 and 0.60%. Molybdenum can induce secondary hardening during the tempering of quenched steels and improves the creep strength of low-alloy steels at elevated temperatures. Alloy steels containing 0.15 to 0.30% Mo and/or V minimize the susceptibility of steel to temper embrittlement (TE) due to the slow precipitation of alloy carbides of increasing stability. It increases corrosion resistance and is thus used greatly with high-alloy Cr steels and with austenitic Cr-Ni steels. High Mo contents reduce the susceptibility to pitting.33 It retards pearlitic transformation from g far more than it does for bainitic transformation from g ; hence, bainite can be formed by continuous cooling of molybdenum-containing steels.

1.24

CHAPTER ONE

Tungsten. Tungsten is a very important carbide former. Tungsten in steel forms very hard, abrasion-resistant carbides. It promotes hot strength and red hardness, and thus the cutting ability. It improves toughness and prevents grain growth. This combination of properties makes it very useful in high-speed cutting tools.31 It has been suggested as a replacement for molybdenum in reducedactivation ferritic steels for nuclear applications. However, W impairs scaling resistance. Vanadium. Vanadium is an excellent deoxidizer, carbide former, and grain refiner, but it is very expensive and scarce.36 It dissolves to some extent in ferrite, imparting strength and toughness. Vanadium increases the deep drawing characteristics for hot-band low-carbon steel and prevents an excessive ferrite grain growth, if the coiling temperature is high.43a Vanadium increases the fatigue strength on one hand, but improves the notch sensitivity on the other hand; it has no appreciable effect on the corrosion resistance. Vanadium also forms nitrides and is present in most nitriding steels. Vanadium addition up to ~0.05% increases the hardenability of steel; larger additions tend to reduce the hardenability, probably due to the formation of vanadium carbides which have difficulty dissolving in austenite. However, a typical V content of 0.1% can increase the yield strength of a controlled-rolled Nb-containing microalloyed plate steel due to a combination of both grain refinement and precipitation strengthening. Vanadium additions up to 0.2% are made into microalloyed medium-carbon forging grades. Vanadium increases abrasion wear resistance, edge-holding quality, and hightemperature strength. It is used, therefore, mainly as an additional alloying element (of 1% and above) in high-speed, hot-forming, and creep-resistant steels. Vanadium additions of up to 0.75% are incorporated into low-alloy Cr-Mo-V steels to improve secondary hardening and good creep strength up to 565°C temperature.44 It promotes the weldability of heat-treatable steels. Vanadium steels exhibit a much finer structure than steels of a similar composition without V. It provides other important alloying effects such as increased hardenability, secondary hardening during tempering (through precipitation hardening), and increased elevated-temperature hardness. The presence of V retards the rate of temper embrittlement in molybdenum bearing steels by a factor of 10; the mechanism has not yet been established. Niobium and Tantalum. These are ferrite formers and, therefore, reduce the austenite phase. Small additions of Nb increase the yield strength and, to a lesser extent, the tensile strength of carbon steel. A 0.02% Nb addition can increase the yield strength of medium-carbon steel by 70 to 100 MPa (10 to 15 ksi). This increased strength may be accompanied by considerably reduced notch toughness unless special measures are employed to refine grain size during hot rolling. Grain refinement during hot rolling involves special thermomechanical treatment techniques such as controlled-rolling practices, low finishing temperature for final reduction passes, and accelerated cooling after the completion of rolling (see also Chap. 15). Aluminum. Aluminum, in small amounts (0.015 to 0.060 wt %), is mostly used as the other principal deoxidizer in steelmaking; however, it also performs as a grain refiner.10 It has the drawback of a tendency to promote graphitization, if present, in excess of 0.06% (e.g., ~0.35%), and dramatically reduces the creep strength and is therefore undesirable in steels to be used for high-temperature applications. As Al forms very hard nitrides with nitrogen, it is usually an alloying element in nitriding steels. It increases scaling resistance and is therefore often added to alloy ferritic heat-resistant steels. Aluminum combines with N in the solid state to minimize

IRON-CARBON ALLOYS

1.25

austenite and to minimize the effects of strain aging. This combination also helps control the plastic strain ratio of sheet products. Calcium. Calcium is sometimes used to deoxidize steels to improve machinability and control shape and distribution of nonmetallic (sulfide) inclusions, thereby improving the toughness.45 Titanium. The effects of Ti are similar to those of V and Nb, but Ti is only beneficial in fully killed (aluminum-deoxidized) steels due to its strong deoxidizing effects. Titanium also lowers the soluble carbon and nitrogen to very low levels and is used to produce the interstitial-free (IF) steels which have improved ductility and extremely high cold formability.46 Microalloying with Ti improves drawability in low-carbon wire rod steels.46a Titanium is used widely in stainless steels as a carbide former for stabilization against intergranular corrosion. Titanium increases creep rupture strength through the formation of special nitrides and tends significantly to segregation and banding.33 Titanium addition is made in boron-treated steels because it combines instantly with any oxygen and nitrogen in steel, thereby increasing the effectiveness of B in improving the steel hardenability.45 Titanium, Zr, and V are effective grain growth inhibitors; however, for structural steels that require heat treatment (quenching and tempering), these three elements may have adverse effects on hardenability, because their carbides are very stable and difficult to dissolve in austenite prior to quenching. Zirconium. Zirconium addition to killed high-strength low-alloy steels is made to obtain improvements in inclusion characteristics, particularly sulfide inclusions where modifications in inclusion shape improve ductility in transverse bending. Zirconium increases the life of heating conductor materials and produces contracted gamma-phase field.33 Its main use is to improve hot-rolled properties in HSLA steels. Zirconium in solution improves slightly the hardenability.47 Arsenic and Antimony. They are ferrite stabilizers. They can render steel susceptible to temper embrittlement. Tin. Tin in relatively small amounts is harmful to steels for deep drawing, but for most uses, the effects of tin in the amounts usually present are negligible.48 It tends toward increased segregation, is a ferrite stabilizer, and limits the gammaphase field.33 It can increase susceptibility of a steel to temper embrittlement and hot shortness.45 Hydrogen. Hydrogen dissolved in steel during manufacturing has an embrittling effect which can result in flaking during cooling from hot-rolling temperatures. However, dissolved hydrogen rarely affects the finished mill products because reheating of the steel prior to hot forming bakes out nearly the entire hydrogen. Nitrogen. Nitrogen increases the strength, hardness, and machinability of steel, but it lowers the ductility and toughness (i.e., raises the ITT) of ferrite-pearlite steels and can give rise to strain aging. In Al-killed steels, N forms AlN particles that control the grain size of the steel, thereby improving both strength and toughness. It can decrease the effect of B on the hardenability of steels. Nitrogen in solid solution is deleterious to the cold formability of low-carbon strip steel, causing low r values, the mean plastic strain ratio (see also Chap. 4). Nitrogen produces a considerable solid-solution hardening and precipitation strengthening reactions which form the foundation of many high-strength steels. Nitrogen additions are also advantageous to the constitution and pitting resistance of austenitic stainless steels.46 Oxygen. Oxygen, which is widely observed in rimmed steels, can slightly increase the strength of steel, but adversely affects the toughness.

1.26

CHAPTER ONE

1.8 STEEL CLASSIFICATIONS Steels can be classified by several different systems depending on (1) the compositions, such as carbon, low-alloy, alloy, or stainless steels; (2) the manufacturing methods, such as basic and acid open hearth, or electric furnace methods; (3) the finishing methods, such as hot rolling or cold rolling; (4) the product shape, such as bar, plate, strip, tubing, or structural shape; (5) the application, such as structural, spring, and high tensile steels; (6) the deoxidation practice, such as killed, semikilled, capped, and rimmed steels; (7) the microstructure, such as ferritic, pearlitic, and martensitic; (8) the required strength level, as specified in ASTM Standards; (9) heat treatment, such as annealing, quenching and tempering, and thermomechanical processing; and (10) quality descriptors/classifications, such as forging quality and commercial quality.30,45,49 Among the above classification systems, quality descriptors and chemical compositions are the widely used basis for designation and will be described in this chapter. Classification systems based on deoxidation practice will be discussed in Chap. 3.

1.8.1 Quality Descriptors/Classifications Quality descriptors are names applied to various steel products to indicate that the particular products possess certain characteristics that make them especially well suited for specific applications or fabrication processes. The quality designations/ descriptors for various carbon steel products and alloy steel plates are listed in Table 1.2. Forging quality and cold-extrusion quality descriptors for carbon steels are self-explanatory. However, others are not explicit; for example, merchant-quality hot-rolled carbon steel bars are made for noncritical applications requiring modest strength and mild bending or forming, but not requiring forging or heat treating operations. The quality classification for one particular steel commodity is not necessarily extended to subsequent products made from the same commodity—for example, standard-quality cold-finished bars are produced from special-quality hotrolled carbon steel bars. Alloy steel plate qualities are described by structural, drawing, cold working, pressure vessel, and aircraft qualities.49 The various physical and mechanical characteristics indicated by a quality descriptor result from the combined effects of several factors, such as (1) internal soundness; (2) uniformity of chemical composition; (3) number, size, and distribution of nonmetallic inclusions; (4) relative freedom from harmful surface imperfections; (5) extensive testing during manufacture; (6) size of the discard cropped from the ingot; and (7) hardenability requirements. Control of these factors during manufacture is essential to achieve mill products with the desired characteristics. The degree of control over these and other related factors is another segment of information conveyed by the quality descriptor. Some, but not all, of the basic quality descriptors may be modified by one or more additional requirements as may be appropriate, namely, macroetch test, special discard, restricted chemical composition, maximum incidental (residual) alloying elements, austenitic grain size, and special hardenability. These limitations could be applied to forging-quality alloy steel bars, but not to merchant-quality bars. Understanding the various quality descriptors is difficult because most of the prerequisites for qualifying a steel for a specific descriptor are subjective. Only

TABLE 1.2

Quality Descriptions* of Carbon and Alloy Steels51 Carbon steels

1.27

Semifinished for forging Forging quality Special hardenability Special internal soundness Nonmetallic inclusion requirement Special surface Carbon steel structural sections Structural quality Carbon steel plates Regular quality Structural quality Cold-drawing quality Cold-pressing quality Cold-flanging quality Forging quality Pressure vessel quality Hot-rolled carbon steel bars Merchant quality Special quality Special hardenability Special internal soundness Nonmetallic inclusion requirement Special surface Scrapless nut quality Axle shaft quality Cold extrusion quality Cold-heading and cold-forging quality Cold-finished carbon steel bars Standard quality Special hardenability Special internal soundness Nonmetallic inclusion requirement Special surface Cold-heading and cold-forging quality Cold extrusion quality

Hot-rolled sheets Commercial quality Drawing quality Drawing quality special killed Structural quality Cold-rolled sheets Commercial quality Drawing quality Drawing quality special killed Structural quality Porcelain enameling sheets Commercial quality Drawing quality Drawing quality special killed Long terne sheets Commercial quality Drawing quality Drawing quality special killed Structural quality Galvanized sheets Commercial quality Drawing quality Drawing quality special killed Lock-forming quality Electrolytic zinc-coated sheets Commercial quality Drawing quality Drawing quality special killed Structural quality Hot-rolled strip Commercial quality Drawing quality Drawing quality special killed Structural quality Cold-rolled strip Specific quality descriptions are not provided in cold-rolled strip because this product is largely produced for specific end use

Alloy steels Tin mill products Specific quality descriptions are not applicable to tin mill products Carbon steel wire Industrial quality wire Cold extrusion wires Heading, forging, and roll-threading wires Mechanical spring wires Upholstery spring construction wires Welding wire Carbon steel flat wire Stitching wire Stapling wire Carbon steel pipe Structural tubing Line pipe Oil country tubular goods Steel specialty tubular products Pressure tubing Mechanical tubing Aircraft tubing Hot-rolled carbon steel wire rods Industrial quality Rods for manufacture of wire intended for electric welded chain Rods for heading, forging, and rollthreading wire Rods for lock washer wire Rods for scrapless nut wire Rods for upholstery spring wire Rods for welding wire

* In the case of certain qualities, P and S are usually finished to lower limits than the specified maximum. Reprinted by permission of Society of Automotive Engineers, Warrendale, Pa.

Alloy steel plates Drawing quality Pressure vessel quality Structural quality Aircraft physical quality Hot-rolled alloy steel bars Regular quality Aircraft quality or steel subject to magnetic particle inspection Axle shaft quality Bearing quality Cold-heading quality Special cold-heading quality Rifle barrel quality, gun quality, shell or A.P. shot quality Alloy steel wire Aircraft quality Bearing quality Special surface quality Cold-finished alloy steel bars Regular quality Aircraft quality or steel subject to magnetic particle inspection Axle shaft quality Bearing shaft quality Cold-heading quality Special cold-heading quality Rifle barrel quality, gun quality, shell or A.P. shot quality Line pipe Oil country tubular goods Steel specialty tubular goods Pressure tubing Mechanical tubing Stainless and head-resisting pipe, pressure tubing, and mechanical tubing Aircraft tubing Pipe

1.28

CHAPTER ONE

limitations on chemical composition ranges, residual alloying elements, nonmetallic inclusion count, austenitic grain size, and special hardenability are quantifiable. The subjective evaluation of the other attributes depends on the experience and skill of the individuals who make the evaluation. Although the use of these subjective quality descriptors might appear impractical and imprecise, steel products made to meet the requirements of a specific quality descriptor can be relied upon to have those characteristics necessary for that product to be used in the suggested application or fabrication operation.30

1.8.2 Steel Classification Based on Chemical Composition 1.8.2.1 Carbon and Carbon-Manganese Steels. In addition to carbon, plaincarbon steels contain the following elements: Mn up to 1.65%, S up to 0.05%, P up to 0.04%, Si up to 0.60%, and Cu up to 0.60%. The effects of each of these elements in plain-carbon steels have been summarized in Sec. 1.7.5. Carbon steels can be classified according to various deoxidation practices (see Sec. 3.10.1.1). Deoxidation practice and steelmaking process have effects on the characteristics and properties of the steel. However, variations in C have the greatest effect on mechanical properties; increased C addition leads to increased hardness and strength. As such, carbon steels are generally grouped according to their C content. In general, carbon steels contain up to 2% total alloying elements and can be subdivided into low-carbon steels, medium-carbon steels, high-carbon steels, ultrahigh-carbon steels, and boron-treated steels; each of these designations is discussed below. As a group, carbon steels are most widely used. Table 1.3 lists various grades of standard carbon and low-alloy steels with SAE-AISI designations. Tables 1.4 through 1.7 show some representative standard plain (nonresulfurized, 1% Mn maximum) carbon steel, free-cutting (resulfurized) carbon steel, free-cutting (resulfurized and rephosphorized) carbon steel, and high-manganese (nonresulfurized) carbon steel compositions, respectively, with SAE-AISI and corresponding Unified Numbering System (UNS) designations.30,50,51 Low-Carbon Steels. They contain up to 0.25% C. The largest category of this class is flat-rolled products (sheet or strip), bar, rod, wire, nut, bolt, tube, and numerous machined parts that are subjected to low stresses. The carbon content for high-formability and high-drawability steels is very low ( t0, the total quantity diffused via the plate becomes J(t - t0). This method does not require the determination of concentration.13

2.2.2 Fick’s Second Law Mostly non-steady-state situations are set up during the solid-state transformations where concentration varies with both distance and time. For simplicity, let us consider a region of solid bar where a concentration profile exists only along one dimension (x), as shown in Fig. 2.3a. The flux due to the concentration gradient at individual values of x is illustrated in Fig. 2.3b. Let us further assume a volume element of material with thickness Dx and cross-sectional area A, as shown in Fig. 2.3c. The flux J1 into this element across plane (1) is greater than J2 out of plane (2). The rate of accumulation of diffusible substance in this volume element is the difference between the inward and outward flux. In other words, whatever comes in and does not go away, remains there. The net increase of concentration of diffusible atoms in a given volume element, A Dx, in a time increment ∂t is given by

2.12

CHAPTER TWO

FIGURE 2.3 The derivation of Fick’s second law, showing (a) an assumed C(x) plot, (b) J(x) for this plot, and (c) the element of volume with the flux J1 entering and J2 leaving.

Ê ∂c ˆ Ê ∂J ˆ J1 - J 2 = DxÁ ˜ = - DxÁ ˜ Ë ∂t ¯ Ë ∂x¯

(2.20)

It is clear from Eq. (2.20) that in the limit of ∂t Æ 0 ∂c ∂J == -—J ∂t ∂x

(2.21)

DIFFUSION IN METALS AND ALLOYS

2.13

Substituting Eq. (2.6) for the flux gives ∂ c ∂ D∂ c = ∂t ∂ x ∂ x

(2.22)

If D is assumed to be a constant (however, this assumption is not typically true), independent of concentration and, therefore, position x in the sample, or if the sample is chemically homogeneous as in self-diffusion, Eq. (2.22) reduces to ∂c ∂ 2c =D 2 ∂t ∂x

(2.23)

The differentials, Eqs. (2.21) through (2.23), are often called (one-dimensional) Fick’s second law of diffusion or simply the diffusion equation. Equation (2.23) has a simple graphical interpretation. If C(x) plot is concave upward (Fig. 2.3a), the concentration at all points on such a curve increases with time and ∂C/∂t becomes positive (that is, ∂2c/∂x2 > 0) at every point. In contrast, if C(x) plot is convex upward, the concentration decreases with time and ∂c/∂t is negative (that is, ∂2c/∂x2 < 0).10 Equation (2.8) can also be developed in the same way to give ∂ c ∂ D∂ c ∂ ( v c) = ∂t ∂ x ∂ x ∂ x

(2.24a)

Thus the general diffusion equation is a second-order partial differential equation. It cannot be solved analytically if D and ·vÒ are functions of concentration and therefore of x and t, which is the situation for systems that remain chemically homogeneous (e.g., self-diffusion).3 Then Eq. (2.24a) can be written ∂c ∂ 2c ∂c =D 2 - v ∂t ∂x ∂x

(2.24b)

If there is no drift term, Eq. (2.24b) reduces to Eq. (2.23). For three-dimensional diffusion, Eq. (2.23) can be expressed by ∂c ∂ 2c ∂ 2c ∂ 2c = Dx 2 + Dy 2 + Dz 2 ∂t ∂x ∂y ∂z

(2.25)

where the three diffusion coefficients allow for anisotropy of diffusion in noncubic metals.14 More usually they can be written in vectorial form ∂c = -— J = div (D grad c) = —( D —c ) = D — 2 c ∂t

(2.26)

where —2c is the Laplacian of c.10 The above differential equation of Fick’s second law can also be transformed into other coordinates. For example, in spherical coordinates, Fick’s second law becomes ∂c Ê ∂ 2 c 2 ∂c ˆ = DÁ 2 + ˜ Ë ∂r ∂t r ∂r ¯ In cylindrical coordinates, this diffusion equation is given by

(2.27)

2.14

CHAPTER TWO

∂c Ê ∂ 2 c 1 ∂c ˆ = DÁ 2 + ˜ Ë ∂r ∂t r ∂r ¯

(2.28)

For a hollow cylinder, this diffusion equation is given by 1 ∂c ∂ 2c 1 ∂c = + D ∂t ∂r 2 r ∂r

(2.29)

where C = C(r,t). The initial condition at t = 0 is C(r,0) = 0 at a < r < b. Fick’s Second Law for Concentration-Dependent Diffusion Coefficient. In general, the diffusion coefficient is concentration-dependent which implies that the diffusion coefficient changes with position in the sample. In this situation, according to Eq. (2.22), Fick’s second law must be written in terms of chemical diffusion ˜ (c):12 coefficient D ∂ c ∂ Ê ˜ ∂ c ˆ ∂ D˜ ∂ c ˜ ∂ 2 c = +D 2 ÁD ˜ = ∂t ∂ x Ë ∂ x ¯ ∂ x ∂ x ∂x

(2.30)

2.2.2.1 Non-steady-state Solutions. In a specific diffusion problem, Eq. (2.23) has to be solved for the given initial and boundary conditions. The equation can be integrated only if the diffusivity is independent of position. We will outline here the solution to the one-dimensional diffusion equation (2.23) for three types of situations which are of technical importance. The first concerns the situation where a complete homogenization is reached within a finite time and the second is for a semi-infinite system where the concentration remains unchanged within a finite period in the regions at large x.13 The third is an application of a thin-film solution or thin-layer method. The practical significance of the degree of homogeneity of alloys is well known throughout the metal industry. 1. Homogenization. It is often necessary to know the time required for an inhomogeneous alloy or casting to reach complete homogeneity. Here we will discuss the diffusive processes in solids of the practical problems of the elimination of periodic microsegregation arising from alloy solidification. Let us consider the situation in which the variation of solute concentration with distance in one dimension is sinusoidal (Fig. 2.4). Here solute atoms diffuse down the concentration gradients such that the solute concentration decreases in the regimes between x = 0 and x = l and increases in the regimes between x = l and x = 2l. The curvature is zero at x = 0, l, and 2l; hence, the concentration at these points remains unchanged with time, assuming D is independent of C, as given in Eq. (2.23). Initially (i.e., at time t = 0) the concentration at any point along the line C(x,0) can be described by C( x, 0) = C0 + Cm sin

px l

(2.31)

where C0 is the overall or average alloy content or concentration, Cm is the amplitude of the initial concentration profile, and l is the average distance between adjacent maxima and minima. The solution to Eq. (2.23), using interdiffusion coefficient D (discussed later), which shows many of the salient features of homogenization in binary systems and satisfies this initial condition, is given by

DIFFUSION IN METALS AND ALLOYS

2.15

FIGURE 2.4 The effect of diffusion on a sinusoidal distribution of solute toward homogenization.

px Ê - Dtp 2 ˆ C( x, t ) = C0 + Cm cosÊ ˆ exp Á ˜ Ë l ¯ Ë l2 ¯

(2.32)

That is, the amplitude Cm of a sinusoidal initial distribution will decay exponentially in time; the rate depends on the wavelength 2l. The time to reach 1/e of the initial amplitude, called relaxation time, t is equal to l2/(p2D). Thus, we can rewrite Eq. (2.32) as px -t C( x, t ) = C0 + Cm cosÊ ˆ exp Ê ˆ Ë l ¯ Ët ¯

(2.33)

Similarly, the concentration profile at x = l/2 and time t is given by

where

l -t px C Ê , tˆ = C + C0.5 m cosÊ ˆ expÊ ˆ Ë2 ¯ Ë l ¯ Ët ¯

(2.34)

-t C0.5 m = Cm expÊ ˆ Ët ¯

(2.35)

Equation (2.33) also shows that the deviation of concentration C at any point from C0 (that is, C - C0) decreases exponentially with time and after a long time approaches zero, i.e., becomes completely homogenized, or C = C0. The rate at which this occurs is a function of t, which, in turn, is a function of l and D. The solute distribution at this stage should appear as a dashed curve in Fig. 2.4. After a time t = 2t, C - C0 is reduced by a total of 1/e2.14,15 In the foregoing derivations it is assumed that the initial solute concentration varies sinusoidally along the x direction. However, usually the initial concentration distribution should be considered as a superposition of a series of different sinusoidal wavelength 2li, associated with a characteristic time t, proportional to the

2.16

CHAPTER TWO

wavelength squared. Thus, the shortest wavelength terms will decay quickly, and the homogenization time will ultimately depend on the largest wavelength present in the initial solute distribution (or spectrum). It can be seen from Eq. (2.33) that the rate of homogenization increases quickly with the short-range variations in concentration: the longer-range variations would produce a 16-fold decrease in the time needed for equivalent homogenization (t µ l2).15 This process can also be augmented by raising the temperature and thus increasing D. Alternative methods of decreasing the magnitude of l include rapid cooling to provide fine-scale segregation and hot working of castings. The following practical conclusions can be derived from the homogenization of alloys:11 • In commercial alloys, with large dendritic arm spacing (DAS) (200 to 400 m), substitutional elements do not homogenize until excessively high temperatures and long diffusion times are used. • It is possible to significantly homogenize substitutional elements at reasonable temperatures and times only if the material has fine DAS ( t1 > 0) when the two semi-infinite bars of different compositions are annealed after welding.

fully g-iron with the formation of a-Fe on the surface in the temperature range of 723 to 910°C is shown in Fig. 2.6.17,18 4. Infinite (or diffusion) couple method. Another variation of Eq. (2.45) occurs when two homogeneous alloys of unequal solute concentrations C1 and C2 ( DG 2v) denotes a positive value of DG b2v. Therefore, a more useful equation for C *2v is given by55 C2*v =

( )

2 DG2bv z C1*v exp kT 2

(2.75)

A positive value of DG b2v means that C*2v increases more rapidly with temperature than when DG b2v becomes equal to zero. A combined field-ion microscopy and resistivity experiments on vacancies in quenched platinum performed by Seidman56 suggests the direct evidence for the presence of divacancies and a positive value of DGb2v. However, these vacancies might correspond to divacancies at the surface or just below the surface and, therefore, do not prove their existence within the bulk of a crystal.57 In a similar way, we can express the equilibrium concentration of trivacancies C*3v as

( )

3

C3*v = h C1*v exp

DG3bv kT

(2.76)

where h is the number of physically distinct orientations of the most stable configb uration of a trivacancy and DG3v is the Gibbs free energy of binding of a trivacancy. If the vacancy remains the main point defect in thermodynamic equilibrium, the total equilibrium vacancy concentration C*tv is given by C*tv = C*1v + 2C*2v + 3C*3v + . . . + nC*nv + . . .

(2.77)

CHAPTER TWO

2.36

where n is the number of vacancies contained in the cluster.55 For equilibrium concentration measurements, higher aggregates than divacancies need not be considered because of their negligibly small concentration in the normal cubic metals.57

2.5.2 Migration Rates of Defects and Atoms In moving from one site to another adjacent site, a defect has to overcome an activation barrier at the saddle point (midpoint position). The work done in this reversible, isothermal process, at constant pressure, is the change in Gibbs free energy of migration (DG md ) from a region when an atom moves from a normal site to the saddle-point position (the height of the barrier to be overcome), which again can be divided into an enthalpy of migration DH md and a vibrational entropy DS md term according to DG md = DH md - T DS md

(2.78) f v

It is assumed that DG md has all the properties held by DG of Eq. (2.72). The equilibrium concentration of activated complexes in the region of the saddle point C*d can be calculated using the same treatment as for C*v in Eq. (2.73). Instead of mixing f into the lattice vacancies that increase the DG v per mole of vacancies, we mix in activated complexes which increase the free energy by DG md per mole of complexes. Since the ideal entropy of mixing remains the same for vacancies and activated complexes, the equilibrium concentration of activated complexes in the neighborhood of the saddle point at any instant can be given by Cd* =

nd*

Ê - DGdm ˆ Ê - DH dm + TDSdm ˆ = expÁ ˜ ˜ = expÁ Ë kT ¯ Ë ¯ kT na + nd*

(2.79)

where C *d is the equilibrium concentration of activated complexes, n*d is the equilibrium number of activated complexes, and na is the number of atoms. A statistical analysis, taking all degrees of freedom into consideration, leads to a simple final result for the probability per second wd of a defect jumping into a particular neighboring site (i.e., the average jump frequency of defect per atom): Ê - DGdm ˆ wd = vCd* = n expÁ ˜ Ë kT ¯

(2.80)

where ␯ is the oscillation (or vibrational) frequency at the bottom of the potential well of the atom that makes the jump, which is usually taken close to the Debye frequency ␯D (~3 ¥ 1012 s-1).28 2.5.3 Experimental Measurement of Enthalpy (DH vf ) and Entropy (DSvf ) in Vacancy Formation During quenching, three perturbing effects arise:44 (1) clustering of vacancies, (2) loss of vacancies to sinks such as dislocations and grain boundaries, and (3) creation of additional defects by plastic deformation. Ideally, one would like to measure directly the individual concentration of various clusters Cnv which are present after quenching. But only the total quenched-in single or monovacancy C1v has been approximately estimated to date.

DIFFUSION IN METALS AND ALLOYS

2.37

Properties of vacancy-type defects are derived from results of measurements of the self-diffusion coefficient, the isotope effect, the lattice parameter, length expansion with temperature, the quenched-in resistivity, and positron annihilation.42 Only the last three are briefly discussed here. 2.5.3.1 Direct Determination. Individual concentrations of monovacancies C1v, divacancies C2v, and higher-order vacancies Cnv present at a quenched metal tip can be determined directly by field ion microscopy (FIM). However, the main problems involved in the measurement of Cnv are (1) existence of artifact defects, (2) the sampling problems linked with the observation of a statistically large defect concentration, and (3) stress-induced defect migration, because of the imposition of large electric field on the specimen at the imaging temperature (T); this causes a reduced concentration of defects.58 Differential Dilatometry. The second direct method for the determination of the total quenched-in equilibrium concentration of vacancies C*v is differential dilatometry. This involves the simultaneous measurement of macroscopic length expansion DL/L and microscopic lattice parameter expansion Da/a as a function of temperature. This gives an additional amount of vacancies if measured at higher temperature in a nonequilibrium condition because the mobility of vacancies increases with increasing temperature. Due to the lack of mobility of vacancies below a certain temperature, the specimen will always contain a nonequilibrium concentration Cv,0 of vacancies at ambient temperature. Hence, the total vacancy concentration at temperature T in isotropic media is59 DL Da ˆ Cv (T ) = Cv ,0 + 3(1 + Cv ,0 )Ê Ë L a ¯

(2.81)

If we assume Cv,0 = 0, then Eq. (2.81) converts the well-known Simmons and Balluffi equation to DL Da ˆ Cv (T ) = 3Ê Ë L a ¯

(2.82)

This equation holds good for close-packed metals or cubic crystals with small concentration of vacancies. A plot of 3(DL/L - Da/a) against 1/T gives approximate f f f f values for DHv and DS v (Fig. 2.19). Table 2.3 lists DHv and DS v values for several f f metals. These values often imply the assumption that DH v and DS v are independent of T. However, this may not always be a sufficient approximation. For example, there appear to be large temperature variations in metals which exist in more than one crystal structure.60 But in all cases, it is presumed that such variations arise from the normal lattice expansion with increasing T and the accompanying changes in interatomic forces. In case both vacancies and interstitials form, the right-hand portion of Eq. (2.82) would be equal to the difference between vacancy and interstitial concentrations Cv and CI, respectively.10 That is, DL Da ˆ Cv - CI = 3Ê Ë L a ¯

(2.83)

This equation assumes that vacancies and interstitials occupy the same related volume; this is not true. It is also noted that interstitials are less likely to be quenchedin.

CHAPTER TWO

2.38 19

DL L

18 DL curve L

DL or Da (¥103) L a

17

cooling run heating run cooling run

Da a

16 15 14 13 cooling run heating run

Da curve a

12 11 10 400

450

500 550 Temperature, °C

600

650

FIGURE 2.19 Differential dilatometry in vacancy equilibrium conditions for aluminum. Relative length change DL/L and relative lattice parameter change Da/a of a sample against 1/T. The difference between the two lines is linearly proportional to the concentration of vacant atomic sites.10 (From R. Simmons and R. Balluffi, Phys. Rev., vol. 117, 1960, p. 52.)

TABLE 2.3 Values of DHVf as Determined by Positron-Annihilation Spectroscopy (PAS) and Quenching Methods for a Number of FCC and BCC Metals62 Hlvf (eV) Metal Al Ag Au Cu Ni V Nb Mo Ta W

PAS

Quenching

0.66 ± 0.02 1.16 ± 0.02 0.97 ± 0.01 1.31 ± 0.05 1.7 ± 0.1 2.1 ± 0.2 2.6 ± 0.3 3.0 ± 0.2 2.8 ± 0.6 4.0 ± 0.3

0.66 1.10 0.94 1.30 1.6 3.2 3.7

DIFFUSION IN METALS AND ALLOYS

2.39

2.5.3.2 Indirect Determination The Quenched-in Electrical Resistivity rq Measurement. A second and more accurate determination of total vacancy concentration can be obtained by measuring the change in quenched-in resistivity Drq as a function of temperature and hydrostatic pressure. (The order of magnitude of pressure dependence is such that for gold, a 6-kbar pressure increment corresponds to a decrease of temperature of about 30 K around 900 K for constant C *v .57) The method consists of heating the metal (such as Au, Cu) samples to a high temperature Tq, followed by rapid quenching to a very low temperature T1 (4.2 or 78 K) at which vacancies become immobilized and incapable of diffusing to sinks. This equilibrium concentration of vacancies corresponding to Tq is maintained at T1. Subsequently, resistivity measurements at low temperatures are performed on the quenched as well as the unquenched (or wellannealed) samples for direct comparison. It is found that the quenched concentration Cv remains smaller than C*v (Tq) because of (1) the loss of migrating vacancies during quenching to sinks such as dislocations, grain boundaries, and surfaces and (2) vacancy clustering. The difference in these resistivities Drq(= Drv) is then taken to be directly proportional to the total concentration of quenched-in vacancies. That is, Drq = Drv = aCv

(2.84)

where a is the resistivity per unit concentration of vacancies. Some deviations up to 10% may occur for vacancy defects at 78 K; however, these deviations become negligible at 4.2 K. Figure 2.20 shows an Arrhenius plot of the quenched-in residual resistivity of single and polycrystals of gold with different dislocation densities Nd.57 Positron-Annihilation Spectroscopy. Another method utilizes positronf annihilation spectroscopy (PAS) to determine DH v and thereby the vacancy concentration in thermal equilibrium in the medium temperature range (T 艐 0.6Tm). High-energy positrons produced by a nuclear reaction injected into a metal specimen are rapidly thermalized within a picosecond by electron-hole excitations and interactions with phonons. The thermalized positrons diffuse through the lattice and end its life either by annihilation with electrons or by trapping with vacancies. The vacancy-trapped positrons cause a marked reduction in the local electron density and produce increased lifetimes by 20 to 80% with respect to the free positrons in the perfect lattice. It is noted that the average lifetime of a positron in the metal crystal is about 200 ps.61 The increased lifetimes of captured positrons are proportional to the vacancy concentration Cv(T). The temperature dependence of the lifetimes is also used for accurate DHvf determination. Figure 2.21 shows the Arrhenius plot of vacancy concentration as measured by the positron annihilation spectroscopy for Au and Cu. The range of measurement by differential dilatometry is indicated for comparison. The slope of the line is a measure of DHvf .61 The PAS data extend to about two orders of magnitude lower vacancy concentration than the differential dilatometry data. Since monovacancies are predominant in this concentration range, PAS studies are very valuable as a complement to differential dilatometry on one side, and to resistivity measurements in quenched specimens on the other side. Table 2.3 lists values of DH vf obtained by PAS and resistivity methods for a number of fcc and bcc metals; usually a good agreement is found between these two methods.62

CHAPTER TWO

2.40 T, °C 100

1000 900

700

800

600

SINGLE CRYSTALS Nd < 2 ¥ 103 cm–2

∆rq , nΩ·cm

50

20

10 POLYCRYSTALS Nd~107 cm–2 5

8

9

10 4

11

12

–1

10 /T, K

FIGURE 2.20 Arrhenius plot of the quenched-in residual resistivity of single crystals and polycrystals of gold, showing the effect of the high dislocation density Nd in causing the curved plot for quenched polycrystals.57 (Source:After B. Lengeler, Phil. Mag., vol. 34, 1976, p. 259.)

2.6 EFFECT OF KEY VARIABLES ON DIFFUSIVITY In this section we will study diffusion as a function of key variables such as temperature T, pressure P, and isotopic mass m.

2.6.1 Effect of Temperature Normal self-diffusion (at elevated temperature) proceeds mainly through vacancies. Empirically it has been observed that the temperature dependence of the diffusivity of many systems can be expressed by the well-known Arrhenius-type equation -Q ˆ D = D0 expÊ Ë RT ¯

(2.85)

where D is the volume diffusivity; D0 is the preexponential or frequency factor for a given diffusion couple which has a value within a range of 10-6 m2◊s-1 to 10-3 m2◊s-1; Q is the total activation energy, within a range of 50 kJ◊mol-1 to 600 kJ◊mol-1, for

DIFFUSION IN METALS AND ALLOYS

2.41

T, °C 600

800

1000

400

500

log cvo

–3

–4

–5

7

8

9

10

11 4

10 /T, K

12

13

14

15

–1

FIGURE 2.21 Arrhenius plot of the vacancy concentration as derived from positron annihilation spectroscopy for copper (open symbols) and gold (full symbols) according to Triftshauser and McGervey. The left-hand side of the arrow corresponds to the range covered by the differential dilatometry. (Source: W. Triftshauser and J. D. McGervey, Appl. Phys., vol. 6, 1975, p. 177.)

self-diffusion; and R is the molar gas law constant. Both D0 (= ␯0a2f) [here ␯0 is an attempt frequency of the order of magnitude of the Debye frequency, a is the cubic lattice parameter, and f is a geometric factor depending on the lattice structure and type of interstitial sites involved12], and Q (= DHf + DHm) will vary with composition, but are independent of temperature (provided the diffusion mechanism does not change) and are together called Arrhenius parameters. The above equation may also be written as log D = log D0 -

Q 2.302RT

(2.86)

When log D is plotted against 1/T, a straight line is obtained with a slope equal to Q/2.302R. The value of Q is observed to be a function of the melting temperature Tm by the expression Q = 34Tm in calories per mole, or Q = 141Tm J/mol and Tm (in kelvins). This behavior forms the basis of the Van Liempt relation and is well obeyed by compact metals (see Fig. 2.22a for fcc metals and Fig. 2.22b for bcc metals). It is noted here that bcc structures exhibit a wider dispersion around the Van Liempt straight line than that of the fcc structures. Table 2.4 lists diffusion parameters D0 and Q for self-diffusion in various pure elements which can be employed to produce the Arrhenius diffusivity plots. Figure 2.23 shows the Arrhenius plot for certain metals. Figure 2.24 shows similar plots for the diffusion of impurity/tracer elements into silicon which are of importance in the fabrication of integrated circuits for the electronic industry. Figure 2.25 shows the Arrhenius plot of the different regimes of behavior which are possible for diffusion within the crystal lattice in a ceramic. For most of the metals the Arrhenius plot is slightly curved, even if contributions from short-circuit diffusion can be excluded. This curvature can be explained with temperature-dependent enthalpies DH and entropy DS (one-defect model) or with the competition of two or more diffusion mechanisms with different diffusion energies (two-defect model). In this case, temperature-dependent self-diffusivity DT near the melting point is given by

CHAPTER TWO

2.42

Self-diffusion fcc 600 Re 500

Ir Ph

Q, KJ

400 Co Th Fe Ni Pt Cu Pd Ce Au Al Ag Pb Pu La

300 200 100 0

1000

0

4000

3000

2000 Tm (a)

Self-diffusion bcc 600 W Mo

500 Cr Q, KJ

400 V

Ta Nb

Fea Fed

300

Tib Zrb Eu Hf b Cd b Lag Ye b Ti Tib b Pb Ug 100 Li Ced Zrb Rb Pue K Na 0 0 1000 2000 3000 200

4000

Tm (b)

FIGURE 2.22 Van Liempt relation for compact metals: (a) fcc metals and (b) bcc metals. The straight lines represent the Van Liempt relation: Q = 0.14TM kJ/mol and illustrates a wide dispersion of bcc structures around the line compared to the fcc structures.30

-Q1v ˆ -Q2 v ˆ D(T ) = DT = D01 v expÊ + D02 v expÊ Ë kT ¯ Ë kT ¯

(2.87)

where subscripts 1v and 2v denote the mono- and divacancy contributions, respectively.28,63

2.6.2 Effect of Pressure The rate of diffusion process is generally given by the expression

DIFFUSION IN METALS AND ALLOYS

FIGURE 2.23

2.43

Arrhenius plots for certain metals.

- DG1v ˆ È -(DG1fv + DG1mv ) ˘ D1Tv = D = fna 2 expÊ = fna 2 exp Í ˙ Ë kT ¯ kT ˚ Î È -(DH 1fv + DH 1mv ) ˘ Ê DS f + DS1mv ˆ expÁ 1v = fna 2 exp Í ˜ ˙ Ë ¯ kT k ˚ Î

(2.88)

where f is the geometric correlation factor (0.728 for Na),64 a is the jumping distance, ␯ is the average attempt frequency, and DG1v is the Gibbs free energy change f m comprising DG 1v and DG1v , corresponding to the formation and migration of defects, respectively. We can obtain a relation between D and the external hydrostatic pressure P by differentiating Eq. (2.88) with respect to pressure P at a constant temperature T as

TABLE 2.4 Self-diffusion Parameters for Pure Elements30

2.44

Element (see comments at end)a

C.S.b

Tm (K)c

D0(m2s-1) ¥ 104d

Q (kJ/mol)e

Ag

fcc

1234

Al Au Be

fcc fcc hex

933 1336 1560

Ca Cd

bcc hex

1116 594

Ceg T < 999 Ced T > 999 Co Cr Cu

fcc bcc fcc bcc fcc

g /d 999 1071 1768 2130 1357

Er

hex

1795

Eu

bcc

1099

D01 = 0.046 D02 = 3.3 2.25 0.084 ⬜c 0.52 //c 0.62 8.3 ⬜c 0.18 //c 0.12 0.55 0.007 2.54 1280 D01 = 0.13 D02 = 4.6 ⬜c 4.51 //c 3.71 1

Q1 = 169.8 Q2 = 218.1 144.4 174.1 157.4 165 161.2 82 77.9 153.2 84.7 304 441.9 Q1 = 198.5 Q2 = 238.6 302.6 301.6 144

Fea T < 1183 Feg 1183 < T < 1663

bcc fcc

a/g 1183 g /d 1663

121 0.49

Fed T > 1663 Gdb

bcc bcc

1809 1585

Hfa T < 2013

hex

a/b 2013

Temp. range (K)f

Q = 0.1422Tm Van Liemptg

D(Tm)(m2s-1)h

594–994

175.5

4.9 ¥ 10-13

Rein and Mehrer (1982)

673–883 1031–1333 836–1342 841–1321 773–1073 420–587 NA 801–965 1018–1064 944–1743 1073–1446 1010–1352

132.7 190 221.8 NAk 158.7 84.5 NA 152.3* 152.3 251.4 302.9 193

1.85 ¥ 10-12 1.3 ¥ 10-12 2.79 ¥ 10-10 1.85 ¥ 10-10 2.36 ¥ 10-11 1.11 ¥ 10-12 1.69 ¥ 10-12

Beyeler and Adda (1968) Herzig et al. (1978) Dupouy et al. (1966)

1475–1685 NA 771–1074

255.2

281.6 284.1

1067–1168 1444–1634

257.2* 257.2*

2.01 0.01

240.7 136.9

1701–1765 1549–1581

257.2 225.4

⬜c 0.28 //c 0.86

348.3 370.1

1538–1883 1470–1883

355.5*

156.2

4.9 ¥ 10-11 2.65 ¥ 10-13 1.86 ¥ 10-12 5.97 ¥ 10-13

D(ph. tr.)(m2s-1)i

Pavlinov et al. (1968) Mao (1972) 5.37 ¥ 10-13 (999 K) 2.36 ¥ 10-11 (999 K)

7.05 ¥ 10-13 6.2 ¥ 10-13 1.43 ¥ 10-11

2.25 ¥ 10-11 3.07 ¥ 10-11

Reference in Ref. 30 j

4.45 ¥ 10-15 (1183 K) 1.4 ¥ 10-17 (1183 K) 5.83 ¥ 10-14 (1663 K) 5.5 ¥ 10-12 (1663 K)

2.56 ¥ 10-14 2.14 ¥ 10-14 (2013 K)

Dariel et al. (1971) Languille et al. (1973) Lee et al. (1988) Mundy et al. (1981) Bartdorff et al. (1978) Spedding and Shiba (1972) Fromont and Marbach (1977) Geise and Herzig (1987) Heuman and Imm (1968) James and Leak (1966) Fromont and Marbach (1977) Davis and McMullen (1972)

2.45

Hfb T > 2013 In

bcc tetr

2500 430

Ir K

fcc bcc

2716 336

Lab T < 1134 Lag T > 1134

fcc bcc

b/g 1134 1193

Li

bcc

454

Mg

hex

922

Mo Na

bcc bcc

2893 371

Nb Ni

bcc fcc

2740 1726

Pb Pd Prb T > 1068 Pt

fcc fcc bcc fcc

601 1825 1205 2042

Pub 395 < T < 480 Pug 480 < T < 588

m ort

Pud 588 < T < 730 Pue T > 753 Rb

5.19 ¥ 10-11 1.08 ¥ 10-13 7.85 ¥ 10-14 1.3 ¥ 10-13 1.32 ¥ 10-11

8.13 ¥ 10-12 (2013 K)

Herzig et al. (1982)

0.0011 ⬜c 3.7 //c 2.7 0.36 D01 = 0.05 D02 = 1 1.5 0.11

159.2 78.5 78.5 438.8 Q1 = 37.2 Q2 = 47 188.8 125.2

2012–2351 312–417 NA 2092–2664 221–335

355.5 61.1 NA 386.2 47.8

923–1123 1151–1183

169.6* 169.6

Q1 = 53 Q2 = 76.2 138.2 139 488.2 Q1 = 35.7 Q2 = 48.1 395.6 Q1 = 278 Q2 = 357 106.8 266.3 123.1 Q1 = 259.7 Q2 = 365 108 118.4

220–454

64.5

3.13 ¥ 10-11

775–906

131.1

1360–2773 194–370

411.4 52.7

2.59 ¥ 10-12 2.37 ¥ 10-12 1.22 ¥ 10-12 1.75 ¥ 10-11

Combronde and Brébec (1971) Maier et al. (1979) Mundy (1971)

1354–2690 815–1193

389.6 245.4

1.5 ¥ 10-12 9.35 ¥ 10-13

Einziger et al. (1978) Maier et al. (1976)

470–573 1323–1773 1075–1150 850–1265

85.4 259.5 171.3 290.3

4.63 ¥ 10-14 4.9 ¥ 10-13 4 ¥ 10-11 1.4 ¥ 10-12

Miller (1969) Peterson (1964) Dariel et al. (1969) Rein et al. (1978)

b/g 480 g/d 588

D01 = 0.19 D02 = 95 ⬜c 1.75 //c 1.78 8 D01 = 57 D02 = 0.72 0.524 D01 = 0.92 D02 = 370 0.887 0.205 0.087 D01 = 0.06 D02 = 0.6 0.0169 0.038

Dariel et al. (1969) Languille and Calais (1974) Heitjans et al. (1985)

409–454 484–546

129.8* 129.8*

fcc

d/d¢ 730

0.0517

126.4

594–715

129.8*

bcc bcc

913 312

0.003 0.23

65.7 39.3

788–849 280–312

129.8 44.4

-11

3.62 ¥ 10

5.22 ¥ 10-11 6.05 ¥ 10-12

Dickey (1959) Arkhipova (1986) Mundy et al. (1971) 3 ¥ 10-13 (1134 K) 1.88 ¥ 10-11 (1134 K)

2.98 ¥ 10-18 (480 K) 4.95 ¥ 10-19 (480 K) 1.15 ¥ 10-16 (588 K) 3.05 ¥ 10-17 (588 K) 4.66 ¥ 10-15 (730 K) 8.3 ¥ 10-12 (753 K)

Wade et al. (1978) Wade et al. (1978) Wade et al. (1978) Cornet (1971) Holcomb and Norberg (1955)

TABLE 2.4 Self-diffusion Parameters for Pure Elements30 (Continued)

2.46

Element (see comments at end)a

C.S.b

Tm (K)c

Re Rh Sb

hex fcc trig

3453 2239 904

Se

hex

494

Sn

tetr

505

Ta Te

bcc trig

3288 723

Tha T < 1636 Tia T < 1155 Tib T > 1155

fcc hex bcc

a/b 1636 a/b 1155 1940

Tla T < 507

hex

a/b 507

Tlb T > 507

bcc

577

Ua T < 941

ort

a/b 941

0.002

167.5

853–923

199.8*

Ub 941 < T < 1048

tetr

b/g 1048

0.0135

175.8

973–1028

199.8*

Ug T > 108

bcc

1405

0.0018

115.1

1073–1323

199.8

9.46 ¥ 10-12

V

bcc

2175 3673

1323–1823 1823–2147 1705–3409

3.05 ¥ 10-12

bcc

331.9 372.4 Q1 = 525.8 Q2 = 665.7

309.3

W

1.79 26.81 D01 = 0.04 D02 = 46

522.3

1.7 ¥ 10-12

D0(m2s-1) ¥ 104d

Q (kJ/mol)e

511.4 391 ⬜c 0.1 149.9 //c 56 201 ⬜c 100 135.1 //c 0.2 115.8 ⬜c 21 108.4 //c 12.8 108.9 0.21 423.6 ⬜c 20 166 //c 0.6 147.6 395 299.8 169.1 6.6 ¥ 10-5 D(m2s-1) = 3.5 ¥ 10-4 ¥ exp(-328/RT) ¥ exp{4.1(Tm/T)2} ⬜c 0.4 94.6 //c 0.4 95.9 0.42 80.2

Temp. range (K)f

Q = 0.1422Tm Van Liemptg

1520–1560 903–2043 773–903

491 318.4 128.5

425–488

70.2

455–500

71.8

1261–2993 496–640

467.5 102.8

998–1140 1013–1149 1176–1893

287.7* 275.8* 275.8

420–500

82*

513–573

82

D(Tm)(m2s-1)h

D(ph. tr.)(m2s-1)i

Reference in Ref. 30 j Noimann et al. (1964) Shalayev et al. (1970) Cordes and Kim (1966)

2.17 ¥ 10-14 1.36 ¥ 10-14 5.18 ¥ 10-17 1.1 ¥ 10-17 1.29 ¥ 10-14 6.9 ¥ 10-15 3.9 ¥ 10-12 2.03 ¥ 10-15 1.3 ¥ 10-15

Brätter and Gobrecht (1970) Huang and Huntington (1974) Werner et al. (1983) Werner et al. (1983)

3.11 ¥ 10-11

1.48 ¥ 10-16 (1155 K) 5.4 ¥ 10-14 (1155 K)

2.3 ¥ 10-12

7.16 ¥ 10-15 5.2 ¥ 10-15 (507 K) 2.29 ¥ 10-13 (507 K) 1 ¥ 10-16 (941 K) 2.35 ¥ 10-16 (941 K) 2.33 ¥ 10-15 (1048 K) 3.29 ¥ 10-13 (1048 K)

Schmitz and Fock (1967) Dyment (1980) Köhler and Herzig (1987) Shirn (1955) Chiron and Faivre (1985) Adda and Kirianenko (1962) Adda et al. (1959) Adda and Kirianenko (1959) Ablitzer et al. (1983) Mundy et al. (1978)

Ya T < 1752

hex

a/b 1752

Yba T < 993 Ybb T > 993 Zn

hex bcc hex

a/b 993 1097 693

Zra T < 1136

hex

a/b 1136

Zr b T > 1136

bcc

2125

⬜c 5.2 280.9 //c 0.82 252.5 0.034 146.8 0.12 121 ⬜c 0.18 96.3 //c 0.13 91.7 No value Curved Arrh. plot D(m2s-1) = 3 ¥ 10-5 ¥ exp(-3.01/RT) ¥ exp{3.39(Tm/T)2}

1173–1573

256.4

813–990 995–1086 513–691

156* 156 98.5

779–1128

302*

1189–2000

302

2.08 ¥ 10-12 9.92 ¥ 10-13 1.59 ¥ 10-12

2.19 ¥ 10-12 2.43 ¥ 10-12 (1752 K) 6.4 ¥ 10-14 (993 K) 5.18 ¥ 10-12 (993 K)

艐5 ¥ 10-18 (1136 K) 6.14 ¥ 10-14 (1136 K) 1.37 ¥ 10-11

2.47

Source: These self-diffusion data have been extracted from the compilation by Mehrer et al. in Diffusion in Solid Metals and Alloys, LandoltBornstein New Series, vol. 26, ed. O. Madelung, Springer-Verlag, 1990. Reprinted by permission of Elsevier Science, Amsterdam, Netherlands. a Symbol of the metal. b Crystal stucture. bcc = body-centered cubic, fcc = face-centered cubic, hex = hexagonal, m = monoclinic, ort = orthorhombic, tetr = tetragonal, trig = trigonal. c Melting temperature. For the phases which do not melt (for instance Ce g, Fe, a etc.) we have given the temperature of the phase transition. d Experimental D0. The value in m2s-1 is multiplied by 104 (so that it is in cm-s-1). For some of the metals the Arrhenius plot is curved and D has the form: D = D01 exp(-Q1/RT) + D02 exp(-Q2/RT), in these cases D01 and D02 are given (they are also multiplied by 104). For Ti and Zr which have strongly curved Arrhenius plots, special expressions are given for D (in m2s-1 without any multiplying factor). e Experimental Q in kJ mol-1. Same remarks as for column 4. f Temperature range of the experimental determination of D. g Empirical value of Q according to the Van Liempt relation. For the phases which do not meet, this value is followed by an *. h Value of D at the melting point. i For metals which display several phases, the values of D are given at the temperature boundaries of the phase. For instance, Ub is stable between 941 and 1048 K, D values at these temperatures are given in column 9. j References mentioned in Ref. 30. k NA = not available.

Gorny and Altovskii (1970) Fromont et al. (1974) Fromont et al. (1974) Peterson and Rothman (1967) Horvath et al. (1984) Herzig and Eckseler (1979)

CHAPTER TWO

2.48

FIGURE 2.24 An Arrhenius plot (log10D versus 104/T) of the diffusion coefficients of a range of foreign/tracer solute elements in solid crystalline silicon and of silicon self-diffusion. The lines Au (1) and s Au (2) s correspond to different effective diffusivities of substitutional gold in silicon. (After U. Gösele and T. Y. Tan, in Encyclopaedia of Materials Science and Technology, vol. 4, Verlagsgesellschaft Chemie, Weinheim, Germany, pp. 197–247.)

È ∂ ln( fna 2 ) ˘ Ê ∂ ln D ˆ È ∂ (DG f + DG m ) ∂ P ˘ -Í Á ˜ =Í ˙ ˙˚ Ë ∂ P ¯T Î ∂P kT ˚T Î

(2.89)

Using the thermodynamic relation (∂G/∂P)T = V, we can define the activation volume DV of the diffusion process as DV =

∂ (DG f + DG m ) ∂ ln( fna 2 ) ∂ ln D = - kT + kT ∂P ∂P T ∂P T T

(2.90)

The second term in Eq. (2.90) can be estimated using the isotherm compressibility and the Gruneisen constant, but its value is only a few percent of the first term; thus, it can be neglected. Since the activation volume DV is the sum of the defect f formation volume DV d[= ∂(DG f )/∂P兩T] and defect migration volume DV md [= m ∂(DG )/∂P兩T], we have DV = DV df + DV md

(2.91)

DIFFUSION IN METALS AND ALLOYS

2.49

DCT DAT region I intrinsic

I Log D T

II

III association c si rin

t ex

IV

II precipitation

T –1(K–1) FIGURE 2.25 Schematic Arrhenius plot showing the typical temperature dependence of cation (DCT) and anion (DAT) tracer diffusion coefficients in a ceramic. (Source: A. Atkinson, in Encyclopaedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, p. 1177.)

Since the amount of DV md associated with the jump process is much smaller than f DV D, we may assume DVdf 艐 DV. Typically, DV varies between 0.5 and 1.3 W, at least in the case of monovacancy mechanism; however, in fcc metals, DV lies in the range of 0.7 to 1.1 atomic volume (W) and DV md is about 0.15 DV. In some cases, DV is very small or even negative, which can be an indication of an interstitial-type mechanism.30 For bcc sodium, a value of DV ~ 0.32 atomic volume reflects a stronger relaxation into the vacancy, whereas a dependence of DV on temperature indicates that more than one defect mechanism is operative, as observed for the curvature of the Arrhenius plot. The thermodynamic equation for DV df has been derived as65 W dE f ˆ DVdf = - Ê ˆ Ê Ë B ¯ Ë dW ¯

(2.92)

CHAPTER TWO

2.50

where W is the atomic volume, Ef is the vacancy formation energy under the constraint of constant volume, and B is the bulk modulus.66

2.6.3 The Isotopic Mass Effect The isotopic mass effect of solute atoms is of large significance in volume diffusion because it provides information about the atomic mechanism of diffusion and experimental means of occurrence of the tracer correlation factor fa of isotope a. It is customary to obtain the isotope effect E parameter (or the strength of the isotope effect) by measuring simultaneously the diffusion coefficients Da and Db of the two isotopes a and b of the same element with masses ma and mb, respectively, and using the relation E=

(Da ( mb

Db ) - 1 ma )

12

-1

= fa DK

(2.93)

where fa is the correlation factor for isotope a and DK is the fraction of kinetic energy in the unstable mode (i.e., at the saddle point residing in the jumping atom) associated with motion in jump direction that belongs to the diffusing atom. Hence, DK is bound between unity and zero (of the order of 0.8 to 0.9 for self-diffusion in simple fcc metals).67 The value of fa suggests the specific mechanism and indicates whether it changes with temperature in a given system. Consequently, a measurement of E should enable one to identify the corresponding diffusion mechanism. If the Cu, Ag, or Au diffuses fast in lead as a simple interstitial atom (like C in fcc iron), then fa = 1 and fa DK 艐 1. However, the value of fa DK for Ag diffusing in lead is 0.25, suggesting that the mass of the activated complex is much greater than most of the Ag atom alone.10 Equation (2.93) is applied to a process involving one atom jump. The more general case of isotopic mass effect involving n atoms jumping simultaneously, e.g., the interstitialcy mechanism (where n = 2), is expressed by E = fDK =

(Da

Db ) - 1

[ mb + ( n - 1) m0 ma + ( n - 1)m0 ]

12

-1

(2.94)

where m0 is the average mass of nontracers. Assuming the same relationship to hold in the case of grain boundary diffusion, we can rewrite Eq. (2.93) as Eb =

(Dba Dbb ) - 1 = fbDKb ( mb ma )1 2 - 1

(2.95)

where the subscript on the various symbols suggests the same quantity for grain boundary diffusion.68

2.7 TYPES OF DIFFUSION COEFFICIENTS There are various types of diffusion coefficients, which are briefly discussed in this section.

DIFFUSION IN METALS AND ALLOYS

2.51

2.7.1 Self-diffusion or Tracer Self-Diffusion Coefficient The self-diffusion or tracer diffusion coefficient D* is measured experimentally by A* introducing a few radioactive isotopes (A*) into pure A and measuring D*A (or DA ) by the relation

(

)

1 1 DAA* = DA* = f DAA ( = DA ) = f Ê ˆa 2 G = f Ê ˆa 2 zG Ë 6¯ Ë 6¯

(2.96a)

or more generally (2.96b)

D* = f Drandom

where G is the jump frequency for both A* and A atoms, z is the number of nearest neighbors, and f is the correlation factor (usually less than unity, which depends on the crystal structure and on the diffusion mechanism). Figure 2.26a shows the A* experimental situation for tracer self-diffusion coefficient DA (=D*A) when A* substitutes B*. The same idea can be extended to alloys and compounds. The tracer self-diffusion coefficients for both A* and B* tracer atoms in a homogeneous binary B* alloy AB (Fig. 2.26b) are represented by DA* AB and DAB, respectively. The A* or B* concentration is always kept very small so that the alloy composition remains unaffected by the diffusant. These coefficients are functions of concentration. According to Darken, simple thermodynamic considerations lead to the relationship between the intrinsic and the self-diffusion coefficients in binary solid solutions as ∂ ln g A ˆ Ê DA = DA* 1 + N A Ë ∂ NA ¯ ln g B ˆ ∂ Ê DB = DB* 1 + N B Ë ∂ NB ¯

and

(2.97) (2.98)

where DA ≠ DB because the solution is not ideal; DA and DB are the intrinsic diffusion coefficients, DA* and D*B are tracer self-diffusion (or tracer diffusion) coefficients, gA and gB are the activity coefficients of A and B, and NA and NB are the atom fractions (or fractional concentrations) of A and B, respectively. According to the Gibbs-Duhem equation, NA ∂ ln gA = -NB ∂ ln gB and since ∂NA = -∂NB, we have B*(or A*)

A

(a)

A* or B*

AB

(b)

A

B

(c)

FIGURE 2.26 Three types of diffusion experiments to determine (a) tracer self-diffusion or impu* and DB* rity diffusion coefficients D*A and D*B in pure metal; (b) tracer self-diffusion coefficients DAAB AB ˜ of two metals A and B. in homogeneous AB alloys; and (c) chemical diffusion coefficient D

CHAPTER TWO

2.52

1 + NA

∂ ln g A ∂ ln g B = 1 + NB ∂ NA ∂ NB

(2.99)

The two factors that give the intrinsic diffusivities when multiplied by the respective tracer diffusion coefficients are actually equal, and it is customary to call this quantity the thermodynamic factor. Experimental values for lattice self-diffusivity are available for about 50 metallic and semimetallic elements. Among these available data, some tend to be unreliable due to extreme difficulty in obtaining results for highly reactive materials.69 Results on self-diffusion in homogeneous dilute binary alloys containing small atomic fraction NB are usually denoted by A* A* ( N B ) = DAA*(1 + b1 N B + b2 N B2 + . . .) DAB = DAB

(2.100a)

A* B* Then DAB is called the solvent self-diffusion coefficient and DAB the solute difA* fusion coefficient. Experimental measurements of DAB (NB) are frequently given A* by Eq. (2.100a); and b1, b2, etc. are called solvent enhancement factors; DA is the tracer self-diffusion coefficient in the pure solvent. Mainly, b1 is determined by perturbations due to isolated solute atoms, b2 by pairs of solute atoms, and so forth. B* Similarly, the solute diffusion coefficient DAB at low concentration can be denoted by a power series dependence B* B* ( N B ) = DBA*(1 + B1 N B + B2 N B2 + . . .) = DAB DAB

(2.100b)

where DAB* is represented by impurity diffusion coefficient of species B in solvent A and B1 and B2 are the solute enhancement factors.12

2.7.2 Impurity and Solute Diffusion Coefficients Impurity diffusion denotes the diffusion of a solute element present in such low concentrations in a solvent (matrix) that the solute atoms may be considered as diffusing quite independently of one another, i.e., with no mutual interaction. It implies the simplest type of binary diffusion and is of special interest in furtherance of theoretical understanding.69a In order to measure the impurity diffusion coefficient, the tracer B* acts now as the impurity that is chemically different from the host or solvent. However, the impurity concentration must be kept extremely low to eliminate a chemical composition gradient (Fig. 2.26a). In reality, the tracer impurity should be allowed to diffuse into the specimen already containing the same concentration of impurity. Since the impurity always remains in stable solid solution (unless implanted), it is often called the solute and the impurity diffusion coefficient is sometimes called the solute diffusion coefficient at infinite dilution. However, the term solute diffusion coefficient is often “reserved” for dilute alloys, where measurement of the solvent diffusion coefficient is often necessary. In these situations, both solute and solvent diffusion coefficients often depend on solute content; and, as in self-diffusion, the chemical composition of the specimen must remain essentially unchanged by the diffusion process—otherwise it turns out to be a chemical diffusion coefficient.

DIFFUSION IN METALS AND ALLOYS

2.53

2.7.3 Chemical or Interdiffusion Coefficient Unlike self- and impurity diffusion coefficients which are measured in the absence of chemical composition gradients, chemical, interdiffusion, collective, or mutual dif˜ denotes the interdiffusion of A and B and is measured from the fusion coefficient D plot of CA or CB versus x. It is usually a function of concentration and depends markedly on temperature. Generally, the chemical diffusion coefficient does not equal the self-diffusion coefficient due to the effects resulting from the chemical composition gradients. Examples of this type of diffusion coefficient in a pseudoone-component system include diffusion of an adsorbed monolayer onto a clean section of a surface, diffusion between two metal specimens with difference in their relative concentration of a highly mobile interstitial atom such as hydrogen, and diffusion between two nonstoichiometric compounds such as Fe1-∂O and Fe1-∂¢O. Chemical diffusion in binary substitutional solid solutions is often known as interdiffusion and occurs when pure metal A is bonded to pure metal B and diffusion is allowed at elevated temperature (Fig. 2.26c). Although both A and B atoms move, only one independent concentration profile, say of A, is established. The resulting diffusion coefficient, which is derived from the profile by the BoltzmannMatano analysis, is called the interdiffusion, chemical diffusion, or mutual diffusion ˜ . For many practical purposes, D ˜ is adequate to explain the diffusion coefficient D behavior of a binary substitutional and is often quoted in the metals property databook. It is pointed out that the rates of diffusion of the two species relative to the local lattice planes are not equal in amount. For chemical diffusion in systems of more than two components, Eq. (2.5) or (2.6) and those following are insufficient.5 ˜ is simply given by an Arrhenius equation In many cases, D -Q ˆ D˜ = D˜ 0 expÊ Ë RT ¯

(2.101)

As the limit of concentration of one metal species approaches zero, the interdiffusion coefficient approaches the intrinsic diffusion coefficient. 2.7.4 Intrinsic Diffusion Coefficient The intrinsic diffusion coefficients (or component diffusion coefficients) DA and DB of an AB alloy denote the diffusion of the two species A and B relative to the lattice planes. In general, the diffusion rates of A and B are not equal.12

2.8 SELF-DIFFUSION IN PURE METALS Self-diffusion is the most basic diffusion process in solids by which substitutional host atoms move through the lattice to other substitutional sites. For the majority of metals, the sum of formation and migration energies of vacancies is equal to the activation energy for the self-diffusion measured for that metal. In this way, the single- or monovacancy mechanism plays a dominant role for self-diffusion in pure metals.70 Experimentally, the diffusion coefficient is measured by employing radioactive pure metals and monitoring their movement through the metal lattice. Self-diffusion parameters for various pure elements are listed in Table 2.4. The pure metals are most extensively studied with respect to point defects and diffusion properties. Self-diffusion of metals is usually divided into normal and anomalous

2.54

CHAPTER TWO

self-diffusion. The characteristics of normal self-diffusion (Fig. 2.27a) are as follows: (1) A single vacancy mechanism is operative, and D (or D*) is represented by the Arrhenius relation: D = D0 exp (-Q/kT) is obeyed which implies the occurrence of a straight-line diffusion data plot of lnD versus 1/T. (2) The D0 (preexponential factor) values lie in the range 5 ¥ 10-6 and 5 ¥ 10-4 m2◊s-1. (3) A relationship between the value of Q (activation enthalpy) and the melting temperature Tm by the expression Q = 34Tm cal/mol or Q = 141Tm J/mol and Tm (in kelvins). Examples of metals in this class include the fcc metals such as Al, Ag, Au, Cu, Ni, and Pt and bcc metals such as Cr, Li, Na, Nb, Ta, and V. This behavior forms the basis of the Van Liempt relation and is well established in compact metals (see Fig. 2.22a for fcc metals and Fig. 2.22b for bcc metals). It is noted here that bcc structures exhibit a wider dispersion around the Van Liempt straight line than that of the fcc structures. Metals whose properties do not conform to normal characteristics are called anomalous. These are usually specific phases of allotropic metals for which several distinct solid phases are present. Examples of the anomalous self-diffusion behavior occurring in bcc metals are V, Cr, b-Hf, b-Zr, b-Ti, g-U, e-Pu, g-La, d-Ce, b-Pr, g-Yb, and b-Gd. All these metals are characterized by very low values of D0 (and Dmelt.point 艐 10-10 m2◊s-1) and Q at low temperatures when compared to normal metals22,71,72 (Fig. 2.27a); and a clearly visible (upward concave) curvature of the “Arrhenius plot.”73 The diffusion coefficient of most bcc metals at the melting temperature is 10-11 m2◊s-1, an order of magnitude higher than in most fcc metals. Various theories for this anomalous behavior have been advanced, and most of them have been discounted. The following explanation is presently the most favored one. The anomalous behavior is usually interpreted as resulting from (1) a strong temperature dependence of the vacancy migration energies, (2) the existence of allotropy or phase transformation associated with very high densities of short-circuiting paths such as dislocations and grain boundaries, (3) lowering of activation (or jump) barrier at lower temperatures, (4) unusually large divacancy binding energy, (5) two possible types of monovacancy jumps such as nn and nnn, and (6) a “dissociative” mechanism, in which mainly the substitutional atoms have a considerable probability of being excited into a highly mobile interstitial state. Among these, the most likely explanation of the anomalies is the lowering of the jump barrier at lower temperature.74 These are linked to the fragments or embryos of lower-temperature w phase occurring near the transition. This w phase, which is structurally related to the bcc lattice when a migrating atom is at the saddle point, could enhance diffusion by providing easier jumps.75 If we assume, for simplification, temperature dependence of Q and D0 values, then Eq. (2.102) is used to explain a curvature in the plot of ln D versus 1/T -Qi ˆ D = Â A i expÊ Ë kT ¯

(2.102)

In such an equation, two or three terms are normally sufficient; each term represents a competing diffusion mechanism. One of the terms is possibly due to a single vacancy mechanism that is dominant in fcc metals. There are several possibilities for the other terms, and different mechanisms can actually work in different systems.

2.8.1 Self-diffusion in FCC and HCP Metals The vacancy mechanism for the diffusion in these metals has now been widely recognized. However, there are still some controversies with respect to the origin of

10

10

–8

–10

g-La

O in b-Ti Co in b-Zr Fe in b-Zr O in b-Zr d-Fe

10

–12

10

–14

b-Hf b-Ti

b-Zr

D (m /s)

Na

2

K Li 10

–16

Ta Nb V

10–18

W 10

1.0

10

Mo Cr

–20

1.2

1.4

1.6 Tm / T (a)

1.8

2.0

–10

10

–12

10

–14

Al

10–16 D (m /s)

Pt

2

Au 10–18 Ag Pd 10

FIGURE 2.27 Self-diffusion coefficients D of (a) bcc metals and diffusion coefficients of the very fast impurity diffusion and (b) fcc metals as a function of their homologous temperature Tm/T.70,71

Cu

–20

Ni

10–22 1.0

2.55

1.2

1.4

1.6 Tm / T (b)

1.8

2.0

2.56

CHAPTER TWO

the possible curvature of the Arrhenius plots even in the so-called normal metals (Al, Cu, Ni, Ag, and Au) profoundly at high temperature and sporadic presence over the entire temperature range from Tm to Tm/2 (Fig. 2.27b). In order to elucidate this curvature, three hypotheses have been put forward: 1. A vacancy mechanism takes place over the whole temperature range. However, D0 and Q increase with temperature due to strong thermal expansion coefficient for vacancies, or variation of elastic constants with the temperature. In these temperature ranges, it was found that the data did not follow an Arrhenius behavior; that is, ln D versus 1/T plots were curved. 2. Both vacancies and divacancies contribute to the diffusion; however, the single vacancy mechanism dominates at low temperatures, whereas there is an increasing contribution from divacancies at the higher temperatures approaching Tm.70,76 3. A vacancy mechanism takes place, and the curvature is attributed to the dynamic correlation between successive jumps (vacancy double jumps).77 Figure 2.27b also exhibits that, in Group IB metals, the D values in Cu became the lowest, followed by those of Ag and Cu, and in the Group VIII metals, the Ni values were the lowest followed by Pd and Pt.70 In hcp metals the limited number of available data are in agreement with a slight decrease of the ratio of the activation energies of the diffusion parallel to perpendicular to the c axis with increasing c/a ratio; the activation energy remains the same in the perfect lattice.78 2.8.2 Self-diffusion in BCC Metals Self-diffusion in bcc metals exhibits three features: (1) There is much larger scatter of diffusivity in bcc metals than in the compact phases, and some exhibit an exceptionally large absolute value of D (Fig. 2.27a). (2) Their Arrhenius plots often display much larger curvature than those of the fcc or hcp metals, mostly accounted for by a divacancy mechanism. (3) They show an orderly variation of D with the position in the periodic table which needs to be elaborated; e.g., metals of the same column, like Ti, Zr, Hf in Group IV, have a very small activation energy and a large curvature (Fig. 2.27a). The self-diffusion in Group IV metals is much faster than in others; the lowest the temperature, the largest this difference.79 The diffusivities of some transition impurities such as Fe, Co, and Ni are orders of magnitude faster than the already fast diffusion coefficient. Many explanations have been proposed to explain these anomalies based on strong contribution of short circuits, presence of extrinsic vacancies due to impurities, interstitial mechanisms, etc. All these hypotheses were excluded by experiments. The very origin of this behavior is now understood to be associated with the electronic structure of the metal and the structural properties of the bcc lattice.80

2.9 DIFFUSION IN DILUTE SUBSTITUTIONAL ALLOYS In this case, dilute solute concentrations imply those below 1 at %, where one can avoid the complications of solute-solute interactions. Although fcc alloys have been mainly investigated in which defect mechanisms are needed, the same approach may

DIFFUSION IN METALS AND ALLOYS

2.57

be valid to all systems where the atomic migration involves motion of vacancies or interstitials or more complex defects formed from them (such as solute-vacancy pairs).81 Analysis has been accomplished in terms of the five-jump-frequency model of fcc alloys involving the various vacancy jumps near an impurity atom, as shown in Fig. 2.28.82 The jump frequencies w0, w1, w3, and w4 all involve solvent-vacancy exchanges, each in a different relationship to the solute atom. Only w2 is the frequency of a solute (impurity)-vacancy exchange, and the solute diffusion coefficient is given by Ê DGvf + DGA ˆ D2 = a 2w2 f2 expÁ ˜ Ë ¯ kT

(2.103)

where DGA is the solute-vacancy association free energy. The major problem, however, is that the correlation factor f2 is no more a constant but now is dependent in a complex fashion on all the various jump frequencies.28 For the normal dilute alloy systems, the experimental results provide an activation energy for D2 within 25% of that of the value for solvent self-diffusion, and a preexponential D02 within one order of magnitude of that of the solvent. This similarity helps to advocate the vacancy mechanism for solute diffusion. The (solvent) self-diffusion coefficient Ds(c2), in a dilute random alloy containing a low concentration c2 of solute/impurity atoms, can be written in the form Ds (c2 ) = Ds (0)(1 + bc2 )

(2.104)

where Ds(0) is the self-diffusion coefficient of the pure solvent and b is a constant (for a given alloy system at a given temperature) called the solvent enhancement factor or coefficient (either positive or negative for particular alloy models). Positive b values may be attributed to (1) the effect of solute atoms in enhancing the total vacancy concentration in the crystal and (2) the effect of changed solvent jump frequencies (w1, w2, and w4) near solvent atoms. Expressions for b can be very complex, based on the degree of approximation used. In entirety, the five-frequency model holds good in explaining data on both solute diffusion and solvent enhancement for a wide range of dilute alloy systems.28 Anomalous Dilute Systems. In some dilute alloys one observes striking departures from the just discussed behavior such as D2 >> Ds by factors of 103 to 105 times,

FIGURE 2.28 The five-jump frequency model for diffusion near an impurity atom in dilute fcc substitutional solid solutions.82

CHAPTER TWO

2.58

TABLE 2.5 Dilute Alloy Systems in which Anomalously Fast Solute Diffusion Occurs80 Solutes

Solvents

1. Low valency groups I and II Cu, Ag, Au, Be, Zn, Cd, Hg, Pd 2. Later transition metals Ni, Co, Fe, Cr, Mn, Pd 3. Noble metals

High valency groups III and IV Pb, Sn, Tl, In Early members of d-transition groups b-Ti, b-Zr, b-Hf, Nb, La, Ce, Pr, Nd, g-U, Pu Li, Na, K

and an activation energy Q2 by only approximately one-third to one-half of Qs. Table 2.5 lists many of these (classified into three groups by Le Claire).80 Such effects are most significantly found for dilute alloys of Pb, Na, and In containing the noble metals as solutes where interstitial solute diffusivity is often much larger than the solvent diffusivity. To elucidate these effects by a vacancy mechanism, it would need a very large solute-vacancy association/binding energy DGA and both a rapid solutevacancy exchange rate (w2) and a rapid vacancy jump rate around the solute (w1). These requirements can be written quantitatively. However, these requirements would impart a very high enhancement factor b [Eq. (2.104)], which is not observed. Alternatively, it was suggested that in such systems there is a fraction yi of solute atoms that fills interstitial sites. The diffusion coefficient D2 is then given by D2 = yi Di + (1 - yi )Ds

(2.105)

where Di and Ds are the interstitial and substitutional diffusivities, respectively. Since Di >> Ds, the first term may dominate even if yi DB. In this situation, across any lattice plane in the concentration gradient, more atoms of A migrate in one direction than do atoms of B in the opposite direction; i.e., a net mass transfer of atoms takes place across the plane and is associated with an equal flux of vacancies in the opposite direction. These unbalanced intrinsic fluxes result in (1) an increase in volume of the B-rich part of the sample to accommodate the net positive inward flux of matter, (2) shift of the lattice plane (or marker movement) toward the A-rich part (i.e., the shrinking side), thereby a state of compression in this region, and (3) generation or agglomeration of vacancies, voids, or pores on the A-rich side of the couple, commonly called Kirkendall voids or porosity (thereby causing the region in a state of tension). With continued exposure to high temperature, the voids increase and coalesce, producing porosity and loss of strength. This shift of crystal lattice (and related phenomena) during mutual diffusion was first observed by Smigelkas and Kirkendall on Cu-brass diffusion couple.88 This experiment rejects the assumption of direct exchange of A ´ B mechanism which was formerly proposed and which would have implied equal diffusivity for both atoms. This phenomenon has been, subsequently, observed to occur in thin-foil couples of Ag-Au, Al-Au, Au-Sn, Au-Pb, Au-Sn/Pb, Cu-Al, Cu-Ni, Cu-Pt, Ni-Mo,89 and Pt-Ir. The Kirkendall effect is time- and temperature-dependent but with some systems can even occur at ambient temperatures. It is concluded, based on the effects mentioned in the previous paragraph, that a composition profile exists after interdiffusion of A and B (Fig. 2.31a); voids or pores form in the diffusion zone area from which there is a mass flow; and a net flow of vacancies occurs from the B-rich side of the bar toward the A-rich side (Fig. 2.31b). This movement increases the equilibrium number of vacancies on the A-rich side and reduces it on the B-rich side. The vacancies are thus created on the side of a couple that gains mass and are absorbed or destroyed on the side that loses mass (Fig. 2.31c). Kirkendall porosity can be eliminated by careful selection of alloy composition. For example, Kirkendall voids observed in Pt-Cu couple can be avoided by select-

100 percent A Pure metal A

Pure metal B Wire

Interface or weld

Original interface

(a)

FIGURE 2.30

Wire

Penetration curve (percent A)

Marker shift in a Kirkendall diffusion couple.10

(b)

O percent A

CHAPTER TWO

2.62

JA JV 1

CA

P

JB 0

A

B

0

(a)

CB

1

X J JA

+

0

X

JV JB – (b)

∂Cv ∂Jv =– ∂t ∂x

Vacancies must be destroyed +

0

X –

(c)

Vacancies must be created

FIGURE 2.31 Schematic diagram showing interdiffusion and vacancy flow. (a) Concentration versus distance curve for A component. (b) The corresponding fluxes of A and B (DA > DB) and vacancies as a function of position x. (c) ∂CV/∂T = -∂JV/∂x. That is, the rate of vacancy creation is equal to the rate of vacancy destruction or absorption.14 [Reprinted by permission of Van Nostrand Reinhold (U.K.) Ltd.]

ing Pt and electrodeposited Ni after annealing for 8 hr at 700°C. Annealing in a hot isostatic press (HIP) has also been used to suppress Kirkendall void formation. One advantage of the Kirkendall effect is the development of deliberately controlled porosity. 2.10.2.2 Darken’s Analysis. Fick’s law contains the implicit assumption that the driving force for diffusion is the concentration gradient. However, a more basic viewpoint assumes that the driving force is a chemical free-energy gradient. There

DIFFUSION IN METALS AND ALLOYS

2.63

is a direct similarity between the two gradients, but it is seldom that the relationship becomes inverted and the so-called uphill diffusion, i.e., diffusion against a concentration gradient, occurs. Darken provided the following analysis to describe the Kirkendall effect. Let us suppose that the vacancy mechanism of transport is operating. In every experiment, the inert markers are invariably made of insoluble materials with high melting temperature. The formation and migration energies of the vacancy in such materials are significantly larger than in the surrounding matrix. As a result, the markers are impermeable to the vacancy flux. In this situation, it can be shown that such a marker shifts along with the lattice planes,90 irrespective of the type of its interface with the matrix (coherent or incoherent). Thus the measurement of Kirkendall shift is considered as the measurement of the lattice plane shift. The above formalism can be easily extended to incorporate the situation where the average atomic volume changes with the concentration of the alloy.91 Since the net intrinsic flux of atoms across a plane is associated with an equal flux of vacancies in the opposite direction, we can write J V = -( J A + J B )

(2.110)

as shown in Fig. 2.31b. The flux can be maintained by the creation of new vacancies at sources upstream of the flux and by their destruction at sinks downstream (Fig. 2.31c), both processes striving to maintain equilibrium concentration of vacancies characteristic of the sample at each composition along the gradient. The most probable sources and sinks include dislocations generating and annihilating vacancies by climb processes. But such generation and annihilation involve, respectively, an increase and a decrease in the number of lattice planes and thus provide the mechanisms for exactly the amounts of expansion and shrinkage necessary to accommodate the net atom flux. ˜ in terms of the intrinsic (or comTo express chemical interdiffusion coefficient D ponent) diffusion coefficients DA and DB, we describe the fluxes J¢A and J¢B of A and ˜ elucidates the B, respectively, with respect to the local lattice plane. Furthermore, D interdiffusion fluxes J˜A = -J˜B measured with respect to laboratory fixed axes. In the simplest case, to analyze this problem, we require the flux of the two components relative to the fixed end of the sample and velocity of a lattice plane referred to these axes; hence, with J A¢ = - DA

and

∂ CA , ∂x

etc.

∂ CA ∂ CB =∂x ∂x

(2.111) (2.112)

we can express the sum of the fluxes of both components Jnet as ∂ CA ˆ Jnet = J A¢ + JB¢ = (DB - DA )Ê = J A + JB - (CA + CB )V Ë ∂x ¯

(2.113)

If we assume that (1) the vacancies condense onto the lattice planes normal to the diffusion flow, (2) the volume per lattice sites is constant, i.e., atomic volume is constant and does not vary with composition, and (3) net flux Jnet is zero by definition when the volume change on mixing is zero, then we find that the Kirkendall or marker velocity will equal the net flow of atoms, which is the vacancy flux times the atomic volume

CHAPTER TWO

2.64

∂ NA ˆ V = (DB - DA )Ê = JV W = -( J A + JB )W Ë ∂x ¯

(2.114)

∂ CA ˆ J A = ( N BDA + N ADB )Ê = - JB Ë ∂x ¯

(2.115)

and

where ∂NA/∂x denotes the concentration gradient at the marker position. ˜ is the same for both species and The mutual, chemical, or interdiffusivity D equal to ˜ = NBDA + NADB D

(2.116)

where NA and NB are the molar (or atomic) fraction of species A and B, respec˜ and tively. So DA and DB can be determined separately from measurements of D ˜ and V can be easily written that allow for volume changes and V. Equations for D are required to be used in accurate work when these are appreciable, but it will not be considered here.92 It is possible to express Darken’s relation between tracer diffusion coefficients and chemical diffusion coefficients by using the Darken’s equation (2.116), which is given by N A ∂ ln g A ˆ N B ∂ ln g B ˆ Ê Ê + N ADB* 1 + D˜ = N BDA* 1 + Ë Ë ∂ NB ¯ ∂ NA ¯

(2.117a)

or

N A ∂ ln g A ˆ Ê D˜ = ( N BDA* + N ADB* ) 1 + Ë ∂ NA ¯

(2.117b)

or

∂ ln g ˆ DA = DA* Ê 1 + Ë ∂ ln c ¯

(2.118)

Because the two forms of the thermodynamic factors are equal and simple, it provides an obvious convenience. This is also called the Darken equation. Figure 2.32 shows how well this equation is obeyed in the Au-Ni system. 2.10.2.3 Vacancy Wind Effect. A very important assumption in the above derivation is that the compensation of atom flux differences by bulk lattice motion is complete and achieved by motion only along the diffusion direction. In the absence of kinetic cross-interactions between diffusing components, intrinsic diffusion can be explained according to a simple atomic mobility model. For systems involving diffusional kinetic interactions among components, Manning expressed the interactions in terms of the vacancy wind effect where the net vacancy flux influences a component’s intrinsic flux.93 For a random concentrated alloy for which there is no solute-vacancy binding, the Manning approximation assumes that, at infinite dilution, the solid solution becomes ideal (j = 1) and the intrinsic diffusion coefficient DA must tend toward the tracer diffusion coefficient DA*. Hence, DA = DA*j

and

DB = DB*j

where j = 1 +

∂ ln g ∂ ln c

(2.119)

These relations are again called Darken’s equations. Furthermore, from thermodynamics of irreversible processes, we know that the off-diagonal terms cannot be ignored. The more general expression can, therefore, be given by

DIFFUSION IN METALS AND ALLOYS

2.65

FIGURE 2.32 (a) Calculated and observed interdiffusion coefficients in gold-nickel alloys at 1173 K. (b) Corresponding thermodynamic factor for interdiffusion. (Source: J. E. Reynolds, B. L. Averbach, and M. Cohen, Acta Met., 1957, p. 29.)

DA =

kT Ê LAA LAB ˆ j N Ë NA NB ¯

and

DB =

kT Ê LBB LBA ˆ j N Ë NB N A ¯

(2.120)

where k is the Boltzmann constant; N is the total number of sites per unit volume; LAA, LAB, LBA, and LBB are the phenomenological coefficients which are functions of temperature, concentration, etc.; and according to the Onsager reciprocity relation, Lij = Lji (i.e., LAB = LBA). In a particular case of a simplified random-alloy model for which j = 1, these expressions are assumed to hold even for a nonrandom alloy in which the thermodynamic factor j is no longer unity. Therefore, the final expression for the intrinsic diffusivities is given by equations DA = DA* ABj rA

and

B* DB = DAB j rB

(2.121)

CHAPTER TWO

2.66

where rA and rB denote the vacancy wind corrections. Hence 2 È 2 N A N B (DA* - DB* ) ˘ D˜ = ( N ADB* + N BDA* )j Í1 + ˙ Î M0 D * ( N ADB* + N BDA* ) ˚

(2.122)

Here, M0 is a pure number that is based on the crystal structure. The rightmost term in the square brackets is known as the Manning vacancy wind term or correction because it represents the coupling between the transport of species A and B through the vacancy flux. It is noted that Manning’s equations predict a chemical diffusion ˜ always greater than that given by Darken’s equations. It is clear that coefficient D the vacancy wind factor rm depends on the physical effect that the isotopes of A and B differ. These vacancy wind factors and phenomenological coefficients can be found for dilute alloys by using a kinetic theory for the mobility of a substitutional tracer.2 Thus the vacancy wind effect can improve or retard atomic flow, respectively, when it is opposite to or in the same direction as the net vacancy flux.93

2.10.3 Ternary Diffusion Diffusion phenomena in multicomponent systems are complicated and encountered in a variety of materials and processes. The systems include steels, high-temperature alloys, coatings and composites, and processes ranging from diffusion bonding, cladding, and controlled heat treatment to surface modifications.94 Diffusion against a concentration gradient in a ternary or higher-order couple can take place due to thermodynamic interactions, Onsager correlation cross-effects, vacancy winds and inhomogeneities, electrostatic neutralization interactions in ionic solids, and electron-solute interactions in semiconductors.95 One problem associated with multicomponent diffusion is the prediction of the diffusion path between two terminal alloys. In general, the diffusion path is defined as a line from one terminal alloy to another on the ternary isotherm, denoting the locus of the average composition in planes parallel to the original interface throughout the diffusion path. It also corresponds to the morphology of the diffusion area. It depends on the gradient of chemical potentials and mobilities of each species, taking into consideration the mass-balance requirements. If the phases are separated by planar interfaces, the diffusion path can cross the two-phase region along a tie line, and along the entire interface; here the same local equilibrium is assumed. For ternary and multicomponent diffusion, the concept of zero-flux planes has been developed by Dayananda and co-workers.96–98 These are defined as the positions where the interdiffusion flux of one component approaches zero, resulting in an interdiffusion flux reversal for that component.99 In a ternary system, Onsager’s phenomenological diffusion coefficients forming an L (3 ¥ 3) matrix have served as the basis for both theoretical and experimental diffusion studies. In substitutional alloys, the frequency of direct atom interchanges is small compared to atom-vacancy interchanges because it is assumed that metal diffusion occurs only by exchange with the neighboring vacancies. If vacancies are in equilibrium in the system and the contributions of vacancy wind and correlation effects are omitted, it is possible to eliminate the nondiagonal, or Onsager coefficients in the Lij matrix. Based on the Kirkendall frame of reference, the intrinsic diffusion fluxes Ji (mol◊cm-2◊s-1) can be given by100

DIFFUSION IN METALS AND ALLOYS

2.67

J1 = - L11

∂ m1 ∂ C1 ∂ C2 = - D11 - D12 ∂x ∂x ∂x

(2.123a)

J 2 = - L22

∂ m2 ∂ C1 ∂ C2 = - D21 - D22 ∂x ∂x ∂x

(2.123b)

J3 = - L33

∂ m3 ∂ C1 ∂ C2 = - D31 - D32 ∂x ∂x ∂x

(2.123c)

where Ci is the concentration (mol◊cm-3), x the distance parameter (cm), mi the chemical potential (J◊mol-1), and Dij the intrinsic diffusion coefficients (cm2◊s-1). The relationship between Onsager diffusion coefficients Lij and tracer diffusion coefficients D*i , according to Darken and Le Claire, is given by Ci Di* RT

(2.124)

Ci Di* ∂ mi N i Di* ∂ mi =RT ∂ x RTVm ∂ x

(2.125)

Lij = which yields Ji = -

where Ni is the atomic fraction of component i and Vm is the molar volume (cm3/mol of atoms), assumed to be constant in this analysis. Since activity ai is written as mi = m0i + RT ln ai

(2.126)

N i Di* ∂ ai Di* ∂ ai =Vm ai ∂ x g iVm ∂ x

(2.127)

this gives Ji = -

where gi = ai/Ni is the activity coefficient of species i. In an ideal ternary system, the relationships among the interdiffusion fluxes J˜i and intrinsic diffusion fluxes Ji; inter˜ ijn , and concentration gradients ∂Ni /∂x; and tracer diffusion diffusion coefficients D coefficients D*i , and chemical potential gradients ∂mi /∂x can be given by 3

J˜i = Ji - N i  J j

(2.128)

j =1

3 D˜ 11 J˜1 = Vm 3 D˜ 21 J˜ 2 = Vm

3 ∂ N 1 D˜ 12 Vm ∂x 3 ∂ N 1 D˜ 22 Vm ∂x

∂ N2 ∂x

(2.129a)

∂ N2 ∂x

(2.129b)

J˜3 = - J˜1 - J˜ 2

(2.129c)

or through Eqs. (2.125) to (2.128) Ê N 1D1* ˆ Ê ∂ m1 ˆ ÈÊ N 1D1* ˆ Ê ∂ m1 ˆ J˜1 = - Á + N 1 ÍÁ ˜ ˜ Ë RTVm ¯ Ë ∂ x ¯ ÎË RTVm ¯ Ë ∂ x ¯ Ê N2 D*2 +Á Ë RTVm

ˆ Ê ∂ m 2 ˆ Ê N3 D*3 +Á ˜Ë ¯ ∂ x ¯ Ë RTVm

ˆ Ê ∂ m3 ˆ ˘ ˜Ë ¯ ∂ x ¯ ˙˚

(2.130a)

CHAPTER TWO

2.68

Ê N2 D*2 ˆ Ê ∂ m 2 ˆ ÈÊ N 1D1* ˆ Ê ∂ m1 ˆ J˜ 2 = - Á + N 2 ÍÁ ˜ ˜ Ë RTVm ¯ Ë ∂ x ¯ ÎË RTVm ¯ Ë ∂ x ¯ Ê N2 D*2 +Á Ë RTVm

ˆ Ê ∂ m 2 ˆ Ê N3 D*3 +Á ˜Ë ¯ ∂ x ¯ Ë RTVm J˜3 = - J˜1 - J˜ 2

ˆ Ê ∂ m3 ˆ ˘ ˜Ë ¯ ∂ x ¯ ˙˚

(2.130b) (2.130c)

Using the known values of tracer diffusion coefficients, the vacancy flux JV can be calculated from the equation 3

JV = - Â J j = j =1

+

N 1D1* ∂ m1 N 2 D*2 ∂ m 2 + RTVm ∂ x RTVm ∂ x

N3 D*3 ∂ m3 RTVm ∂ x

(2.131)

Using the Boltzmann-Matano analysis for multicomponent systems, the interdiffusion fluxes J˜i can be calculated from the experimental diffusion profile as give by Ê 1 ˆ J˜i = Ë 2tVm ¯

Ni¢

Ú

x dN i

(2.132)

Ni ( -•)

where the origin (x = 0) is the Matano plane or interface. At the original (Kirkendall) plane, a relation between the vacancy flux and the marker shift Dxm is given by the equation Dxm = Vm JV 2t

(2.133)

In all analyses, note that SiNi = 1, Si J˜i = 0, and SiNi ∂mi = 0, which implies that according to Eq. (2.125), S(Ji/D*i ) = 0. Since the interdiffusion and intrinsic fluxes are related by the lattice or marker velocity V, the interdiffusion coefficients can be represented in terms of the intrinsic coefficients by 3

D˜ ij3 = Dij3 - N i  Dkj3

(i, j = 1, 2)

(2.134)

k =1

where Ni is the molar fraction of component i and the molar volume Vm is assumed to be constant. Usually, all the diffusion coefficients are dependent on composition. The Dij3 are interrelated by three atomic mobilities and thermodynamic data.100a For a ternary system, a pair of solid/solid couples A/B and C/D with intersecting diffusion paths is required, as shown in Fig. 2.33. Here, the composition identified by the intersection point I is common to both couples, and it is only at this compo˜ 3ij . sition we can measure the four D Ternary diffusion couple experiments elucidated by Darken involve welding a bar of g-Fe–0.4% C with g-Fe–0.4% C–4% Si and diffusion annealing (or austenitizing) at 1050°C. Figure 2.34 shows that initially a sudden rise in the carbon concentration develops on both sides of the weld plane because of the very rapid diffusion of interstitial carbon with respect to silicon which, although contrary to Fick’s first law, decreases the mc (chemical potential of carbon) gradient caused by the difference in Si concentration. In the long term, after equilibrium of Si content

DIFFUSION IN METALS AND ALLOYS

2.69

FIGURE 2.33 Schematic diffusion paths for a pair of ternary solid/solid diffusion couples A/B and C/D with a common composition point (C1, C2) of intersection I where four ternary interdiffusion 3 ˜ 311, D ˜ 12 ˜ 321, and D ˜ 322 coefficients D ,D can be determined.96 (Reprinted by permission of Springer-Verlag, Berlin, Germany.)

0.6 Carbon, percent

0.586%C 3.80%Si 0.5

0.4

0.478%C

0.441%C 0.315%C

0.3 –1.0

–0.5

0

0.5

1.0

FIGURE 2.34 Distribution of carbon after a 13-day anneal at 1050°C in a diffusion couple between g-Fe-0.44% C (right) and g-Fe-0.4% C-4% Si (left). The chemical potential of carbon is continuous and monotonic across the couple throughout the diffusion anneal. (Source: L. S. Darken, Trans. AIME, vol. 180, 1949, p. 430.)

on either side of the weld plane, the carbon content is redistributed and maintains a constant chemical potential again. Figure 2.35 shows the concentration-time paths for two points to the left and to the right of the weld plane.

2.11 ELECTRO- AND THERMOMIGRATION Electromigration is a diffusion-controlled mass transport phenomenon where the forced movement of individual atoms (or metal ions) occurs along the length of a conducting element under the direct influence of an applied electric field, current flow, or high current density. In metals and alloys, electrons carry practically all the electric current, and the ratio of electrons to atomic currents is high. It is clear that most of the atomic transport arises from the momentum transfer between large flux of electrons and lattice ions (electron winds) (which is attributed to the collision of electrons with atoms).10 The direction of the drift movement in the electric field depends primarily on whether the alloy conducts by electrons or holes. In reality,

2.70

CHAPTER TWO

FIGURE 2.35 Schematic diagram showing the change in composition with time of two points A and B on opposite sides of the weld plane (diffusion couple) in Darken’s diffusion couple of g-Fe-0.4% C and g-Fe-0.4% C-4% Si. C is the final equilibrium composition of the whole bar. (Source: L. S. Darken, Trans. AIME, vol. 180, 1949, p. 430.)

the electron wind effect tends to drag atoms which are changing places toward the anode. However, the greatest influence is observed in the case of a-iron where highly mobile interstitial solute carbon atom in metals moves toward the cathode.43 This is an important mechanism responsible for several different kinds of failure in many electrical and microelectronics systems such as tungsten lightbulb filaments and computer chips. The most common are void failures along the length of narrow conducting lines (called internal failures) and diffusive displacements at the terminals of the line that destroy electrical contact. Recent research has determined that both of these failure modes are attributed to the microstructure of the conducting line and can, therefore, be inhibited by controlling or changing the microstructure.101 This has assumed increasing importance in the performance and reliability of packaging systems that involve electrical contacts (Fig. 2.36). Thus, electromigrationinduced damage in metallic conductors is a complex field which ranges from microscopic damage mechanisms to reliability modeling.102 Electromigration often occurs in the presence of uneven temperature distributions that develop at various sites within device structures, such as at locations near hot spots, which are caused by occasional nonadhesion of the stripe to the substrate; in areas of different thermal conductivity, such as metal-semiconductor contacts or interconnect-dielectric crossovers; at nonuniformly covered steps; and at terminals of increased cross section. In addition to the effect of microstructure and phase morphology, the temperature gradient plays a key role in adding complications.103 Matter flow induced by a temperature gradient is called thermotransport.104

DIFFUSION IN METALS AND ALLOYS

Before

2.71

After

FIGURE 2.36 SEM micrograph showing void formation after reliability test in thin-film metallization due to electromigration. (Source: A. Christou and M. C. Peckerar, in Electromigration and Electronic Device Degradation, ed. A. Christou, Wiley, New York, 1994, pp. 105–137.)

TABLE 2.6 Different Modes of Metallic Electromigration105 A. Electrolytic (ionic) Ambient temperatures (104 A/cm2)

Thermotransport is usually a second-order effect with respect to electromigration and, therefore, can only occur when electromigration has been subdued as, for example, under an ac stressing or bidirectional pulsing.104 Table 2.6 shows the different modes of metallic electromigration. Electrolytic electromigration occurs primarily under normal ambient conditions when the local temperatures and current densities are low enough to allow water to be present on the surface. Solid-state electromigration has received much attention in microelectronics due to its role in producing failures in integrated circuits. For more details, see the article by Krumbein.105 Thermomigration, i.e., the diffusion of atoms under the influence of a temperature gradient, is also known to contribute to voids in thin-film metallization. In conclusion, the thin-film connecting stripes between the individual transistors, etc., are

2.72

CHAPTER TWO

especially susceptible to failures due to diffusion-controlled processes.104 Although the magnitude of thermomigration is usually small compared to electromigration, there can be occasions where temperature gradients are very steep to effect diffusion. Finally, it is a method to study the electronic structures of point defects (vacancies, impurity atoms) at elevated temperatures and its variation during a jump. To understand and solve numerous diffusion-related problems such as electromigration, thermomigration, hillock growth, etc., the availability of diffusion data in thin films at low temperatures is of great significance.

2.11.1 Electromigration in Thin Films Material transport phenomena on thin-film surfaces are of fundamental and technological interest. These processes include electromigration, thermomigration, and mechanodiffusion (diffusion in stress gradients). In multiphase materials, diffusion caused by a concentration gradient is often associated with these phenomena. Thin metallic films are used for interconnects in integrated circuits, for superconducting devices, magnetic memories, etc., and their grain structure and preferred orientations can have important bearings on their performance. Thin-film metallic conductors are normally characterized by a polycrystalline microstructure comprising dense crystalline grains separated by an interconnected system of relatively open grain boundaries (GBs), as shown schematically in Fig. 2.37. These GBs usually provide short-circuit paths for interdiffusion, which permits the conductor to conform with external (driving) forces, thereby accelerating the degradation of requisite physical properties and reducing ultimate time to failure.106 The atomic or ionic flux J under usual drift conditions (i.e., induced by internal mechanical stress caused by nonequilibrium deposition and processing conditions,

FIGURE 2.37 Schematic representation of a small segment of an idealized (two-dimensional) microstructure in thin-film conductor, comprising impenetrable (hexagonal) grains isolated by highconductivity grain boundaries. Usually the corresponding mass flux (denoted by arrows) can be directed toward or away from individual grain boundary triple junctions, causing localized mass accumulation or depletion.

DIFFUSION IN METALS AND ALLOYS

2.73

differential thermal expansion and contraction of the film, and substrate and/or chemical/interfacial reactions) is given by the Nernst-Einstein equation NDb ˆ NDb dm F= J = NMF = Ê Ë kT ¯ kT dx

(2.135)

where N is the number of atoms per unit volume; M, the atomic mobility of the rate controlling species (= Db/kT); F, the electromigration induced force on the metal ions; Db, the grain boundary diffusion coefficient; and dm/dx the chemical potential gradient (force per atom) (where stress-dependent component of m = Ws, where W is the atomic volume = 1/N and s the uniaxial stress). The phenomenon of electromigration involves diffusion-controlled mass transport induced by current flow in a thin-film conductor, either through coulombic force applied on individual lattice ions or through momentum transfer between electrons and lattice ions (electron wind). The driving force on the ions due to applied electric field E (= rj, where r is the electrical resistivity and j the current density) is given by F = -z*qE = -qz*rj

(2.136)

where q is normal coulombic (ionic or electronic) charge, z* an empirical (dimensionless) parameter, qz* the effective metal ion charge, as negative in sign, and E the electric field strength. Thus the electromigration force (chemical potential gradient) is directly related to the current density. The atomic flux induced by current flow along GB i is obtained by substituting dm/dx with F so that Db ˆ Ji = N Ê qz * rj cos j i Ë kT ¯

(2.137)

where Db is the GB diffusivity and ji is the angle subtended by the direction of current and the inclination of grain boundary i. The net atomic flux, due to both stress- and current-induced mass transport along GB, is given by Db ˆ Ê Wds Ji = N Ê + qz * rj cos j i ˆ Ë kT ¯ Ë dx ¯

(2.138)

For low values of j, the corresponding driving force for electromigration is just balanced by the induced stress gradient so that the net atomic flux along the GB disappears. If we assume that l is the length of the thin-film conductor and ds/dx > 2s0/l (a critical value), stress at the GB extremities (adjacent grain boundary triple junctions) can no longer be supported. In that situation, matter is either absorbed or ejected, leading to the formation of either a hole or a hillock. Consequently, the threshold condition for the formation of either a hole or a hillock is given by lB j = 2s 0 qz * r cos j i

(2.139)

where lB is called the Blech length. When the threshold condition is exceeded (i.e., when l > lB), the effective current density j is diminished by 2s0/lBqz*r cos ji. It may be noted that stress- and current-induced degradations are interrelated and depend strongly on the corresponding microstructure of thin-film conductors.106 Figure 2.38 shows three common examples of electromigration in which atomic transport occurs by (a) grain boundary movement, (b) saddle displacement, and (c) flux divergence at a grain boundary triple junction.107 In the first case (Fig. 2.38a), grain boundary velocity Vgb (=Vb) is easily obtained from the condition for mass continuity between grains I and II and is given by

2.74

CHAPTER TWO

FIGURE 2.38 Three common examples of electromigration, in which atomic transport is manifested by (a) grain boundary movement, (b) saddle displacement, and (c) flux divergence at a grain boundary triple junction.107

Vb =

Db F kT

(2.140)

Thus, grain boundary migration arises from the net mass transport (net atomic flux across the boundary) attributed to electromigration. Although this effect is very small, it clearly illustrates the influence of a localized flux imbalance on a microscopically observable event (i.e., grain boundary migration). This specific mode is associated with the stability of bamboo structures in narrow conducting lines.108 In the second case (Fig. 2.38b), the velocity of a high conducting (thin-film) saddle, defined by length l (l >> d), width w, and thickness d, is located on a lowconductivity (thin-film) rail, so that current is diverted by the saddle. The resultant saddle velocity Vs is obtained from the condition for mass continuity and is given by Vs =

Ai Di F wd kT

(2.141)

where Di is the appropriate diffusivity for mass transport through cross-sectional area Ai. (Ideally, Vs is independent of saddle length.) If the principal mechanism for mass transport is controlled by volume diffusion of the film, Ai = wd and Eq. (2.141) converts to that for the atomic drift velocity (V = MF). However, cross-sectional areas for surface and grain boundary diffusion are approximately equal to 2aw and dwd/l, respectively, where a, d, and l are, respectively, nearest-atom distance, effective grain boundary thickness, and average grain size (l >> d). Thus, saddle velocity is a function of the principal mechanism for mass transport, thereby, of the characteristics of the rate-controlling species and saddle dimensions, because volume, surface, and grain boundary diffusions are the competing mechanisms for mass transport. This particular mode is related to the stability of multilevel interconnections in thin-film circuits. In the third case (Fig. 2.38c), mass transport at the triple junction can be analyzed in a similar manner. If we assume the transport to be limited to the grain boundary, net flux at a given triple junction is given by

DIFFUSION IN METALS AND ALLOYS

D Jb =

Nb F Nb F -Q ˆ Db = D0 expÊ Ë kT ¯ kT kT

2.75

(2.142)

where Nb = number of atoms per unit volume at the grain boundaries, Db is the GB diffusivity, D0 is the preexponential diffusivity (0.1 cm2◊s-1 for Al self-diffusion), and F is the electromigration-induced force on the metal ions.104 Equation (2.142) may also be written as D Jb =

Nb F (D1 cosq 1 + D2 cosq 2 - D3 cosq 3 ) kT

(2.143)

where Di and qi are, respectively, diffusivities of an angle formed by grain boundary i and the applied electric field (q3 = 0 in Fig. 2.38c). Usually a flux divergence occurs at individual triple junctions, leading to preferential formation of microscopic holes and hillocks to various degrees at GB triple points or where the GBs intersect the edge of the metal stripe. Unlike the first two examples, however, a flux divergence favors growth or decay of voids, rather than displacement of a macroscopic entity. This particular mode is linked to the stability of thin polycrystalline conducting films.107 Formation of holes and hillocks is a function of a gradient in temperature, current density, grain size, geometric features, crystal orientation of the grains, etc. The growth of these holes provides ultimately the interruption of the continuity of a conducting metallic stripe, which is damaged by high current densities and high temperature.101

2.11.2 Microelectronic Device and Electromigration In microelectronics, metallization, i.e., application of metals and metallike layers, plays an important role in controlling the device circuit performance. Basically metallization application can be divided into two groups, as illustrated in the schematic cross section of a completed metal-oxide semiconductor field-effect transistor (MOSFET). Metallization in contact with Si, such as on source and drain regions of Fig. 2.39, is termed contact metal whereas the one interconnecting various devices is termed the interconnect metal. Sometimes the metal on top of the gate of MOSFET can also act as the interconnect at that level. However, in a multilevel scheme of interconnects, the upper-level interconnects can be different from this gate-level interconnect.109 The effect of stress due to the ease of different processing temperatures of the metal and the dielectric on electromigration is of importance in fine multilayered aluminum interconnecting thin-film lines or stripes (typically, 1 mm or less in thickness and width) which are confined in quartz in VLSI or microelectronic devices.110 These thin-film lines contain a high density of defects such as grain boundaries, dislocations, and normal vacancies. Also, multilevel thin-film metallizations are employed to produce steep composition gradients.111 Electromigration in integrated circuits, especially VLSI and ultralarge-scale integration (ULSI) circuits, describes the development of structural damage due to metal ion transport in thin metal films under high current density (≥106 A/cm2). In integrated circuits, electromigration damage takes place either in the metallization or at the metal-semiconductor contacts (in GaAs MOSFET high-power electronic device). The former leads to an open-circuit localized mass depletion (holes) or short-circuit mass accumulation (hillocks), and the latter produces a poor perfor-

2.76

CHAPTER TWO

FIGURE 2.39 Schematic illustration of a cross section of (a) MOSFET with same contact and interconnect metal and (b) a contact with different contact and interconnection metals.109 (Reprinted by permission of Pergamon Press Plc., Oxford, after S. P. Murarka.)

mance or malfunction in semiconductor devices due to large power dissipation and high current density arising from the deterioration of ohmic and Shottky contacts.112 (In ohmic contact, the current-voltage characteristics are linear; they are usually formed by alloying or annealing several metallization structures on the semiconductor.113 High-power applications require low or ohmic barriers with low specific contact and parasitic resistance. Optimal Shottky barriers are the most demanding requirements of desirable metal-semiconductor interface properties.114) These accelerate degradation of requisite physical properties and reduce ultimate time to failure. Figure 2.40 shows examples of electromigration damage in thin films.

2.12 DIFFUSION ALONG SHORT CIRCUITS It is generally recognized that polycrystals and even single crystals contain regions such as dislocations, grain boundaries, interfaces, and free surfaces where the diffusivity is much higher than that through the lattice. These areas interact chemically

DIFFUSION IN METALS AND ALLOYS

2.77

(a)

(c)

(b)

FIGURE 2.40 Examples of electromigration damage in Al films. This leads to preferential formation of microscopic (a) hillock growth and holes,103 (b) whisker bridging two conductors,108 and (c) nearby mass accumulation (hillocks) and depletion (cracks).108 (Reprinted with permission from Academic Press, Inc., Boston, Massachussets.)

with the point defects, diffusing species, and the components of the alloy; the concentrations in the short circuits are quite different from those in the bulk. They can be modified by the diffusion process itself, which can produce changes in the ledge and kink densities on a surface, diffusion-induced migration of a grain boundary, etc. In the next section, we describe diffusion along dislocations, GB diffusion, diffusion-induced grain boundary migration (DIGM), and surface diffusion.

2.12.1 Dislocation Diffusion Several methods have been used to measure dislocation diffusion rates (with single crystals to avoid the grain boundary effect); most provide directly the product Dda2,

2.78

CHAPTER TWO

where a is the effective radius of dislocation “pipe” within which the mean effective diffusion coefficient is Dd. When segregation of the diffusant to the dislocation occurs, the product usually becomes Dda2s, where s is the segregation coefficient (or factor). To obtain the dislocation diffusion coefficient Dd, it requires the knowledge of a and, if relevant, of s. Because these quantities are commonly unavailable, it is common practice to report directly the Dda2 or Dda2s. (The value of a = 5 ¥ 10-10 m is sometimes assumed to calculate Dd.114a) Let us consider now the effect of randomly dispersed dislocations or a fine grain size which is of great interest in the study of kinetics of diffusion-enhanced process at or below ( –12 )TM and is vital in determining the diffusion rate in fine-grained thin films. A close similarity is observed between the results for diffusion enhanced by a three-dimensional array of dislocations and that resulting from a three-dimensional array of GBs. We first describe the findings for dislocations and then compare with those for boundaries. A proper mathematical description of penetration curves in dislocated crystals was provided by Le Claire and Rabinovitch, based on Smoluchowski’s model, which is shown in Fig. 2.41a. Let us consider the regular array of dislocation pipes of radius a and average separation 2Z, as shown schematically in Fig. 2.41b. They are perpendicular to the free surface, which has solute source on the top of it. The section can exhibit three different types of solute distributions based on the ratio of the mean diffusion distance in the lattice (Dlt)1/2 to the separation Z, and remembering that Dd is always much higher than Dl. A Kinetics. Case I: If L = (Dlt)1/2 >> R, an atom interacts with several dislocations in diffusing far enough in the lattice over the time t, and the effect of dislocations is to increase the jump frequency of the atoms in an isotropic manner. Thus, the dislocations enhance the effective diffusivity Deff for the solid, and the resulting penetration depth becomes larger than it would be without dislocations. The advancing isoconcentration lines are relatively flat near the dislocations. To obtain a relation between Deff and Dl, the concentration profiles obey generally the solution of Fick’s equation for homogeneous medium (Gaussian profile for a thinfilm source). However, we can measure an apparent (or effective) diffusion coefficient given by È Ê Dd¢ ˆ˘ Deff = Dl Í1 + rDpa 2 Á - 1˜ ˙ ÍÎ ¯ ˙˚ Ë Dl

(2.144)

where rD is the (line) dislocation density, Dd¢ the diffusion coefficient inside the dislocation pipe, and Dl the diffusion coefficient in sound crystal. Alternatively, Deff for a single crystal is given by, for no segregation of solute at dislocations, È Ê Dd ˆ ˘ Deff = fDd + Dl (1 - f ) = Dl Í1 + f Ë Dl ¯ ˙˚ Î

(2.145)

For diffusion of solute segregation at dislocations È Ê Dd ˆ ˘ Deff = fsDd + Dl (1 - fs) = Dl Í1 + fs Ë Dl ¯ ˙˚ Î

(2.146)

where f is the fraction sites on dislocation. Since Dd >> Dl, it follows that Dda 2 s =

Deff - Dl prD

(2.147)

DIFFUSION IN METALS AND ALLOYS

2.79

Surface 2R

2a D

D¢

(a)

Dt >> R

Type A

Type B

Dt

2R

2a

a
a, the situation is similar to GB diffusion in bicrystals; the lateral diffusion zones surrounding the dislocations are not influenced by neighboring dislocations.115 C Kinetics. If (Dlt)1/2 > average interboundary spacing l between the GBs, each diffusant atom diffuses along a large number of GBs as well as in the crystals in between or a multiple boundary diffusion zone. This condition, called Harrison’s A regime, is met for small-grained materials or very long diffusion anneal times and/or a volume

10–4

Titanium L = 69 mm

s/c

E

°

=

10 –

10

S –1

Climb

10–5 Coble

N-H

10–6 1.7

1.9

2.1

2.3

2.5

Tm / T (a)

Log (neck radius / particle radius)

0

Full density reached

–0.5

Volume diffusion from boundary

Boundary diffusion from boundary

–1.0

Surface diffusion from surface

–1.5

–2.0

0.5

0.6

0.7

0.8

0.9

1.0

T / Tm (b)

FIGURE 2.42 (a) Coble creep that occurs as a result of diffusional flow along grain boundaries for a-Ti of grain size 69 mm. A constant strain rate contour is exhibited by dashed lines.120 (Source: G. Malakonmdaiah and P. Rama Rao, Acta Metall., vol. 29, 1983, p. 1263.) (b) Sintering map for aggregate of Cu spheres of radius 88 mm. Mechanism corresponding to diffusional flow along GBs dominates in the shaded area of the map.121 (Source: M. F. Ashby, Acta Metall., vol. 22, 1974, p. 275.)

2.82

DIFFUSION IN METALS AND ALLOYS

2.83

X

f

Y

d (a)

y

Grain Boundary

Accumulation surface

o

x

Lattice

L

Source

l

(b)

FIGURE 2.43 (a) Typical isoconcentration contours resulting from a rapid diffusion along a grain boundary slab of thickness d.123 (b) Diffusion path geometry used for the accumulation method.123 (Source: J. C. M. Hwang and R. W. Balluffi, J. Appl. Phys., vol. 50, 1979, p. 1339.)

diffusion coefficient not much smaller than the grain boundary diffusion coefficient. A simple expression of the effective or true diffusion coefficient Deff can be written, taking into consideration the fraction f of the lattice sites which belong to the grain boundaries (or dislocations as the case may be).125 Deff = fDb + (1 - f )Dl

(2.148)

CHAPTER TWO

2.84

FIGURE 2.44

Harrison’s A-B-C classification.124

where f is the volume fraction of grain boundaries in the specimen and Dl is the lattice diffusion coefficient. Equation (2.148) is generally known as the HartMortlock equation. Le Claire has also given a rough calculation using this equation. Assuming an effective dislocation area of ~10-14 cm2 and a typical dislocation density of annealed metals of ~10-6 cm-2, we find that f = 10-8. The dislocation contribution to a measured or effective D will then exceed ~1% (about the limit detectable), when Deff/Dl > 106. From the activation energy Q (about 34TM for volume diffusion) and the rough generalization that Qd equals 0.5Qv, we estimate that Dd/Dl > 106 occurs for temperatures below about 0.5TM. That is why experiments where the lattice diffusion coefficient Dl only is of importance are always made at temperatures well above 0.5TM. At lower temperatures, Deff may be enhanced and may be

DIFFUSION IN METALS AND ALLOYS

2.85

seen as a slight upward curvature of the Arrhenius plot.13 For the diffusion of solute impurities that are bound or attracted to dislocations with a binding energy Eb, the Hart-Mortlock equation, Eq. (2.148), is written as D¢ = fDd exp

Eb + (1 - f )Dl kT

(2.149)

and the diffusion coefficients refer to impurities. The Hart equation can be shown to follow fairly accurately, if l/(Dlt)1/2 ≤ 0.3 and also if l/(Dlt)1/2 ≥ 100 (this represents the regime C kinetics as discussed below).126,127 Dislocations may start to contribute to the measured diffusion coefficient in these cases at higher temperatures than for self-diffusion.13 In reality, the same Eq. (2.148) with Db replacing Dd can be used for grain boundary-enhanced diffusion measurements in polycrystalline materials, provided grain size l is much less than L. According to the Lavine-MacCallum model, Hassner derived the following equation for the effective volume diffusion coefficient in the Type A kinetics regime: Deff = Dl + 4dDb/3d

(2.150)

where d is the grain size in polycrystalline materials. Other situations require more detailed considerations, but a rough working rule is that in well-annealed polycrystals, contributions from GB diffusion are usually negligible at temperatures above about 0.75TM. For very fine-grained material, the limit may be greater. kinetic regime c. Under these situations, diffusion may be assumed to occur only within the grain boundaries with negligible sidewise leakage into the neighboring crystals (Fig. 2.44c). Short diffusion anneal times and/or negligibly small values of the volume diffusion coefficient compared to the grain boundary diffusion coefficient cause volume diffusion lengths much shorter than the grain boundary width [(Dlt)1/2 0.75) several mechanisms have been reported. The first one is nonlocal surface diffusion of adsorbed atoms or complexes thereof. This mechanism describes a nonlinearity in the Arrhenius plot. The second mechanism depends on the orderdisorder transition (below Tm) at the surface which results in a nonexponential increase in the number of diffusible species. This process, also called surface melting, leads to high activation energies of surface diffusion at high temperatures and, therefore, very large surface diffusion coefficients (which have been experimentally observed).167a If mass transfer surface diffusion takes place by an adsorbed (adatom) or terrace vacancy mechanism, the macroscopic surface diffusivity Ds (for self-diffusion or diffusion of an impurity layer) can be expressed by168,169 Ds = and

1 n 2 Âa l G i 4 i =1

-Qd ˆ Ds = Ds0 expÊ Ë kT ¯

(2.154) (2.155)

From an atomic approach, Eq. (2.154) relates to an atomic jump distance ai between neighboring lattice sites, total number of jump types n, and a mean atomic jump frequency Gi. If the elementary diffusion process is viewed as the one where the atoms are activated over a free-energy barrier Qd from one potential well to another, then according to absolute rate theory, Eq. (2.154) is converted to the Arrhenius equation, Eq. (2.155), in which the preexponential term Ds0 = –14 ␯ sa2 exp (DSd/k). Here, ␯s is the atomic vibration frequency, and DSd and Qd may include terms related to the formation of a suitable diffusible atom and its motion.157a However, this equation holds only when (1) all the diffusion mechanisms must contribute independently to mass transport and (2) all the surface sites are equivalent. The defect should be in equilibrium all over the surface, and the concentration should be uniform everywhere, without any preferential trapping or occupancy sites. This requirement can be fulfilled by the close-packed perfect surfaces without ledges or kinks, e.g., a (111) surface in the fcc lattice.30 Adsorption of surface contaminants, such as oxygen, sulfur, and halogen and lowmelting-point metals, can enhance the value of Ds very greatly—by factors of up to 104 in some cases.169 This effect is of great technical significance in sintering. Surface free-energy mass transport can also be found in compounds (e.g., oxides), but it is more difficult to interpret this phenomenon if more than one chemical species is involved.

2.13 APPLICATION OF THIN FILMS TO DIFFUSION STUDY Atomic diffusion in thin films is of great technical importance in the processing of electronic and electrooptic devices as well as from the standpoint of their performance and reliability.170 Diffusion studies can be used to exploit the unique capability of control of structure and composition in thin films. This may consist of molecular beam epitaxial (MBE) growth of defect-free (dislocations and

DIFFUSION IN METALS AND ALLOYS

2.93

grain boundaries) single-crystal films and superlattices (SLs), ultrahigh-vacuum deposition of very pure thin films, the fine-tuning of composition in local regions by ion implantation, and the patterning (at submicrometer dimension) of sample by lithographic techniques employed in the electronic industry. In the following section we will only outline a few applications of thin films to diffusion study, such as irreversible processes, nonlinear diffusion, and diffusion in metastable phases.110 Irreversible Processes. This is the area where the effect of multiple driving forces (such as chemical affinity, stress, electric field, and temperature gradient) on diffusion in thin films and their interference and/or interactions can be studied. Examples include metallic thin-film deposition on inert substrate under stress due to the thermal mismatch between them; low-temperature reaction between Cu and Sn thin films leading to whisker and hillock growths in Sn films; and the effect of stress on electromigration in fine lines in microelectronic devices (see also Sec. 2.10.1). Nonlinear Diffusion. Effect of nonlinear diffusion study is of importance due to the trend of miniaturization in the electronic industry. Very high gradients of concentration, temperature, and electrical field are present in localized regions in the structure of devices during operation. It is now possible to grow superlattices of Si/SiGe and GaAs/AlGaAs by MBE to a large degree of perfection with respect to the control of periodicity and defect density. They could be employed to study the nonlinear effect of large concentration gradients on diffusion, using better samples and high-intensity synchrotron radiation sources for X-ray diffraction. Diffusion in Metastable Phases. Metastable phases such as amorphous alloys171 and icosahedral crystals172 can be prepared by thin-film deposition and preparation. Pseudoepitaxial growth can yield metastable crystalline phases. Self-diffusion and impurity diffusion in these metastable phases can be studied by using shallowprofiling techniques of thin films.173 Structure-kinetics correlation should be set up by performing a parallel study of these phases by electron microscopy and X-ray diffraction. The effect of atomic movement on phase stability is interesting.

2.14 DIFFUSION IN IONIC SOLIDS Extensive diffusion studies of ionic solids, involving the transport of electric charge, have contributed greatly to our understanding of defect solid-state physics. Various precise experimental techniques have emerged to probe the lattice defect structures, their transport, and defect interactions. In an ionic crystal, diffusion and electric (or ionic) conductivity s occur by the same defect mechanism. The two are related by the Nernst-Einstein equation D* =

sfkT NZ 2 e 2

or

Ds =

skT NZ 2 e 2

(2.156)

where D* is the diffusivity of the solute measured by radio tracer, Ds is the “charge” diffusion coefficient, f is the correlation factor (usually between 0.5 and 1.00) accounting for the nonrandom motion of tracer ions, k is the Boltzmann constant, and T, N, and e are the absolute temperature, number of mobile ions (of charge Ze) per unit volume, and electronic charges, respectively.174 A large number of ionic solids with exceptionally large ionic conductivity are called superionic conductors,

CHAPTER TWO

2.94

fast-ion conductors, and solid electrolytes. Among these, fast-ion conductor is the term most used.174 In the next section, diffusion in ionic crystals, namely, oxides, will be briefly discussed, because of their technological importance in the oxidation process and in superionic materials, as used in fuel cells, etc. It is also important in the control of ceramics and, thereby, their properties.176 2.14.1 Defects in Ionic Crystals Let us first consider stoichiometric crystals, say, the oxide MO, and “intrinsic” defect production. The Shottky defect (actually a pair of defects) contains a vacant anion site and a vacant cation site. It arises from thermal activation without undergoing interaction with the atmosphere. The Shottky defect production (assuming fully ionized defects) is expressed as a chemical reaction, i.e., 0 ∫ V≤M + VO..

(2.157)

where 0 refers to a perfect crystal, V≤M is a vacant (V) metal (M) site, the primes denote effective negative charges (with respect to perfect crystal), VO.. is a vacant (V) oxygen (O) site, and the over dots denote effective positive charges. This is a Kröger-Vink defect notation that is used as a shorthand for atomic point defects and electronic defects in compounds.3 We can express an equilibrium constant Ks for the reaction (also called the Shottky product) in the following form that is valid for low defect concentrations: Ê - DGsf ˆ .. K s = [VM¢¢ ][VO ] = expÁ ˜ Ë kT ¯

(2.158)

f

where the brackets [ ] denote concentrations and DG s is the Gibbs free energy of f formation of the Shottky defect, which can be split into its enthalpy DH s and entropy f DS s parts. The other type of defects observed in the stoichiometric ionic crystal is the Frenkel defects. The Frenkel defect (actually a pair of defects) consists of a cation interstitial and a cation vacancy, or an anion interstitial and an anion vacancy. The latter is called an anti-Frenkel defect, although presently this nomenclature is seldom used. Like the Shottky defect, the Frenkel defect is thermally activated. The Frenkel defect production is also expressed as a chemical reaction; for example, for cationic disorder ..

MM ∫ Mi + VM≤

(2.159) .. i

where MM is a metal atom (M) on a metal site (M), M is an effectively doubly positively charged metal ion on interstitial i, site and VM≤ is an effectively doubly negatively charged metal ion vacancy. We have assumed double charges here solely for illustrative purposes. The equilibrium constant for the cation Frenkel defect reaction KcF is given by Ê - DGcFf ˆ .. KcF = [VM¢¢ ][ Mi ] = expÁ ˜ Ë kT ¯ f

(2.160)

where DG cF is the free energy of formation of the cation Frenkel defect. Equation (2.160) is valid for low defect concentration. This equation is often termed the cation Frenkel product. Likewise for anion Frenkel defects, we have

DIFFUSION IN METALS AND ALLOYS

[VO.. ][O¢¢i ] = KaF = expÊÁË

f f f - DGaF Ê - DH aF Ê - DSaF ˆ ˆ ˆ ˜ = expÁ ˜ expÁ ˜ Ë kT ¯ Ë k ¯ kT ¯

2.95

(2.161)

where KaF is the equilibrium constant for the anion Frenkel defect reaction, O≤i .. represents doubly negatively charged anion interstitials, VO is a doubly positively f charged oxygen vacancy, DG aF is the free energy of formation of the anion Frenkel f f defect, and DH aF and DSaF are the corresponding enthalpy and entropy terms.25,177

2.14.2 Diffusion Theory in Ionic Crystals In most diffusion mechanisms, except the interstitial mechanism, an atom has to “wait” for a defect to arrive at a nearest-neighbor site prior to a possible jump. Thus the jump frequency contains a defect concentration term such as the vacancy concentration CV. Let us investigate an example case for diffusion comprising a Frenkel defect. Although both an interstitial and a vacancy are formed, in oxides one appears to be much more mobile, i.e., to possess a lower migration energy, than the other. In the case of stoichiometric UO2, for example, theoretical calculation of migration energies indicates a much smaller migration energy for the O2 vacancy than for the interstitial (by either interstitial or interstitialcy mechanisms).25,176 At the stoichiometric composition,

[ Mi.. ] = [VM¢¢ ] Ê DScFf ˆ Ê - DH cFf ˆ Ê - DGcFf ˆ where [ VM¢¢] = Cv = expÁ ˜ ˜ expÁ ˜ = expÁ Ë 2k ¯ Ë 2kT ¯ Ë 2kT ¯

(2.162) (2.163)

The measured activation enthalpy for diffusion consists of the migration enthalpy plus one-half the Frenkel defect formation enthalpy. Another interesting process is the intrinsic ionization process in which an electron is advanced from the valence band to the conduction band, leaving behind a hole in the valence band. When the conduction electron and hole are localized at atoms, it is normal to distinguish the process with a self-ionization reaction such as M2+ Æ M3+ + M+. In the Krœger-Vink notation, we usually express the intrinsic ionization process by . 0 ∫ e¢ + h (2.164) Usually all ionic crystals are capable of becoming nonstoichiometric. The limit of nonstoichiometry is primarily determined by the ease with which the metal ion can change its valence and the ability of the structure to “absorb” defects without undergoing reversion to some other structure and, thereby, changing phase. Nonstoichiometry can occur by either (1) an anion deficiency, which is accommodated by either anion vacancies such as UO2-x or metal interstitials such as Nb1+YO2 or (2) an anion excess, which is accommodated by either interstitials such as UO2+x or metal vacancies such as Mn1-YO.25 The extent of nonstoichiometry and the associated defect concentration are dependent on the temperature and partial pressure of the components. The defects generated in this manner are sometimes termed extrinsic; however, this nomenclature should be avoided because they are still intrinsic to the material. In an oxide, it is the usual practice to consider only the partial pressure of O2 at the temperatures of interest. In carbides, where the temperatures of diffusion are quite high, the

CHAPTER TWO

2.96

10–6

Fe-tracer self-diffusion coefficient, cm2/s

10–7

1300

1400°C

1200

10–8 1100 10–9 10–10

1000 900°C Fe3-y O4

10–11 –16 –14 –12

–10

–8

–6

–4

–2

–0

Log p02, atm FIGURE 2.49 The iron tracer self-diffusion coefficient in magnetite as a function of oxygen activity.16 (Source: After Dieckmann and Schmalzried, Ber/Bunsenges, Phys. Chem., vol. 81, 1977, p. 344.)

metal partial pressures can be comparable to the carbon partial pressure, and therefore either can be controlled. Another important example is that of diffusion in magnetite, Fe3O4, where several diffusion mechanisms are operative. Magnetite has the inverse spinel structure, with Fe2+ on the octahedral sites and one-half of the Fe3+ on the tetrahedral sites at room temperature. At elevated temperatures the distribution of these two cations is random in the two types of sites. There exist four octahedral and eight tetrahedral sites and three cations. At high oxygen activities there is a metaldeficient oxide with cation vacancies as the dominant defect. At low oxygen activities, a cation excess is present with iron ions on interstitial sites as the dominant defect. The reactions controlling the species in magnetite are 2+ 8FeFe + 2O2 = 8Fe3Fe+ + 3VFe + 4O0

3FeFe + 4O0 = 3Fei + 2O2

( high oxygen activity)

(low oxygen activity)

(2.165) (2.166)

In both reactions the stoichiometric ratio of lattice sites of the two species is conserved. Hence, the diffusivity should vary as the –23 power of the oxygen partial pressure at high oxygen activity and as the minus –23 power of the oxygen partial pressure at low oxygen activity. The data shown in Fig. 2.49 are consistent with these relations.16

2.15 DIFFUSION AND DIFFUSION-INDUCED DEFECTS IN SEMICONDUCTORS Diffusion and diffusion-induced defects are important in the manufacture of solid-state circuits in semiconductor materials, and an appreciation of their impor-

DIFFUSION IN METALS AND ALLOYS

2.97

tance is necessary to control the growth of single crystals and to regulate multilayer materials. Silicon and gallium arsenide are presently the most widely used semiconductors for fabricating microelectronic and optoelectronic devices.178,179 A short introduction to diffusion in silicon comprising point defects, phenomenological description of diffusion processes, diffusion mechanisms, and diffusion in GaAs and AlAs-GaAs materials will be given in this section.

2.15.1 Intrinsic Point Defects Equilibrium Condition. Intrinsic point defects are defects of atomic dimensions, free from any foreign atoms. The most basic and simple intrinsic point defects are vacancies (V) and self-interstitials (I) in Si. A Si crystal can achieve a thermodynamically more suitable state by introducing a specific concentration of intrinsic point defects under thermal equilibrium conditions. The thermal equilibrium concentration C xeq (in atomic fractions) of an electrically neutral intrinsic point defect x (x = V, I, . . .) is expressed in terms of Gibbs free energy of formation DG xf by Ê DGxf ˆ Cxeq = Z expÁ ˜ Ë kT ¯

(2.167)

where DG xf = DH xf + DS xf and Z is the dimensionless quantity characterizing the number of different configurations of the intrinsic point defects and is equal to 1 for vacancies. Nonequilibrium Condition. Surface reactions such as thermal oxidation, nitridation, and silicidation of Si may also lead to the injection of nonequilibrium intrinsic point defects. In contrast to the case of implantation, where both vacancies and interstitials are produced in supersaturation, these surface reactions normally produce only one type of point defect (such as self-interstitials by the surface oxidation process and vacancies by the surface nitridation process). In this case, the resulting perturbed self-interstitial concentration CI and vacancy concentration CV are expressed by the local equilibrium relationship CICV = C IeqC Veq

(2.168)

which holds good for prolonged times and elevated temperatures. These nonequilibrium point defects influence both dopant diffusion and nucleation, growth or shrinkage of dislocation loops, and are therefore of great significance for modern process simulation programs for silicon devices, according to Antoniadis, Fichtner, Kump, and Dutton.179 Intrinsic point defects may also occur in a charged form xr, where the superscript r denotes a positive or negative integer. In this case, the concentration of the charged intrinsic point defects is a function of the Fermi level, i.e., of the electron concentration n. Thus, a relation involving intrinsic electron concentration ni, electron concentration n, and the corresponding thermal equilibrium concentrations eq eq Cxr (ni) and Cxr (n), of charged intrinsic point defect and charged point defect xr, is given by Cxeqr ( n)

Cxeqr ( ni )

=

Ê nˆ Ë ni ¯

-r

(2.169)

2.98

CHAPTER TWO

We can use the hole concentration p instead of the electron concentration n, which is related to n by np = n i2

(2.170)

2.15.2 Diffusion in Silicon Silicon is the most important electronic material presently used. The dopant (impurity) diffusion technique is an important step in the fabrication of (doped) p-n- and hetero-junctions in semiconductor or microelectronic devices. These dopant atoms are usually substitutionally dissolved elements on Si lattice sites such as group III elements (B, Al, Ga) for p doping or group V elements (P, As, Sb) for n doping. Their (dopant) diffusion takes place mainly by processes involving lattice/ intrinsic point defects such as monovacancies and self-interstitial atoms.178 Hence, from the atomistic approach, diffusion involves the transport of atoms from one region of the Si crystal to another by the interaction of atoms with these point defects. Dopant diffusion from doped polysilicon or dopant-implanted silicides are used to avoid implantation-induced intrinsic point defects in the fabrication of ultrashallow junctions in silicon devices. The diffusion of various dopants such as Zn or Si results in greatly enhanced disordering of GaAs/AlAs and related III–V compound superlattices or multiquantum wells.180 2.15.2.1 Phenomenological Concepts. In the simplest case, Fick’s first law, the Arrhenius relationship, and Fick’s second law hold for diffusion in silicon. The diffusion coefficient of substitutionally dissolved elements can be shown to be primarily the product of the lattice vibration frequency G, the atomic jump distance a, the point defect fraction xpd, and the fraction of atoms that are activated to make a jump xm.181 1 D = a 2 xpd xm G 2

(2.171)

Thus, the activation energy for diffusion consists of the energy necessary to form the point defect and the migration energy of the diffusing atom. The concentration profile for a species diffusing with a constant diffusivity into Si from a surface source at a constant surface concentration Cs can be expressed by the traditional complementary error function, as shown in a normalized form in Fig. 2.50.179,182 In many cases, impurity diffusion in Si and other semiconductors can be best described by a concentration-dependent diffusivity of the form D = Ds

ÊCˆ Ë Cs ¯

g

(2.172)

where Cs is the concentration at the surface, Ds is the diffusivity at the surface, and g is the parameter describing the concentration dependence. Figure 2.50 is thus the resulting indiffusion normalized concentration profiles [in the log (C/Cs) versus x/ 4Dst form] corresponding to the actually occurring cases of g = 0, 1, 2, 3, and -2.179 For g = 0, the diffusion equation for constant diffusivity reduces to x ˘ C( x, t ) = C serf ÈÍ Î 4Dt ˙˚

(2.173)

For g > 0, the diffusivity decreases with decreasing concentration; g = 1 is observed for high-concentration diffusion of B and As in Si; g = 2 for high-concentration P

DIFFUSION IN METALS AND ALLOYS

FIGURE 2.50 Normalized concentration profiles for dependencies of the diffusion coefficient as indicated.179,182

different

2.99

concentration

diffusion in Si and for Zn diffusion in GaAs. For g = -2, the diffusivity increases with decreasing concentration which produces concave profile shape (in the semilogarithmic plot of Fig. 2.50). Such concave concentration profile shapes have been noted for Au, Pt, and Zn in Si and for several elements in Group III–V compounds such as Cr in GaAs.179 The concentration dependence of D can be determined from measured concentration profiles by Boltzmann-Matano analysis, as described in Sec. 2.10.1. 2.15.2.2 Diffusion Mechanisms. The dominant diffusion mechanisms in Si are vacancy diffusion, the direct interstitial diffusion mechanism, and the interstitialcy mechanism. In Si, self-diffusion via the interstitialcy mechanism predominates and occurs at higher temperature above about 1270 K. The interstitialcy mechanism also plays an important role in the diffusion of substitutional solutes such as P, B, Al, and Ge. Moreover, the ability to vary vacancy and self-interstitial concentrations by the addition of donors and acceptors can materially influence self-diffusivity. The open lattice of covalently bonded materials makes interstitial solute diffusion much more common.16 Interstitially dissolved foreign atoms can jump from one interstitial site to another. Intrinsic point defects do not play a role in this direct interstitial mechanism. However, the diffusion of substitutional foreign atoms such as p- and n-dopants and of host atoms (self-diffusion) requires intrinsic point defects. In the vacancy diffusion mechanism, the substitutional foreign atoms move by jumping into an adjacent vacant site. In the direct interstitial or interstitialcy diffusion mechanism, a selfinterstitial drives out a substitutional atom, which then returns to a substitutional lattice site at a neighboring position. Generally, diffusion by the direct interstitial mechanism, which occurs for elements such as Cu, Ni, or H, is very fast compared to the diffusion of substitutionally dissolved elements such as Group III and Group V dopants via the interstitialcy or vacancy mechanism, as shown in Fig. 2.24.

CHAPTER TWO

2.100

Silicon Self-diffusion. Uncorrelated self-diffusion and tracer self-diffusion coefficients of Si atoms, DSD and DT, respectively, in polycrystalline silicon are given by DSD = DICIeq + DVCVeq T

and

eq I

(2.174a) eq V

(2.174b)

D = fIDIC + fVDVC

where DI and DV are the diffusivities of self-interstitials and vacancies, respectively, and fI and fV are correlation factors for self-diffusion via self-interstitial (interstitialcy) and vacancy diffusion, respectively. The interstitial- and the vacancy-related parts have been measured predominantly by indirect methods comprising elements diffusing via interstitial-substitutional diffusion mechanisms, as shown in Fig. 2.51. However, it remains very difficult to determine the individual factors of these parts

1300

1100

T (°C) 900

700

800

10–16 DICIeq Self-diffusion in silicon

DICIeq or DVCVeq (m2.s–1)

10–18

10–20 DVCVeq 10–22

10–24

10–26

6

7

8 9 104/T (K–1)

10

FIGURE 2.51 Self-interstitial contribution DIC Ieq (full symbols) and vacancy contribution DVC Veq as a function of reciprocal absolute temperature.178 (Source: U. Gösele and T. Y. Tan, in Encyclopedia of Materials Science and Technology, vol. 4, Verlagsgesellschaft Chemie, Weinheim, Germany, 1991, pp. 197–247.)

DIFFUSION IN METALS AND ALLOYS

2.101

of the right side of Eq. (2.174) with good certainty, due to the dynamic interactions of vacancies and self-interstitials, which result in a single effective point defect diffusivity instead of two individual values of DI and DV, usually expected.178 Dopant Diffusion interstitial-related contribution. The diffusivity Ds of substitutionally dissolved atoms such as dopants comprising an interstitial-related part D Is and a vacancy-related part DVs is given by Ds = D Is + DVs

(2.175)

The normalized interstitial-related fractional diffusion part jI = D Is /Ds is close to unity for the small P and B atoms. The large-sized Sb atoms diffuse primarily by the vacancy mechanism, whereas the medium-size As atoms diffuse equally by both vacancy and self-interstitial mechanisms. The normalized jI part tends to decrease with decreasing temperature. For n-type dopants such as P, the jI part tends to decrease with increasing n-doping level. For perturbed intrinsic point defects such as during surface oxidation (CI > C Ieq) or surface nitridation (CV > C Veq), the pers turbed dopant diffusivity D per , is given by s = DIs Dper

Ê CI ˆ Ê CV ˆ + DVs Ë CIeq ¯ Ë CVeq ¯

(2.176)

In many instances, the local dynamic equilibrium of intrinsic point defects according to Eq. (2.170) is dominant and permits further simplification. Equation (2.176) may also be employed to simulate diffusion of dopants after ion implantation provided the nonequilibrium concentration of implanted-induced defects may be calculated. fermi level effect. The diffusivities Ds of all dopants in Si depend on the Fermi level. The experimentally observed dopant-dependent diffusivity, resulting from the interaction of dopants with charged and neutral point defects, can be described by the relation ni Ê nˆ Ê nˆ D s ( n) = D0s + D+s Ê ˆ + D-s + D=s Ë n¯ Ë ni ¯ Ë ni ¯

2

(2.177a)

which for intrinsic conditions n = ni converts to D s ( ni ) = D0s + D+s + D-s + D=s

(2.177b)

where the various subscripts denote the charge state of the intrinsic point defects forming a diffusing complex with the dopant atoms. Usually dopants tend to diffuse with neutral or oppositely charged intrinsic point defects. Table 2.7 lists the suitable diffusivities in the different charge states for B, P, As, and Sb dopants. dopant diffusion-induced nonequilibrium point defects. The nonequilibrium concentrations of supersaturated intrinsic point defects may also be induced by the in-diffusion of some dopants (e.g., P) starting from a high surface concentration. The highest supersaturation of self-interstitials is produced by highconcentration P in-diffusion. High-concentration B in-diffusion is accompanied by

CHAPTER TWO

2.102

TABLE 2.7 Diffusion of Various Dopants Fitted to Eq. (2.177a). Each term is fitted to D•0 exp(-Q/kT); D•0 values are 10-4 m2s-1 and Q values in electronvolts (1 eV = 1.6 ¥ 10-19 J).178 Element B P As Sb

D0•

Q0

D+•

Q+

0.037 3.85 0.066 0.214

3.46 3.66 3.44 3.65

0.72

3.46

D-•

Q-

D=•

Q=

4.44 12.0 15.0

4.00 4.05 4.08

44.20

4.37

Reprinted by permission of Pergamon Press Plc., Oxford.

a much smaller but still perceptible supersaturation. The self-interstitial supersaturation sI may be predicted from the equation s

sI =

CI - CIeq hf I D ( ns )CDs = (g + 1)DICIeq CIeq

(2.178)

where Ds and C Ds are the diffusivity and the surface concentration of the dopants, respectively. Here it is assumed that the dopant diffusivity obeys a simple linear or quadratic relationship with the electron concentration n, resulting in g = 1 or 2, respectively. For P, g = 2 has to be used.179 The supersaturated self-interstitials (SSI) can also condense into interstitial-type dislocation loops with or without a stacking fault or affecting oxygen or carbon agglomeration. diffusion-induced misfit dislocations. Both P and B (with smaller size than Si atoms) are dopants which can lead to the formation of an array of misfit dislocations formation provided they are present in very high concentrations. For most applications, it is preferred to avoid the formation of misfit dislocations.178 Interstitial-Substitutional Diffusion. As shown in Fig. 2.24, it seems that the much higher diffusivities of some elements are likely to be due to the presence of these solutes as interstitial atoms. A detailed discussion of these relative diffusivities is given in the article by Frank et al.38,39 Note that Au and Pt (Fig. 2.24), used in power devices to decrease the minority carrier lifetime in a controlled fashion, diffuse in Si via the interstitial-substitutional diffusion mechanism. At high temperatures (> ~800°C) the kick-out diffusion prevails in their diffusion behavior with a strongly concentration-dependent effective diffusion coefficient D(I) eff which is given by (I ) Deff =

DICIeq Ê C s ˆ Á ˜ C seq Ë CI ¯

2

(2.179)

yielding a fairly unusual concentration profile shown in Fig. 2.50 as the case of g = -2. In Eq. (2.179), Cs and C eq s are the actual and equilibrium concentrations (solubilities), respectively, of substitutional Au or Pt, and CI and C Ieq are the actual and equilibrium concentrations, respectively, of self-interstitials. At lower temperatures, the dissociative diffusion mechanism dominates with a concentrationindependent diffusivity D(V) eff

DIFFUSION IN METALS AND ALLOYS

(V ) Deff =

DVCVeq C seq

2.103

(2.180)

which represents a typical complementary error function profile (Fig. 2.50). In the event where the supersaturation of self-interstitials is canceled by a high concentration of internal sinks such as dislocations, for both the kick-out and the dissociative mechanisms, the in-diffusion of foreign interstitials prevails and leads to a concentration-independent diffusivity (I ) Deff =

eq DC i i eq Cs

(2.181)

where Di is the diffusivity and C eq i the solubility, of interstitial Au or Pt.

2.15.3 Diffusion in GaAs and AlAs-GaAs Materials Gallium arsenide is the most important compound semiconductor which has four principal advantages over silicon: (1) its higher electron mobility and saturated drift velocity, (2) ease of manufacture into semi-insulating substrate form, (3) the contribution of GaAs-based devices for much greater temperature tolerance and radiation hardness due to its larger band gap, and (4) use of GaAs as a direct band gap material. These properties make GaAs and related materials promising candidates to produce a wide range of high-frequency electronic and optoelectronic devices. To fabricate or even simply to operate the devices, self-diffusion and impurity diffusion of atoms are involved. Hence, a good understanding of both diffusion mechanisms and the behavior of point defect species controlling the diffusion processes is essential. For more details on this subject, see the article of Tan and others.183 Impurity diffusion profiles in GaAs usually exhibit complex, non-erfc function shapes. Laidig et al.184 have shown that the diffusion of the p-type dopant Zn into a GaAs-AlAs superlattice (SL) dramatically increased the Al-Ga interdiffusion. Later, it was found that n-type dopants, e.g., Si, also give rise to an improved Al-Ga interdiffusion; however, the effect is small compared to that due to Zn. Because of the technological significance of using dopant-enhanced SL disordering for lateral patterning of device structures, the doping and As4 pressure dependencies of Al-Ga interdiffusion have been studied in great detail.185,186 For the dominant native point defect species controlling the diffusion of self and impurity species on the Ga (such as gallium vacancy VGa and gallium-self interstitial species IGa) or group III sublattice, the following characteristics are very important:185,186 (1) Both VGa and IGa species are contributors; (2) the point defects, VGa and IGa, species are charged; (3) the point defects can occur in nonequilibrium concentrations; and (4) point defect thermal equilibrium concentrations are functions of the GaAs crystal composition, which is readily determined by the pressure of a vapor species (e.g., that of As4) in thermal equilibrium coexistence with the crystal.183

2.16 RADIATION EFFECTS AND DIFFUSION Bombardment with energetic particles such as electrons, protons, neutrons, and light and heavy ions produces changes in the physical properties of metals and alloys.

2.104

CHAPTER TWO

These property changes include formation of defects such as point defects, defect clusters, voids, and bubbles; void swelling; enhanced diffusion; induced segregation or precipitation; inverse Kirkendall effects; induced phase transformation; embrittlement; enhanced creep; surface modifications; sputtering at surfaces; changes in thermal and electrical conductivity; and impurity atom production.187–191 These physical property changes depend on the particular metal or alloy, its metallurgical history, the conditions of irradiations (namely, temperature and stress state), nature of irradiations, type of irradiation particles, particle energy or distribution of energies, instantaneous flux of particles, and the accumulated fluence of particles incident on the metal. Hence, an understanding of the various effects of irradiation is essential in several fields of materials science, namely, nuclear industry, surface treatment, semiconductors,188 superconductors,189 microelectronic devices, and so forth.30 This section briefly discusses only six areas of the physical property changes due to irradiation because of space limitations. However, readers should find the remainder elsewhere for further reading.

2.16.1 Types of Radiation Electron and proton irradiations produce primarily isolated point defects and are used to study basic properties and interactions of point defects. Ion bombardment is most useful for studies of microstructural changes and is used for radiation creep and fatigue studies on thin specimens. Light-ion irradiations produce some combination of isolated Frenkel pairs and defect cascades over longer ranges. Heavy-ion irradiation is usually carried out to study the properties of defect cascades over a shorter range. Neutron irradiations produced by fission of 235U have an average energy of 1.98 MeV and can exhibit defect cascade damage in metals and alloys. In a thermal reactor, fission neutrons become slowed or moderated to energies typical of reactor temperatures ( ~50 keV, defect cascades, instead of simply getting larger, split into distinct, well-separated damage regions or subcascades. These subcascades are closely spaced, and each exhibits typically a 25- to 30-keV primary recoil event. The frequent production of subcascades is typical of irradiations by fusion neutrons. Planar channeling of primary or secondary recoils also influences the subcascade formation when high-energy recoils discharge far from an earlier energetic collision.195,196 A small addition of impurities (50% in some grades of austenitic stainless steels.200–202 Figure 2.53 shows the schematic diagram relating swelling with dose dependence or fluence for different materials.203 Alloying elements or impurities may have considerable influence on swelling.202 The excellent swelling resistance of ferritic steels,

30 Solution treated austenitic steels

Standard cold-worked austenitic steels

Swelling, %

20 Commercial ferritic steels

Molybdenum (void lattice)

Nimonic PE16

10

Advanced cold-worked austenitic steels 0

0

50

100

150

Fluence, dpa FIGURE 2.53 Swelling versus fluence for different materials and conditions (schematic).203 (After B. L. Eyre and J. R. Matthews, J. Nucl. Mater., vol. 205, 1993, p. 1.)

DIFFUSION IN METALS AND ALLOYS

2.109

and of other bcc metals and alloys, is attributed to the trapping of interstitials near dislocations and their availability for recombination.204 It has been concluded, based on experimental results, that impurity trapping has greater effect on the vacancy supersaturation than the corresponding reactions occurring in fcc metals and alloys.205 The interstitials C and N and the substitutional Si atoms are particularly found to be effective. The C-vacancy binding enthalpy appears to be as high as 0.85 eV and would, therefore, serve as an effective recombination catalyzer even at high temperatures. Also, Cottrell atmosphere around dislocations may screen the sinks and reduce climbing rates. Transmutation-produced He affects both solute segregation and phase instability; the latter may exert strong effects on swelling.

2.16.4 Radiation-Enhanced Diffusion Radiation-enhanced diffusion (RED) is responsible for inducing phase transformations in a wide range of alloys under various irradiation conditions. The concept of RED results from the creation of excess numbers of point defects and defect clusters by irradiation and their subsequent free migration. RED is useful to (1) describe thermodynamic force-driven microstructural changes and (2) determine the characteristics of phase diagrams in the temperature range where diffusion is usually too slow to achieve equilibrium in an experimentally viable time. The MeV electron and ~100-keV proton irradiations are commonly used to remove energetic cascade effects and to provide a large fraction of freely migrating defects.206 When defects annihilate primarily at sinks, the RED coefficient becomes temperature-independent and is linearly proportional to the dose rate. This implies that the number of jumps a defect makes between generation and annihilation at a fixed sink is constant irrespective of how long the defect takes to make each jump. When defects annihilate mainly by direct recombination, the RED coefficient displays an Arrhenius temperature dependence, with an apparent activation enthalpy equal to one-half of the migration enthalpy of the slowest-moving defect, and is proportional to the square root of the defect-production rate.207 At intermediate temperatures, irradiation-induced point defects form much faster than they can anneal (or migrate) to dislocations, grain boundaries, and other point defect sinks. The resulting vacancy and interstitial concentration are often orders of magnitude greater than their equilibrium values. RED coefficients can be many orders of magnitude enhanced by this random migration process. At high temperatures, the additional defects introduced by irradiation become trivial and have no effect on diffusivity, because of the very rapid production of equilibrium thermal vacancies. The activation energy for diffusion is the sum of the energies of vacancy formation and migration. As stated earlier, at lower temperatures, Frenkel pairs annihilate by direct recombination, and D has an activation energy of one-half that for vacancy migration.30

2.16.5 Irradiation-Induced Segregation and Precipitation In general, an irradiation-induced flux of point defects to dislocations, grain boundaries, and voids usually produces a corresponding mass flux. In dilute alloys, the smaller and usually faster-diffusing atomic species is segregated toward defect sinks, whereas in concentrated alloys, the faster-diffusing species is segregated in the

2.110

CHAPTER TWO

opposite direction. Radiation-induced segregation (RIS) toward a sink appears to cause precipitation in single-phase alloys whereas segregation away from sinks has produced precipitate dissolution. RIS is a nonequilibrium process, which induces nonequilibrium concentration gradients rather than nonequilibrium phases.208 The large concentrations of vacancies and interstitials so created exchange differently with the atoms of different elements and produce redistribution of the different atomic species in the regions. Strong segregation of solute atoms toward point defect sinks and irradiation-induced instabilities can lead to oscillations in the composition of irradiated alloys even when it appears that very few sinks are available. RIS can be attributed ultimately to the formation of mobile defect-solute complexes and/or inverse Kirkendall effects. Both mechanisms couple a net flux of solute to the defect fluxes. The former mechanism is especially predominant in dilute alloys, for example, the formation of Zn precipitates in an irradiated Al–1.9% Zn alloy209 and occurrence of spinodal decomposition in an irradiated more complex Fe-Cr-Ni system.210,211 The extent of segregation produced by quenching and annealing is usually minor, because of the formation of a smaller amount of point defects (mostly vacancies) and transient fluxes. In contrast, the incessant creation and subsequent migration of vacancies and interstitials during irradiation at high temperatures produce continuous defect fluxes. Preferential coupling of individual solute elements to these persistent fluxes favors large localized compositional changes that undergo a net influx or outflow of defects. For example, austenitic grains of stainless steel transform completely into ferrite during neutron irradiation because of Ni segregation toward, and Cr segregation away from, grain boundaries. Also, during neutron irradiation, coatings of brittle silicide phases form on grain boundaries in many alloy systems due to strong RIS of Si or tramp impurities. Such large microstructural changes significantly influence the in-service performance of irradiated components. This defect-flux-driven segregation during irradiation has the greatest technological importance.207

2.16.6 Inverse Kirkendall Effects In the traditional Kirkendall diffusion mechanism, a gradient in composition can induce a net flux of defects (i.e., vacancies) across a diffusion couple with alloy components of unequal diffusion rates. In contrast, inverse Kirkendall segregation can occur in irradiated alloys where a vacancy (or interstitial) flux gives rise to a solute flux (i.e., a gradient in defect concentration can induce a solute concentration gradient). Both vacancy and interstitial defects can induce inverse Kirkendall effects. The inverse Kirkendall effect from a vacancy flux always results in the depletion of the faster-diffusing species at a sink, due to the atom flux associated with vacancies in the opposite direction to that of the defect flux. The inverse Kirkendall effect induced by an interstitial flux always causes enrichment of the faster-diffusing species at a sink, due to the same directions of both the interstitial and solute fluxes. However, the two inverse Kirkendall effects may assist or oppose each other in producing segregation at any specific sink.207 During the irradiation of austenitic stainless steels, the vacancy flow to grain boundaries leads to Ni enrichment and Cr and Fe depletion at grain boundaries. These composition changes are in agreement with fast diffusion of Cr and slow diffusion of Ni in austenitic stainless steels.211–213 Figure 2.54 is a schematic of the solute and defect concentration gradients near a vacancy sink.214 Heavy-ion and proton irradiation experiments have proved that the inverse Kirkendall segregation mechanism is quantitatively in agreement with several

DIFFUSION IN METALS AND ALLOYS

2.111

FIGURE 2.54 Inverse Kirkendall segregation: a schematic of concentration profiles for solute atoms and defects near irradiated grain boundaries. Fast-moving solutes move away from the boundary, and slow-moving solutes become segregated near the boundary.214 (Courtesy of E. P. Simonen.)

hundred measurements of grain boundary composition in irradiated austenitic alloys.214,215

2.16.7 Radiation-Induced Phase Transformation Irradiation of metals and alloys with energetic particles can introduce up to 108 J/mol of energy in the form of atomic displacements which is finally available to produce a range of phase and microstructural changes: These phenomena cannot be observed under thermal conditions.216 The production of metastable phases during irradiation is an active research area, especially dealing with the ion implantation.192 Many theoretical and experimental studies have been carried out on radiationinduced phase transformations.216 Russell has provided a brief review of phase stability under irradiation.217 Experimental results have been briefly outlined here for various ferrous and nonferrous alloy systems.216 Irradiation-induced amorphization is not discussed, but can be found elsewhere.216 2.16.7.1 Experimental Results Ferrous Alloys. Irradiation of martensitic stainless steel gives rise to the formation of Mo6C, Mo2C, c, Laves, M23C6, and a¢ phases. Irradiation of low-alloy ferritic steels gives severe embrittlement of the steel due to irradiation-induced defect aggregates such as dislocation loops, voids, various kinds of precipitates, and point defect-solute complexes. Cu and P used as tramp elements are known to increase greatly the embrittling effect of irradiation. Neutron irradiation of type 316 stainless steel provides g ¢ and G phases, which do not occur thermally; increases the formation of the h, Laves, and MC phases; and changes the composition of the Laves phases. The s and Fe2P phases may also be

2.112

CHAPTER TWO

influenced by irradiation. The temperatures of appearance and abundances of these phases are very sensitive to minor changes in alloy composition. Irradiation of types 316, 316L, 321, and 347 stainless steels has caused the formation of two kinds of magnetic phases. Some irradiations produce blocky ferrite particles, whereas others give a very high concentration of fine supermagnetic particles which are believed to be Ni- and Fe-rich. Neutron, heavy-ion, and electron irradiations induce copious NbC precipitation in Nb-containing stainless steels. Irradiation of Fe-Ni or Fe-Ni-Cr Invar-type alloys provides a spinodallike decomposition of the matrix into Fe-rich and Ni-rich regions. Nonferrous Alloy Systems. Irradiation of Al-Cu solid solutions appears to favor the formation of q¢ phase and depress the formation of q≤. Irradiation leads to a large increase in precipitation nucleation rates in Al-Ge and Al-Si alloys. Electron irradiation raises the solvus temperature in Al-rich Al-Zn alloys by tens of degrees, based on the displacement rate. Strong segregation effects take place in Cu-base alloys for Fe, Be, and Ag solutes. Irradiation usually augments GP zone formation and precipitation in Cu-Be alloys and leads to coherency loss in Cu-Co and Cu-Fe alloys. It also results in a decrease in resistivity of Cu-Ni alloys, which can be attributed to the decomposition of solid solution, perhaps by a spinodal-type transformation. Strong solute segregation of Al, Be, Cr, Cu, Mo, Mn, Si, and Ti has been observed in irradiated Ni-base alloys. Precipitation has been noticed in irradiated, thermally single-phase Ni-Be and Ni-Ge alloys. A displacement rate-dependent threshold for precipitation is found in thermally single-phase Ni-Si alloys. Irradiation of Inconel 706 and 718 provides complex interplay between g ¢, g ≤, and h phases. Irradiation has been found to induce disorder in several Ni-base alloys. An interesting finding is the disordering of Ni3Si and its replacement by Ni5Si2, which is not thermally stable, but appears to be more irradiation-resistant than Ni3Si. Irradiation induces strong segregation of Al, Mo, and V in Ti-base alloys. Segregation of Al in irradiated Ti-Al alloys produces redistribution of a2 precipitates and formation of b phase. Neutron irradiation of thermally single-phase W-rich W-Rh alloy causes the formation of c, WRe3 phase rather than the equiatomic s phase. The effect of irradiation in w-phase precipitation in Zr-Nb alloys is ambiguous, in some cases favoring w-phase formation and in other cases exerting no effect. Neutron irradiation greatly increases the rate of white-to-gray (tetragonal-tobcc) tin transformation, perhaps by the vacancy-rich cores of displacement cascades serving as nucleation sites for the less dense gray phase. High-fluence neutron irradiation of U-Mo and U-Nb alloys produces a stable aU and causes g ¢ phases to revert to the metastable high-temperature g phase. The reversion probably takes place by irradiation disordering of the g ¢ phase to yield the g phase.

REFERENCES 1. D. Lazarus, in Materials Science Forum, vol. 1, 1984, Diffusion in Solids, guest eds. A. L. Laskar, G. P. Tiwari, E. C. Subba Rao, and R. Krishnan, Trans Tech Publications, Switzerland, 1984. 2. J. P. Stark, in Treatise of Materials Science and Technology, vol. 4, Academic, New York, 1974, pp. 59–111; Solid State Diffusion, Wiley, New York, 1976.

DIFFUSION IN METALS AND ALLOYS

2.113

3. J. Philiburt, Atom Movements: Diffusion and Mass Transport in Solids (English translation by S. J. Rothman), Les Editions de Physique, Paris, 1991. 4. E. A. Brandes and G. B. Brook, eds., Smithells Metals Reference Book, 7th ed., Butterworth-Heinemann, Oxford, 1992, pp. 13.1–13.119. 5. J. Karger and D. M. Ruthven, Diffusion in Zeolites and Other Microporous Solids, Wiley, New York, 1992. 6. J. F. Shakelford and W. Alexander, CRC Materials Science and Engineering Handbook, CRC Press, Boca Raton, Florida, 1994, pp. 236–247. 7. C. P. Flynn, Point Defects and Diffusion, Clarendon Press, Oxford, 1972. 8. G. E. Murch, in Materials Sc. Technolog., vol. 5, Phase Transformations in Metals and Alloys, chap. 2, VCH, Weinheim, 1991, pp. 75–142. 9. R. P. Smith, Acta Met., vol. 1, 1953, p. 578; W. F. Smith, Principles of Materials Science & Engineering, McGraw-Hill, New York, 1986. 10. P. G. Shewmon, Diffusion in Solids, The Metallurgical Society, Warrendale, Pa., 1989. 11. G. H. Geiger and D. R. Poirier, Transport Phenomena in Metallurgy, Addison-Wesley, Reading, Mass., 1973. 12. H. Mehrer, in Landolt-Börnstein, vol. 26, Diffusion in Solid Metals and Alloys, ed. H. Mehrer, Springer-Verlag, Berlin, 1990, pp. 1–31. 13. A. D. LeClaire, in Encyclopedia of Materials Science and Engineering, Pergamon, Oxford, 1986, pp. 1186–1193. 14. R. A. Porter and K. E. Easterling, Phase Transformations in Metals and Alloys, Van Nostrand Reinhold (UK) Ltd., 1981. 15. M. Bishop and K. E. Fletcher, Int’l. Met. Reviews, vol. 17, 1972, pp. 203–225. 16. E. S. Machlin, Thermodynamics and Kinetics/Materials Science, Giro Press, New York, 1991. 17. Decarburization, ISI Publication133, Gresham Press, Old Working Surrey, England, 1970. 18. G. E. Totten and M. A. H. Howes, eds., Steel Heat Treatment Handbook, Marcel Dekker, New York, 1996. 19. M. C. Richman, An Introduction to Science of Metals, Blaisdell, Waltham, Mass., 1967. 20. S. J. Rothman, Diffusion in Crystalline Solids, eds. G. Murch and A. S. Nowick, Academic Press, Orlando, Fla., 1984, pp. 1–61. 21. H. Zabel, in Nontraditional Methods in Diffusion, eds. G. E. Murch, H. K. Birnbaum, and J. R. Cost, The Metallurgical Society, Warrendale, Pa., 1984, pp. 1–37. 22. W. Petry, A. Heiming, C. Herzig, and J. Trampenau, Defect Diffusion Forum, vol. 75, 1991, pp. 211–228. 23. H. T. Stokes, in Nontraditional Methods in Diffusion, eds. G. E. Murch, H. K. Birnbaum, and J. R. Cost, The Metallurgical Society, Warrendale, Pa., 1984, pp. 39–58. 24. H. T. Stokes, in Conf. Proceed. AIME-TMS, 1984, pp. 39–58. 25. G. E. Murch, in The Encyclopedia of Advanced Materials, Pergamon Press, Oxford, 1994, pp. 125–130; in Phase Transformations in Materials, chap. 2, vol. ed. P. Haasen, VCH, Weinheim, Germany, 1991, pp. 74–141. 26. G. K. Wertheim, Mossbauer Effect: Principles and Applications, Academic Press, New York, 1964. 27. J. C. Mullen, in Nontraditional Methods in Diffusion, eds. G. E. Murch, H. K. Birnbaum, and J. R. Cost, The Metallurgical Society, Warrendale, Pa., 1984, pp. 59–81. 28. A. S. Nowick, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 1180–1186. 29. J. L. Snoek, Physica, vol. 6, 1939, p. 591.

2.114

CHAPTER TWO

30. J. L. Bocquet, G. Brebec, and Y. Limoge, in Physical Metallurgy, chap. 7, vol. 1, 4th ed., eds. R. W. Cahn and P. Haasen, Elsevier Science BV, Amsterdam, 1996, pp. 535–668. 31. C. Zener, Trans. AIME, vol. 152, 1943, p. 122; Phys. Rev., vol. 71, 1947, p. 34. 32. J. Volkl, Ber. Bunsen Gesell, vol. 76, 1972, p. 797. 33. L. Neel, J. Phys. Rad., vol. 12, 1951, p. 339; vol. 13, 1952, p. 249; vol. 14, 1954, p. 225. 34. S. Taniguchi, Sci. Rep. Res. Inst. Tohoka Univ., vol. A7, 1955, p. 269. 35. W. A. Lanford, R. Benenson, C. Burman, and L. Weilunski, in Nontraditional Methods in Diffusion, eds. G. E. Murch, H. K. Birnbaum, and J. R. Cost,The Metallurgical Society, Warrendale, Pa., 1984, pp. 1–38. 36. W. T. Petuskey, in Nontraditional Methods in Diffusion, eds. G. E. Murch, H. K. Birnbaum, and J. R. Cost, The Metallurgical Society, Warrendale, Pa., 1984, pp. 179–202. 37. J. A. Kilner, Mater. Sci. Forum, vol. 7, 1986, pp. 205–222. 38. W. Frank, Defect and Diffusion Forum, vol. 75, 1991, pp. 121–148. 39. W. Frank, U. Gösele, H. Mehrer, and A. Seeger, Diffusion in Crystalline Solids, eds. G. Murch and A. S. Nowick, Academic Press, Orlando, Fla., 1984, pp. 63–142. 40. W. Schilling, J. Nucl. Mater., vol. 69–70, 1978, p. 65. 41. A. Seeger, in Proc. Conf. Fundamental Aspects of Radiation Damage in Metals, Gathinburg, eds. M. T. Robinson and F. W. Young, Jr., ERDA Report Conf. 751006, P1 and P2, U.S. Energy and Research and Development Association, Washington, D.C. 42. W. Schule, Defects and Diffusion Forum, vols. 66–69, 1989, pp. 313–326. 43. P. Haasen, Physical Metallurgy, 3d ed., Cambridge University Press, London, 1993. 44. G. D. Watkins and J. W. Corbett, Phys. Reviews, vol. A134, 1964, p. 1359. 45. G. D. Watkins, J. W. Corbett, and R. M. Walker, J. Appl. Phys., vol. 30, 1959, p. 1198. 46. P. H. Dederichs, C. Lehmann, H. R. Schober, A. Scholtz, and R. Zeller, J. Nucl. Mater., vols. 69 and 70, 1978, p. 176. 47. U. Gösele, W. Frank, and A. Seeger, Appl. Phys., vol. 23, 1980, p. 361. 48. F. C. Frank and D. Turnbull, Phys. Rev., vol. 104, 1956, p. 617. 49. U. Gösele and F. Morehand, J. Appl. Phys., vol. 52, 1981, p. 4617. 50. A. H. Van Ommen, J. Appl. Phys., vol. 54, 1983, p. 5055. 51. V. M. Vorob’ev, V. A. Murav’ev, and V. A. Panteleev, Sov. Phy. Solid State, vol. 23, 1981, p. 2055. 52. D. Weiler and H. Mehrer, Phil. Mag., vol. A49, 1983, p. 309. 53. N. V. Doan and Y. Adda, Phil. Mag., vol. A56, 1987, p. 269. 54. H. Bakker et al., in Diffusion in Solids: Recent Developments, eds. M. A. Dayananda and G. E. Murch, TMS-AIME, Warrendale, Pa., 1984, pp. 39–65. 55. D. N. Seidman, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 3591–3596. 56. D. N. Seidman, J. Phys., vol. F3, 1973, pp. 393–421. 57. H. J. Wollenberger, in chap. 18, Physical Metallurgy, vol. 2, 4th ed., eds. R. W. Cahn and P. Haasen, Elsevier Science BV, Amsterdam, 1996, pp. 1621–1721. 58. R. W. Balluffi, K. H. Lie, D. N. Seidman, and R. W. Siegel, Vacancies and Interstitials in Metals, eds. A Seeger et al., North-Holland Publishing Co., Amsterdam, 1970, pp. 125–166. 59. H. Bakker et al., in Diffusion in Solids: Recent Developments, TMS-AIME, Warrendale, Pa., 1984, pp. 39–65. 60. C. Herzig and U. Kohler, Mater. Sci. Forum, vols. 15–18, 1987, p. 301. 61. A. Seeger, J. Phys., vol. F3, 1973, p. 248.

DIFFUSION IN METALS AND ALLOYS

62. 63. 64. 65. 66. 67. 68. 69. 69a. 70. 71. 72. 73. 74. 75. 76.

77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93.

2.115

R. W. Siegel, Annu. Rev. Mat. Sci., vol. 10, 1980, pp. 393–425. G. Neumann, in Defects and Diffusion Forum, vols. 66–69, 1989, pp. 43–64. R. B. McLellan and Y. C. Argel, Scripta Metall., vol. 23, no. 67, 1995, pp. 945–948. M. W. Finnis and M. Sachdev, J. Phys., vol. 6, 1976, p. 965. F. J. Kedves and G. Edelyi, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 175–188. A. D. Le Claire, Phil. Mag., vol. 14, 1966, p. 271. I. Kaur, Y. Mishin, and W. Gust, Fundamentals of Grain and Interphase Boundary Diffusion, 3d ed., Wiley, Chichester, 1995. J. R. Cahoon and O. D. Sherby, Met. Trans., vol. 23A, 1992, pp. 2491–2500. A. D. Le Claire and G. Neumann, in Landolt-Börnstein, vol. 26, Diffusion in Solid Metals and Alloys, ed. H. Mehrer, Springer-Verlag, Berlin, 1990, pp. 85–212. J. N. Mundy, Defects and Diffusion Forum, vol. 83, 1992, pp. 1–18. N. L. Peterson, J. Nucl. Mater., vol. 69–70, vol. 3, 1978; Proc. Int. Conf. on the Properties of Atomic Defects in Metals, Argonne, eds. N. L. Peterson and R. W. Siegel, 1976. A. D. Le Claire, in Diffusion in Body Centered Cubic Metals, eds. J. A. Wheeler, Jr. and F. R. Winslow, ASM, Metals Park, Ohio, 1965, pp. 3–25. J. N. Mundy, T. E. Miller, and R. J. Porte, Phys. Rev., vol. B3, 1971, p. 2445. L. Slifkin, Met. Trans., vol. 20A, 1989, pp. 2577–2582. J. M. Sanchez and D. De Fontaine, Phys. Rev. Lett., vol. 35, 1975, p. 227. A. Seeger and H. Mehrer, in Vacancies and Interstitials in Metals, eds. A. Seeger, D. Schumacher, W. Schilling, and J. Diehl, North-Holland Publishing Co., Amsterdam, 1970, p. 1. A. Da Fano and G. Jacucci, Phys. Rev. Lett., vol. 39, 1977, p. 950. G. M. Hood, Proc. Int. Conf. on Diffusion in Materials, 1992, Kyoto, Japan, 1993, p. 755. W. Petry, A. Heiming, J. Trampenau, and G. Vogl, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 157–174. A. D. LeClaire, J. Nucl. Mater., vols. 69–70, 1978, p. 70. A. R. Allnatt and A. B. Lidiard, Atomic Transport in Solids, Cambridge University Press, London, 1993. N. L. Peterson, in Diffusion in Solids: Recent Developments, eds. A. S. Nowick and J. J. Burton, Academic Press, New York, 1975, pp. 115–170. W. K. Warburton and D. Turnbull, in Diffusion in Solids, eds. A. S. Nowick and A. S. Burton, Academic Press, New York, 1975, pp. 171–229. H. Nakajima, O. Ohshida, K. Nonaka, Y. Yoshida, and E. E. Fujita, Scripta Metall., vol. 34, no. 6, 1996, pp. 949–953. H. Araki et al., Met. Trans., vol. 27A, 1996, pp. 1807–1814. R. J. Borg and G. I. Dienes, An Introduction to Solid State Diffusion, Academic Press, New York, 1988. R. K. Kapoor and T. W. Eager, Met. Trans., vol. 21A, 1990, pp. 3039–3047. A. Smigelskas and E. Kirkendall, Trans. AIME, vol. 171, 1947, p. 130. T. C. Chou and L. Link, Scripta Metall., vol. 34, no. 5, 1996, pp. 831–838. M. A. Krivoglaz, Theory of X-ray and Thermal Neutron Scattering by Real Crystals, Plenum, New York, 1969. R. W. Balluffi, Acta Metall., vol. 8, 1960, p. 871. J. Crank, Mathematics of Diffusion, 2d ed., Oxford University Press, London, 1975. M. C. Petri and M. A. Dayananda, Philosophical Magazine A, vol. 76, no. 6, 1997, pp. 1169–1185; Defect & Diffusion Forum, vols. 143–147, 1997, pp. 575–579.

2.116

CHAPTER TWO

94. M. A. Dayananda, in Defects and Diffusion Forum, vol. 83, 1992, pp. 73–86. 95. D. J. Young, R. C. Doward, and J. C. Kirkaldy, in Fundamentals and Applications of Ternary Diffusion, 29th Ann. Conf. of Metallurgists of CIM, ed. C. Purdy, Pergamon, New York, 1990, pp. 3–14. 96. M. A. Dayananda, in Diffusion in Metals and Alloys, Landolt-Börnstein, SpringerVerlag, Berlin, 1991. 97. M. A. Dayananda, Met. Trans., vol. 27A, 1996, pp. 2504–2509. 98. M. A. Dayananda, Trans., TMS-AIME, vol. 242, 1968, pp. 1369–1372. 99. F. J. J. Van Loo, G. F. Bastin, and J. C. Vrolijk, Metal Trans., vol. 18A, 1987, pp. 801–809. 100. J. S. Kirkaldy and D. J. Young, on Diffusion in Condensed State, The Institute of Metals Society, London, 1987, pp. 172–272. 100a. M. A. Dayananda, in Landolt-Börnstein, vol. 26, Diffusion in Solid Metals and Alloys, ed. H. Mehrer, Springer-Verlag, Berlin, 1990, pp. 372–436. 101. J. W. Morris et al., JOM, May 1996, pp. 43–46. 102. O. Kraft and E. Artzt, Acta Mater., vol. 46, 1998, pp. 3733–3743. 103. M. Ohring, The Materials Science of Thin Films, Academic Press, Boston, 1992. 104. R. E. Hummel, International Materials Reviews, vol. 39, no. 3, 1994, pp. 97–111. 105. S. J. Krumbein, in Electromigration and Electronic Device Degradation, ed. A. Christou, Wiley, New York, 1994, pp. 139–166. 106. C. L. Bauer, in The Encyclopedia of Advanced Materials, vol. 1, Pergamon, Oxford, 1994, pp. 711–717. 107. C. L. Bauer and P. F. Tang, in Defect and Diffusion Forum, vols. 66–69, 1989, pp. 1143–1152. 108. S. Vaidya, T. T. Sheng, and A. K. Sinha, Appl. Phys. Lett., vol. 36, no. 6, 1980, p. 464. 109. S. P. Murarka, in The Encyclopedia of Advanced Materials, vol. 4, Pergamon, Oxford, 1994, pp. 2912–2919. 110. K. N. Tu, Defect and Diffusion Forum, vol. 83, 1992, pp. 141–143. 111. D. Gupta, in Defect and Diffusion Forum, vol. 59, 1988, pp. 137–150. 112. J. H. Zhao, in Electromigration and Electronic Device Degradation, ed. A. Christou, Wiley, New York, 1994, pp. 167–233. 113. C. J. Palmstrom, in The Encyclopedia of Advanced Materials, vol. 1, Pergamon Press, Oxford, 1994, pp. 473–478. 114. L. J. Brillson, The Encyclopedia of Advanced Materials, vol. 1, Pergamon Press, Oxford, 1994, pp. 478–486. 114a. A. D. Le Claire, in Landolt-Börnstein, vol. 26, Diffusion in Solid Metals and Alloys, ed. H. Mehrer, Springer-Verlag, Berlin, 1990, pp. 626–629. 115. H. Mehrer and M. Lubbehusen, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 591–604. 116. J. Bernardini, S. Bennis, and G. Moya, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 801–810. 117. K. Viergge and C. Herzig, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 811–818. 118. I. Kaur and W. Gust, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 765–786. 119. N. L. Peterson, Int’l. Metals Review, vol. 29, no. 9, 1983, p. 65. 120. G. Melakondaiah and P. Rama Rao, Acta Metall., vol. 29, 1983, p. 1263. 121. M. F. Ashby, Acta Metall., vol. 22, 1974, p. 275. 122. J. Singh, S. Lele, and S. Ranganathan, Zeitschrift für Metallkunde, vol. 72, 1981, p. 469.

DIFFUSION IN METALS AND ALLOYS

2.117

123. M. Mohan Rao and S. Ranganathan, Materials Science Forum, vol. 3, 1984, pp. 43–58; Trans Tech Publications, 1984, pp. 43–57. 124. L. G. Harrison, Trans. Faraday Soc., vol. 57, 1961, p. 1191. 125. E. W. Hart, Acta Metall., vol. 5, 1957, p. 597. 126. G. E. Murch and S. J. Rothmann, Diffusion Defect Data, vol. 42, 1985, p. 17. 127. D. Gupta, D. R. Campbell, and P. S. Ho, Thin Films: Interdiffusion and Reactions, eds. J. M. Poate, K. N. Tu, and J. W. Meyer, Wiley, New York, 1978, p. 161. 128. T. Suzuoka, Trans. Jpn. Inst. Met., vol. 2, 1961, p. 25. 129. A. D. Le Claire, Br. J. Appl. Phys., vol. 14, 1963, p. 351. 130. D. Gupta and J. Oberschmidt, Diffusion in Solids: Recent Developments, eds. M. A. Dayananda and G. E. Murch, TMS-AIME, Warrendale, Pa., 1984, p. 121. 131. A. D. Le Claire and A. Rabinovitch, in Diffusion in Crystalline Solids, eds. G. E. Murch and A. S. Nowick, Academic, New York, 1984, pp. 259–319. 132. N. L. Peterson, Mater. Sci. Forum, vol. 1, 1984, p. 85. 133. R. W. Balluffi, in Diffusion in Crystalline Solids, eds. G. E. Murch and A. S. Nowick, Academic, New York, 1984, pp. 320–378. 134. J. Philibert, Diffusion et Transport de Matiere dans les Dolides, Les Editions de Physique, Paris, 1985. 135. Y. Mishin, Chr. Herzig, J. Bernardini, and W. Gust, Inter. Mater. Reviews, vol. 42, no. 4, 1997, pp. 155–178. 136. R. A. Fournelle, Materials Science and Engineering, vol. A138, 1991, p. 133. 137. F. J. A. den Broeder, Acta Metall., vol. 20, 1972, p. 319. 138. A. H. King, Int’l. Mater. Review, vol. 32, 1987, p. 173. 139. R. D. Doherty, The Encyclopedia of Advanced Materials, Pergamon Press, Oxford, 1994, pp. 1695–1698. 140. P. G. Shewmon and G. Meyrick, in Diffusion in Solids: Recent Developments, eds. M. A. Dayananda and G. E. Murch, The Metallurgical Society, Warrendale, Pa., 1985. 141. P. Shewmon, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 727–734. 142. R. W. Balluffi and J. W. Cahn, Acta Metall., vol. 29, 1981, p. 493. 143. P. G. Shewmon, M. Abbas, and G. Meyrick, Met. Trans., vol. 17A, 1986, pp. 1523–1527. 144. M. Hillert and G. R. Purdy, Acta Metall., vol. 26, 1978, p. 338. 145. D. B. Butrymowicz, J. W. Cahn, J. R. Manning, and D. E. Newbury, in Advances in Ceramics, vol. 6, Grain Boundaries and Interfaces in Ceramics, American Ceramic Society, Westerville, Ohio, 1983, p. 202. 146. R. S. Ahy and B. Evans, Acta Metall., vol. 35, 1987, p. 2049. 147. L. Liang and A. H. King, Acta Metall., vol. 44, no. 7, 1996, pp. 2893–2988. 148. J. Sommer, Y. M. Chiang, and R. W. Balluffi, Scripta Metall., vol. 33, no. 1, 1995, pp. 7–12. 149. M. Hillert and R. Lagenborg, J. Mater. Sci., vol. 6, 1971, p. 208. 150. I. G. Solorzano, G. R. Purdy, and G. C. Weatherly, Acta Metall., vol. 32, 1984, p. 1709. 151. C. Y. Ma, W. Gust, R. A. Fournelle, and B. Predel, Mater. Sc. Forum, vol. 3/7, 1993, pp. 126–128. 152. Y.-J. Baik and D. N. Yoon, Acta Metall., vol. 38, 1990, p. 1525. 153. W. H. Rhee and D. K. Yoon, Acta Metall., vol. 37, 1989, pp. 221–228. 154. T. Tashiro and G. R. Purdy, Scripta Metall., vol. 21, 1987, pp. 361–364. 155. D. A. Smith and A. H. King, Phil. Mag., vol. A44, 1981, pp. 330–340.

2.118

CHAPTER TWO

156. G. R. Purdy, in Phase Transformations in Materials, vol. 5, vol. ed. P. Haasen, VCH Publishers, Weinheim, Germany, 1991. 157. C. Li and M. Hillert, Acta Metall., vol. 29, 1981, pp. 1949–1960. 158. C. A. Handwerker, in Diffusion Phenomena in Thin Films and Microelectronic Materials, Noyes, Park Ridge, N.J., pp. 245–322. 159. D. N. Yoon, Annu. Rev. Mater. Sci., vol. 19, 1989, pp. 43–58. 160. D. N. Yoon and W. J. Huppmann, Acta Metall., vol. 27, 1979, pp. 973–979. 161. M. K. Burton, N. Cabrera, and F. C. Frank, Phil. Trans. Roy. Soc., vol. A243, 1951, p. 299. 162. H. J. Leamy and G. H. Gilmer, J. Crystal Growth, vols. 24/25, 1974, p. 499. 163. J. P. Van der Eerden, P. Bennema, and T. A. Cherepanova, Prog. Crystal Growth Charact., vol. 1, 1978, p. 219. 164. B. Perraillon, I. M. Torrens, and V. Levy, Scripta Metall., vol. 6, 1972, p. 611. 165. P. G. Flahive and W. R. Graham, Surface Sci., vol. 91, 1980, p. 449. 166. C. L. Liu, J. M. Cohen, J. B. Adams, and A. F. Voter, Surface Sci., vol. 253, 1991, p. 334. 167. C. L. Liu and J. B. Adams, Surface Sci., vol. 265, 1992, p. 262. 167a. H. P. Bonzel, in Landolt-Börnstein, vol. 26, Diffusion in Solid Metals and Alloys, ed. H. Mehrer, Springer-Verlag, Berlin, 1990, pp. 717–747. 168. J. Y. Choi and P. G. Shewmon, TMS/AIME, vol. 224, 1962, p. 589. 169. G. E. Rhead, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 237–241. 170. J. M. Poate, K. N. Tu, and J. W. Mayor, eds., Thin Films: Interdiffusion and Reactions, Wiley-Interscience, New York, 1978. 171. S. R. Herd, K. N. Tu, and K. Y. Ahn, Appl. Phys. Lett., vol. 42, 1983, p. 597. 172. D. M. Follstaed and J. A. Knapp, Phys. Rev. Lett., vol. 56, 1986, p. 1827. 173. D. Gupta and P. S. Ho, eds., Diffusion Phenomena in Thin Films and Microelectronic Materials, Noyes, Park Ridge, N.J., 1988. 174. A. L. Laskar, Defect and Diffusion Forum, vol. 83, 1992, pp. 207–234. 175. A. V. Chadwick, Defect and Diffusion Forum, vol. 83, 1992, pp. 235–258. 176. C. R. A. Catlow, Proc. Roy. Soc., London, Ser. A353, 1977, p. 533. 177. A. S. Nowick, Defect and Diffusion Forum, vols. 66–69, 1989, pp. 229–256. 178. U. Gösele, in The encyclopedia of Advanced Materials, Pergamon Press, Oxford, 1993, pp. 629–635. 179. U. M. Gosele and T. Y. Tan, in Electronic Structure and Properties of Semiconductors, vol. 4, vol. ed. W. Schroter, VCH, Weinheim, Germany, 1991, pp. 197–247. 180. U. Gösele and T. Y. Tan, Defect and Diffusion Forum, vol. 83, 1992, pp. 189–206. 181. R. B. Fair, in Semiconductor Materials and Process Technology Handbook, ed. G. E. McGuire, Noyes, Park Ridge, N.J., 1988. 182. U. Gösele, in Microelectronic Materials and Processes, Kluwer Academic Press, Dordrecht, The Netherlands, 1989, pp. 583–634. 183. T. Y. Tan, in The Encyclopedia of Advanced Materials, Pergamon Press, Oxford, 1993, pp. 629–635. 184. W. D. Laidig et al., Appl. Phys. Lett., vol. 38, 1981, pp. 776–778. 185. D. G. Deppe and N. Holonyak, Jr., J. Appl. Phys., vol. 64, 1988, pp. R93–113. 186. T. Y. Tan, U. Gosele, and S. Yu, Critical Rev. Sol. State and Mater. Sci., vol. 17, 1991, pp. 47–106. 187. M. A. Kirk and R. C. Bircher, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4016–4022.

DIFFUSION IN METALS AND ALLOYS

2.119

188. J. W. Corbett, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4034–4036. 189. B. S. Brown, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4036–4038. 190. K. Farrell and M. B. Lewis, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 2407–2409. 191. A. Barbu, A. Dunlop, D. Lesueur, and R. S. Averback, Europhys. Lett., vol. 15, 1991, p. 37. 192. W. Shilling and H. Ullmaier, in Nuclear Materials, vol. 10B, vol. ed. B. R. T. Frost, VCH, Weinheim, Germany, 1994, pp. 179–241. 193. H. L. Heinish, JOM, Dec. 1996, pp. 38–41. 194. R. S. Averbach, T. Diaz de la Rubia, and R. Benedek, Nucl. Inst., Meth., vol. B33, 1988, p. 693. 195. M. T. Robinson, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 2516–2518. 196. L. E. Rehn, Met. Trans., vol. 24A, 1993, pp. 1941–1945. 197. J. L. Brimhall, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 1459–1462. 198. S. Amelinckx and D. Van Dyck, in Encyclopedia of Materials Science and Engineering, supp. vol. 1, Pergamon Press, Oxford, 1988, pp. 77–85. 199. L. K. Mansur, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4834–4838; JOM, Dec. 1996, pp. 28–32. 200. R. E. Stoller, JOM, Dec. 1996, pp. 23–27. 201. J. A. Wang, F. B. K. Kam, and F. W. Stallman, in Effects of Radiations on Materials,ASTM STP 1270, eds. D. S. Gelles et al., ASTM, Philadelphia, 1996, pp. 500–521. 202. R. E. Stoller, in Effects of Radiations on Materials, ASTM STP 1270, eds. D. S. Gelles et al., ASTM, Philadelphia, 1996, pp. 25–28. 203. H. J. Wollenberger, in Physical Metallurgy, 4th ed., Elsevier, 1996, pp. 1621–1723. 204. E. Kuramoto, J. Nucl. Mater., vols. 192–194, 1992, p. 1297. 205. E. A. Little, J. Nucl. Mater., vol. 206, 1993, p. 324. 206. L. E. Rehm, Met. Trans., vol. 20A, 1989, pp. 2619–2626. 207. L. E. Rehm, in Encyclopedia of Materials Science and Engineering, supp. vol. 3, Pergamon Press, Oxford, 1993, pp. 1941–1945. 208. K. C. Russell, in Encyclopedia of Materials Science and Engineering, supp. vol. 1, Pergamon Press, Oxford, 1988, pp. 380–386. 209. R. Cauvin and G. Martin, J. Nucl. Mater., vol. 83, no. 1, 1979, pp. 67–78. 210. H. R. Brager and F. A. Garner, Effects of Radiations on Materials, ASTM STP 870, eds. F. A. Garner and J. S. Perrin, ASTM, Philadelphia, 1985, pp. 139–150. 211. S. M. Murphy, Met. Trans., vol. 20A, 1989, pp. 2599–2607. 212. E. P. Simonen and S. M. Bruemmer, JOM, Dec. 1998, pp. 52–55. 213. S. J. Rothman, L. J. Norwicki, and G. E. Murch, J. Phys. F. Metal Phys., vol. 10, 1980, pp. 383–398. 214. E. P. Simonen, JOM, Dec. 1996, pp. 34–37. 215. R. Allen et al., 7th International Symp: Environmental Degradation of Materials in Nuclear Power Systems—Water Reactors, NACE, Houston, 1995, pp. 997–1008. 216. K. C. Russell, in Encyclopedia of Materials Science and Engineering, supp. vol. 1, Pergamon Press, Oxford, 1988, pp. 380–386. 217. K. C. Russell, “Phase Stability under Irradiation”, Progr. Mater. Sc., vol. 28, 1984, pp. 229–434.

CHAPTER 3

SOLIDIFICATION

3.1 INTRODUCTION Solidification is the most important method of material preparation and a very familiar phase transformation that is usually associated with the formation of crystalline metals and alloys from liquid upon cooling.1 The microstructure is determined largely by the solidification process. Solidification and melting play important roles in many processes used in the fields ranging from production engineering to solid-state physics. Examples include ingot casting, foundry casting, single-crystal growth, directionally solidified alloys, rapidly solidified alloys, and glasses.2 This chapter discusses heat transfer in solidification, thermodynamics of solidification, nucleation, interface kinetics, solidification of alloys, cellular and dendritic solidification, polyphase solidification, solidification processes and casting structure, new solidification processes, and structure manipulation and control.

3.2 HEAT TRANSFER IN SOLIDIFICATION When hot metal is poured into a mold, the rate of heat extraction from the melt is of special interest because it usually (but not always) constitutes the ratedetermining process for the progress of solidification.3 Heat transfer may take place in different portions of the metal-mold system by conduction, convection, and radiation.4 The solidification rate influences directly the coarseness or fineness of dendritic structures and, therefore, controls the spacing and distribution of microsegregates such as coring, second phases, and inclusions. Thermal gradient during freezing is also of great importance because of its relation with the formation of microporosity in alloys.5 The analysis of heat transfer during solidification is complex due to continuous evolution of latent heat of fusion at the moving solid-liquid (S-L) interface, the behavior of S-L interface configuration, and the change in physical properties of the metal-mold system with temperature.5 In this section we briefly describe solidification in conducting molds, solidification limited by interface resistance, and solidification in insulating molds. 3.1

3.2

CHAPTER THREE

FIGURE 3.1 Temperature distribution during the solidification of a metal in a conductive mold.5

3.2.1 Solidification in Conducting Molds If poured into metal molds, castings freeze rapidly and temperatures change fast in both the mold and the casting. An understanding of the factors influencing solidification in metal molds is of paramount importance in permanent and pressure diecasting molds. The analysis of heat transfer during pouring of metal against a chill wall is more complex than that when pouring into a sand mold (discussed later), because metal molds are superior heat conductors to sand molds and the size of the main thermal resistance, the solid itself, increases with the progress of solidification. As shown in Fig. 3.1, a temperature drop takes place at the solidified metal-mold interface, because of the thermal contact resistance. In this case, it is assumed that the mold is held at a temperature T0 and the liquid metal is at its melting temperature Tm at the S-L interface. The surface temperature of the metal mold remains well below the melting point, because of high thermal conductivity of the metal being cast, while a considerable thermal gradient exists within the solidifying metal. More total heat extraction occurs during solidification due to the cooling of the solidified metal. The solution to the heat-transfer equation, corresponding to the above boundary conditions, shows the amount of solid growth as the square root of time and is given, in simple form, by M = 2 B K S r SC S t

(3.1)

where M is the thickness of the solidified material, t is the time, and KS, rS, and CS, respectively, are the thermal conductivity, density, and specific heat (capacity) of the solid metal. As in the case of insulating molds, the solidification time is proportional to the square of thickness or modulus. Here B, a pure number, is the integration constant, which has no specific algebraic form and must be numerically known for each particular condition. However, the values of B can be expressed, in good approximation, as CS ˘ È B = f Í(Tm - T0 ) H f ˙˚ D Î

(3.2)

SOLIDIFICATION

3.3

FIGURE 3.2 Measured solidified thickness and time for pure iron in a cast iron mold.6 (Reprinted by permission of Butterworths, London.)

where DHf is the latent heat of fusion. Here B increases with the increase of Tm T0 and CS or with the decrease of DHf. Since the function in Eq. (3.2) is not linear, it is difficult to calculate the precise effect of thermal properties on the rate of solidification in this situation. However, the parabolic law given by Eq. (3.1) holds well with experimental results, as seen from Fig. 3.2. Metals or conductive molds are best suited to thin-walled castings because of yielding short cycle times. To overcome the tooling costs in a reasonably short time, short cycle time and larger production runs are required. Therefore, the metal-mold requirements for pressure die castings must include conductivity and durability. This analysis is also justified for injection molding where thick sections have long cycle times and are difficult to mold.6

3.2.2 Solidification Limited by Interface Resistance In a large number of casting processes such as permanent mold casting, die casting, splat cooling, and powder manufacturing processes, there exists an insulating layer (air gap) between the solid and the mold due to shrinkage of the casting; then heat flow is controlled to a large extent by resistance at the mold-metal interface.6,7 Figure 3.3 shows the temperature distribution across the solidifying metal and mold, which illustrates that the temperature of the solid (effectively Tm) and that of mold (T0) are both practically constant. At the S-L interface, heat is generated due to the evolution of latent heat of solidification. The rate of heat production is thus regulated by the volume of the material solidifying in a particular time. Thus the heat flow rate across this interface for metal poured at its melting temperature Tm is Q dM = DHf r s A dt

(3.3)

CHAPTER THREE

3.4

FIGURE 3.3 Temperature distribution across the solidifying metal and mold in which interface resistance is dominant.5 (Reprinted by permission of Addisson-Wesley Publishing, Inc.)

where A is the area and dM/dt the velocity of the S-L interface. We can predict the rate of heat transfer across the solid-mold interface using a modified (form of) experimental law for heat flow Q = hA(Tm - T0 )

(3.4)

where h is the heat-transfer coefficient of the interface and A the interface area. By combining Eqs. (3.3) and (3.4) for the case of large, flat mold wall and integrating from M = 0 at t = 0, we obtain the general solution for M as a function of t: M=

h(Tm - T0 )t r s DH f

(3.5)

Thus the solid thickness increases linearly with time, and its growth rate is a function of the interfacial heat-transfer coefficient. Since the shape of the casting does not affect in any way the heat transfer across the interface, Eq. (3.5) can be generalized to calculate the solidification time tf for a simple-shaped casting in terms of its volume-to-surface-area ratio: tf =

r s DHf M r s DH f V = h(Tm - T0 ) h(Tm - T0 ) A

(3.6)

where V/A (= M) is called the modulus of a casting and is widely employed in the foundry industry. It allows comparison of freezing times of castings with different shapes and sizes and assumes that castings with the same modulus will have the same solidification time. It is also beneficial in the determination of the feeding requirements of castings.6 In case of castings such as gravity die casting where heat transfer occurs as a result of interface resistance, the solidification time tf is proportional to the casting thickness. Therefore, large castings are more economical than thin section castings. Solidification times are usually longer, and multiple molds are generally employed to increase the productivity. The large casting size implies the use of friable cores and increases of the range of shapes that can be produced.6

SOLIDIFICATION

3.5

3.2.3 Solidification in Insulated Molds Sand casting and investment casting are two important commercial processes for making shaped castings which employ relatively insulated molds.7 The largest tonnage of metal is cast in sand molds, except the quantity of steel cast in ingot molds. The following analysis is used for solidification of melt in refractory mold such as sand molds, molds made of plaster, granulated zircon, or various other materials that have poor thermal conductivity. Hence mold itself is the main resistance to heat flow. Consider pure liquid metal without superheat poured against a flat sand mold wall. Figure 3.4 depicts the temperature distribution during solidification of metal in a sand mold at some time. Since all the resistance to heat flow is entirely confined within the mold and the thermal conductivity of the casting is very much higher than that of its mold, mold surface temperature TS is nearly equal to the melting point of the metal Tm. That is, the temperature drop during freezing through the solidified metal is small, and at the metal-mold interface a constant temperature TS 艐 Tm is maintained. We assume an infinitely thick mold whose outer surface is at T0 and inner surface is immediately heated to Tm at time t = 0. Finally (at t = •) the temperature gradient in the mold will be linear, but the metal solidification takes a very long time before this happens. The solution arising from this boundary condition is based on the assumption that the amount of heat which flows in the mold must equal the latent heat evolved during solidification. This solution can be finally expressed as M=

V 2 Ê Tm - T0 ˆ = Á ˜ K m r mCm t A p Ë r s DH f ¯

(3.7)

where Km, rm, and Cm are the thermal conductivity, density, and specific heat capacity of the mold material, respectively. In this case the amount of solidification depends on certain metal characteristics (Tm - T0)/(rS DHf) and the mold’s heat diffusivity KmrmCm (which is a measure of the ability of mold to absorb heat at a certain rate) and is proportional to the square root of time. Note that since a high melting temperature favors solidification, steel castings solidify faster than similar cast iron castings. Similarly, low DHf favors rapid solidification, so that, in spite of their similar melting temperatures, Mg alloy castings solidify faster than Al alloy castings.6 The thickness of the solid metal is a parabolic function of time, which implies the solidification rate is initially very rapid and decreases as the mold becomes heated.

FIGURE 3.4 Temperature distribution during solidification of a metal in a sand mold.5 (Reprinted by permission of Addison-Wesley Publishing, Inc.)

CHAPTER THREE

3.6

FIGURE 3.5 Distance solidified as a function of square root of time for several metals in insulating molds.7 (Reprinted by permission of McGraw-Hill, Inc.)

Figure 3.5 shows the wide range of solidification rates achieved in practice, based on metal and mold and mold temperature.7 The product KmCm is a useful parameter for evaluating the ability of mold material to absorb heat and is called heat diffusivity. If we rearrange to give solidification time tf of a casting in terms of its volume-to-surface-area ratio (or modulus) V/A, we obtain the following equation: V t f = CÊ ˆ Ë A¯ where

C=

p Ê r s DHf ˆ 4 Ë Tm - T0 ¯

2

(3.8)

2

K m r mCm

(3.9)

Equation (3.8) is the well-known Chvorinov’s rule, and C is a Chvorinov constant for a given metal-mold material and mold temperature. This works well for casting configurations where none of the mold material becomes saturated with heat such as in internal cores or internal corners. The success of this relationship relies on the mold material absorbing the same amount of heat per unit area exposed to the metal. As said earlier, it holds for castings with similar shapes but different sizes. For shapes such as cylinders and spheres, a more appropriate expression than Eq. (3.8) is derived without retaining the assumption of nondivergency of heat flow. In this situation, the partial differential equation for heat flow in the mold is given by ∂T ∂ 2T n ∂T = am 2 + ∂t ∂r r ∂r

(3.10)

where am (= Km/rmCm) is the thermal diffusivity of the mold, r is the casting radius, and n is 1 for cylinder and 2 for sphere. The resulting equivalent equation, similar to Eq. (3.7), is V Tm - T0 Ê 2 K m r mCm t f nK m t f ˆ = + A r s D Hf Ë p 2r ¯

(3.11)

By comparing Eqs. (3.7) and (3.11), it seems that the simple Chvorinov approximation becomes increasingly valid with the decrease in thermal diffusivity am

SOLIDIFICATION

3.7

(= Km/rmCm). It holds more for cylinders than for spheres. For a certain V/A ratio, a sphere solidifies more rapidly than a cylinder and a cylinder more rapidly than a plate.7

3.3 NUCLEATION DURING SOLIDIFICATION Nucleation during solidification is defined as the formation of a small crystal from the melt that is capable of continued growth.4 Nucleation process plays a vital role in the solidification of castings by controlling to a large extent the initial structure type, size scale, and spatial distribution of the product phases. Nucleation effects in the solidification microstructure exert a great influence on the grain size, morphology, extent of segregation, and compositional homogeneity. The final microstructure is also modified by the crystal growth, fluid flow, and structural coarsening processes that dominate in the latter stages of ingot freezing.8 These microstructural features have important bearings on the mechanical properties of an as-solidified material. Hence, it is necessary to understand and control the nucleation behavior. Nucleation can occur homogeneously and/or heterogeneously. Homogeneous nucleation takes place in pure liquid without the aid of foreign particles. Heterogeneous nucleation implies that nucleation preferentially takes place on foreign substances.9 (See Chap. 6 for nucleation in solids.)

3.3.1 Homogeneous Nucleation The consideration of nucleation and the extent of undercooling introduce another type of deviation from full equilibrium, called metastable equilibrium. From the thermodynamic point of view, the solidification process cannot occur at full L-S equilibrium. Solidification is possible only when there is a departure from full equilibrium that causes the liquid to remain in a metastable, undercooled state. For a pure metal at the thermodynamic melting point Tm, the free energy change per unit volume, which acts as the driving force, is given by DGv = GvS - GvL = H S - H L - Tm (SS - SL ) = 0

(3.12)

or

DGv = DHf - Tm DS = 0

(3.13)

so that

DH f =

DS Tm

(3.14)

where GvS and GLv , HS and HL, and SS and SL are free energy, enthalpy, and entropy per unit volume for the solid and liquid, respectively. The enthalpy change DHf is the latent heat of fusion Lv per unit volume. At temperature T, other than Tm, DGv π 0, and Eqs. (3.12) and (3.14) can be combined to give DGv =

D Hf (Tm - T ) D Hf (DT ) Lv DT = = = DS f DT Tm Tm Tm

(3.15)

where DT is the undercooling or supercooling. The total free energy change per unit volume DG to form a solid embryo of spherical shape of radius r from pure liquid consists of the change of volume free energy (which has a negative contribution) DGv and the interfacial free energy DGi and is given by

3.8

CHAPTER THREE

DG = DGv + DGi = -

4pr 3 D Hf ( = Lv )DT + 4pr 2g SL 3Tm

(3.16)

where gSL is the S-L interfacial free energy. The critical radius r* occurs, when DG has a maximum given by the condition [∂DG(r)/∂r]r=r* = 0 as r* =

2g SL 2g SL Tm = Lv DT DGv

(3.17)

Figure 3.6, due to Kurz and Fisher,1 shows a comprehensive plot of free energy change of an embryo as a function of its radius and DT: (a) At T > Tm, both DGv and DGi increase with r. Hence, the total DG increases monotonically with r.(b) At Tm, DGv = 0, but DGi still increases monotonically. (c) Below Tm, the sign of DGv becomes negative due to the metastable liquid whereas the nature of DGi remains unchanged, as in (a) and (b). At large values of r, the cubic dependence of DGv dominates over DGi, and DG passes through a maximum at the critical radius r*. When a thermal fluctuation results in a growth of embryo larger than r*, the total free energy of the system decreases if the growth of the solid occurs. Such clusters (or embryos) thus grow spontaneously and are termed nuclei, whereas clusters

FIGURE 3.6 Free energy change of an embryo as a function of its radius at temperatures (a) T > Tm, (b) T = Tm, and (c) T < Tm.

SOLIDIFICATION

3.9

smaller than r* can lower its free energy by dissolution of the solid. Unstable solid particles with r < r* are called clusters or embryos whereas stable particles with r > r* are called nuclei and r* is called the critical radius. Nucleation in a homogeneous melt (containing no solid phase) is called homogeneous nucleation, and, from Eq. (3.16), the critical free (or activation) energy for nucleation of a spherical nucleus of radius r* in a pure melt is given by DG* = DG *hom =

3 3 16p g SL 16p g SL Tm2 = 2 2 3 DGv 3 Lv DT 2

(3.18)

Equation (3.16) shows effectively how the maximum cluster size rmax is present in the liquid and how it increases with the decrease in temperature, as shown schematically in Fig. 3.7. This indicates the undercooling DTN at which nucleation occurs. The maximum cluster size exceeds r* at an undercooling DTN, but the probability of occurrence of clusters only slightly larger than rmax is very small. Also note that as undercooling DT increases, both DG* (DG* µ DT-2) and r* (r* µ DT-1) are reduced.2,10 A high value of gSL implies a high value of r*, a high DG*, and a nucleation problem. The value of gSL is considered independent of temperature. It is noted that undercooling phenomena are of great importance in the nucleation of both equilibrium and nonequilibrium crystalline phases and in the formation of amorphous phases. The amount of undercooling is a critical factor in the determination of many practical aspects of solidification practice, including morphological evolution, final solidification structure, phase selection or manipulation, and grain refinement.8

3.3.2 Homogeneous Nucleation Rate in Liquid To calculate the nucleation rate, we consider the entropy of mixing between a small number Nn of crystalline clusters, each of which contains n atoms, and NL atoms per

FIGURE 3.7 Schematic plot of critical radius r* and the maximum cluster size radius in the liquid rmax as a function of undercooling DT. This indicates the undercooling DTN at which homogeneous nucleation occurs.

3.10

CHAPTER THREE

unit volume in the liquid. The metastable equilibrium concentration of these spherical clusters or embryos Nn of a given size (Nn 2 grow facets. Polymers with large a values grow as spherulites having complex internal structures. The parameter a that controls the equilibrium interface consists of two parts: (1) x, which is a function of the structure of the solid and the orientation of the interface, is always 2, the edge energy becomes nonzero and the growth will be lateral with faceted interface. At some value of increased supercooling, the driving force is comparable to the work of creating a new disk, and the advancing interface becomes rough. The kinetic roughening then allows the occurrence of the continuous growth. Growth by Propagation of Twin Plane. Twin-plane reentrant corner (TPRE) is another source of generation of layers which occurs at both low and high supersaturation due to low edge energy on the rough twinned area and the higher nucleation rate along the reentrant corner because of the reduction in the formation of the partial MNG at the reentrant corner (Fig. 3.15d).38,39 This mechanism of step growth arises from the existence of reentrant grooves or angles on the surface which form when multiple twins terminate on the surface and which generate a source for the initial growth of steps.46,47 This twinned crystal requires smaller undercooling for a given velocity than the 2DN and the screw dislocation mechanism. Twin boundary, rotational twin boundary, and twin planes are found as the operating growth mechanisms in silicon plate and wafer growth48,49 and diamonds, eutectic Al-Si, and graphite growth.50 Recently, it has been shown that stacking faults and twin lamellae, like screw dislocations, can also act as self-perpetuating step-generating sources during crystal growth.

3.4.3 Interface Structure of Single-Phase Binary Alloys Kerr and Winegard51 extended the Jackson model to include alloy melt, and they demonstrated that the entropy of solution DSaL should be employed instead of the entropy of fusion in the derivation, which is given by DS aL = (1 - xea )S1L + xea S2L - S a

(3.41)

where xae is the concentration of component 2 in the liquid, SL1 and SL2 are the partial molar entropies of components 1 and 2, respectively, in the liquid alloy, and Sa is the molar entropy of the solid. Taylor et al.52 and Croker et al.53 have employed this approach to describe the various microstructures. More general approaches of Jackson54,55 show that DSaL is really a significant parameter for the determination of the faceting behavior as well as the rate of crystal growth and its anisotropy. Mamta et al.23 and Ramachandran and Srikanth56 have confirmed the requirements of both a high value of entropy of solution and a strong temperature dependence of the entropy of solution for promoting faceting behavior in the primary phases during eutectic solidification of the Al-Sn, Ag-B, Ag-Pb, Ag-Pd, Sn-Zn, and In-Zn systems, but fails in the case of the Al-Ge system.23,56 Usually Jackson’s a criterion is useful in predicting the system for the occurrence of facet formation, whereas the criterion of strong temperature dependency of entropy of solution is advantageous in predicting the composition range for facet formation. However, Jackson’s criterion describes most of the experimental results with a few exceptions. For example, crystals of Al2Cu and white P exhibit faceting in spite of their a parameters being less than the critical value.57

CHAPTER THREE

3.24

3.5 INTERFACE GROWTH Growth of the interface depends on the heat and solute flow and the conditions of the S-L interface. During crystal growth, the instability of the smooth and planar S-L interface will result in its breakdown and the formation of cellular or dendritic interface morphologies. Linear morphological stability theory deals with the conditions under which the interface becomes unstable.58 Stability conditions describe the pattern formation of cellular and dendritic growth. Here we briefly describe the interface conditions and heat and solute transport during the unidirectional solidification of a single-phase alloy from its melt. 3.5.1 Interface Conditions Under normal conditions, the growth rate of the interface is lower than the rate of atom transport; the local equilibrium is assumed to apply at the interface. When a liquid solidifies, the interface temperature TI is related to the interface liquid composition CI, the surface curvature, and the departure of the interface from the equilibrium, as given by59 DTI = T0 - TI = DTS + DTR + DTk

(3.42)

where T0 is the convenient reference temperature and DTS, DTR, DTk are the undercoolings resulting, respectively, from solute, curvature, and kinetics. For local equilibrium at the S-L interface, the coefficients obtained from the phase diagram are the liquidus slope m and the distribution coefficient k. If Tm is the melting temperature for an alloy of composition C0, the solute undercooling DTS resulting from the solute buildup around the planar interface or tip is given by DTs = m(C0 - CI )

(3.43)

The local liquid composition CI deviates from the bulk composition due to the rejection of solute at the S-L interface. The distribution coefficient, for low-growth velocities, is defined by CS = kCI

(3.44)

where CS is the solid solute concentration at the interface. During fast growth, such as in splat quenching, the usual assumption of local equilibrium at the S-L interface does not hold good, and k departs from the equilibrium value, that is, k = f(V), leading to solid trapping at high growth rates. The curvature or capillary undercooling DTR is characterized by the equilibrium conditions at a curved interface which is associated with the depression in freezing temperature due to the Gibbs-Thomson curvature effect. It is given, for isotropic surface energies, by 1ˆ Ê 1 DTR = A + Ë R1 R2 ¯

(3.45)

where A is the Gibbs-Thomson coefficient and R1 and R2 are the principal radii of curvature. For a planar or faceted solidification, DTR = 0. The kinetic undercooling at the interface DTk, resulting from the nonequilibrium effects, is necessary to drive the growth process or interface reaction in which atoms are transferred from the liquid to the solid. (DTk Æ 0 as the velocity V Æ 0.) Kinetic

SOLIDIFICATION

3.25

undercooling DTk has been predicted to be about 10-4 K for typical crystallization rates and can be totally neglected relative to other terms.7 For nonfaceting materials, this term is assumed to be small and fairly isotropic, which is given by DTk = BV

(3.46)

where B is a constant. For a faceting material, the normal velocity of a facet depends on the rate of production of steps on the face and thus is dependent in a much more complex manner on the local undercooling kinetics. In the case of nonfaceted growth, DTk is normally omitted except for very fast growth rates. In nonfaceted growth, DTk and DTR should be fairly isotropic. However, certain crystallographic features can be described only by incorporating a slight anisotropy in these terms. Both the solute and heat flow are described at the S-L interface, resulting in the final two interface equations: For solute, For heat flow,

Ê ∂c ˆ Ê ∂c ˆ V (CS - CT ) = DLC Á ˜ - DSC Á ˜ Ë ∂n¯ Ë ∂n¯

(3.47)

Ê ∂T ˆ Ê ∂T ˆ VL = K S Á ˜ - K L Á ˜ = K SGS - K LGL Ë ∂n ¯ Ë ∂n ¯

(3.48)

where V is the growth rate; DLC and DSC are the liquid and solid diffusivities, respectively; n is normal to the interface; L is the latent heat of fusion per unit volume; KL and KS are the thermal conductivities of liquid and solid, respectively; and GS and GL are the normal components of the thermal gradient in solid and liquid, respectively. If k < 1, the solute is rejected at the interface and thus concentrated in the liquid ahead of the interface. If k > 1, on the other hand, solute is depleted in the liquid ahead of the interface.

3.5.2 Heat and Solute Transport In the simplest case, where fluid motion is neglected, heat and solute flow occurs by conduction and diffusion.60 The equations ∂T ∂t ∂C DiC — 2C = ∂t

DiT — 2T =

(3.49) (3.50)

must be solved in both the solid and liquid phases (t is time, DiT and DiC are the thermal and solute diffusivities in the relevant phases, respectively). For steady-state growth (i.e., planar front, cellular and lamellar eutectic growth, and dendrites in the vicinity of tip), these equations may be transformed to coordinates moving with the interface, which then become61 DiT — 2T + V

∂T =0 ∂x

(3.51)

DiC — 2C + V

∂C =0 ∂x

(3.52)

3.26

CHAPTER THREE

where x is the steady-state growth direction. A steady-state solution, therefore, involves solving Eqs. (3.51) and (3.52) for an interface shape that satisfies Eqs. (3.42), (3.47), and (3.48). When fluid flow occurs in most solidification processes, the complex effect of convection should be incorporated in the transport equations. The effect of fluid flow is discussed later.

3.6 PLANE FRONT SOLIDIFICATION OF ALLOYS The solidification rate in a pure metal is almost always a controlled heat flow problem; however, in single-phase and eutectic alloys, it is a heat and solute flow problem. The controlled solute diffusion plays an important role in the solidification of alloys. The assumption of local equilibrium or metastable equilibrium at the L-S interface also has been found useful in explaining solidification at rates normally encountered in commercial practices. This may be shown by considering the solidification of a single phase and a eutectic alloy. Plane front solidification (or growth) is widely used to grow semiconductor single crystals, jewelry, and oxide crystals for laser systems and other applications; refine material (zone refining); obtain uniform or nonuniform composition within the solidified alloy by controlling solute redistribution; and understand the more complex interface morphologies involving cellular and dendritic growth. Planar growth of the S-L interface may be macroscopically or microscopically planar. The former is achieved by controlled directional solidification, good furnace design, and no convection in the melt. The latter is achieved by removing interface instabilities due to constitutional supercooling. This section addresses solute redistribution during unidirectional solidification under different conditions. 3.6.1 Solute Redistribution during Unidirectional Solidification We will consider a one-dimensional quantitative treatment of the solute redistribution during the solidification of a single phase or a eutectic alloy with simple phase diagram in order to (1) describe the methods of purifying metals (based on this principle) and (2) understand the effects of this redistribution on the micro- and macrostructure of the solidifying alloy. Here it is essential to make the following simple assumptions:62,63 1. The S-L interface corresponds to the cross-sectional plane of a long, narrow crucible and advances at a constant velocity V. 2. In the idealized binary region of the phase diagram, the liquidus and solidus are straight lines. The ratio of their slopes is a constant, given by CS/CL = k0 (Fig. 3.18), which implies that local equilibrium always exists at the S-L interface. The solute concentration gives rise to either a decrease (Fig. 3.18a) or an increase in melting temperature (Fig. 3.18b). 3. Solute diffusion is of vital importance for complete, limited, or partial mixing of the melt (rather than convection). 4. No solid-state diffusion occurs. Figure 3.19 shows the solute distribution in the solid bar after unidirectional plane front solidification under four different limiting conditions, which are described below.

SOLIDIFICATION

3.27

FIGURE 3.18 Schematic portion of binary-phase diagrams with linear solidus and liquidus curves and very small Tm - Ti value where (a) solute lowers (k0 < 1) or (b) raises (k0 > 1) the melting point of the alloy relative to the pure solvent.

FIGURE 3.19 Solute distribution curves for a bar solidified unidimensionally under various conditions: (a) complete diffusion in solid and liquid; (b) complete mixing in the liquid, no solid diffusion; (c) diffusion in liquid only; and (d) partial mixing in the liquid, including convection.4 (Courtesy of Elsevier Science.)

3.6.1.1 Equilibrium Solidification. Let us consider that the first solid of composition k0C0 begins to form at the liquidus temperature TL (Fig. 3.20a). The S-L plane front moves so slowly that equilibrium conditions at the interface are reached at temperature T* ( 1, there is usually a solute enrichment at the front end of the specimen. The zone refining technique performed without a crucible on a vertical rod, which relies on the surface tension to support the molten zone, is of great significance in the production of pure semiconductors.

3.6.2 Lateral Segregation The segregation in the lateral direction cannot be treated in a unidirectional manner, because of (1) the unavoidable presence of slightly curved S-L interface during steady-state unidirectional solidification; (2) the difference between the apparent distribution coefficients at the center and at the interface during the progress of solidification with a partially faceted front; and (3) a significant role of convective flow. The two-dimensional lateral segregation treatments present some difficulties in that such results do not give the information as to whether axially symmetric descriptions indicate reality. It is thus essential to study, first experimentally and then theoretically, what type of flow geometries appears under different thermal boundary conditions.72,81

3.6.3 Morphological Stability of a Planar Solidification Front So far the solidification front of a single phase and a eutectic alloy has been assumed to be a microscopically planar S-L interface, coinciding with the solute redistribution during unidirectional solidification. However, actually, preferential rejection or incorporation of solute and local temperature fluctuation, mechanical shocks, and convection within the melt occur at the interface and may produce small morphological perturbations of the interface. This morphological instability of the front interface influences and leads to the development and pattern formation of cellular and dendritic growth. Thus morphological stability theory deals with the conditions under which the smooth and planar S-L interface becomes unstable. Such morphologies along with the associated microsegregation of solute or impurities determine the microstructure and defect structure of the growing phase. There are many reviews of the morphological stability; some of the recent ones are given.1,55,85–87 3.6.3.1 Constitutional Supercooling (CS). A thermodynamic criterion for interface instability on the principle of constitutional supercooling (Fig. 3.25) was proposed by Rutter and Chalmers,88 Tiller et al.,61 and Winegard.89 This criterion considers steady-state solute distribution in the liquid CL at any position x in front of an S-L interface (Fig. 3.25a), as given by Eq. (3.59), and the corresponding equilibrium liquidus temperature TL (Fig. 3.25b), which is given by VX ˆ ˘ 1 - k0 TL = Tm - mCL = Tm - mC0 ÈÍ1 + expÊ Ë DL ¯ ˙˚ k0 Î

(3.63)

where m is the slope of the liquidus line corresponding to the solidifying alloy and Tm is the melting temperature of the pure metal. If we assume that freezing occurs only at an advancing S-L interface, the interface temperature TI for the flat front, i.e., for concentration C0 /k0, is

SOLIDIFICATION

3.37

FIGURE 3.25 (a) Solute concentration CL ahead of the L-S interface and the corresponding liquidus temperature TL. (b) Building of constitutional supercooling showing actual temperature profile where part of the liquid ahead of the S-L interface is below its actual or normal freezing as the growth velocity is raised above the critical velocity VC (V2 > VC > V1).

TI = T0 -

mC0 k0

(3.64)

The actual temperature profile T in the liquid is considered linear to a good approximation and is given by T = TI + GLx

(3.65)

where GL is the temperature gradient in the liquid ahead of the S-L interface. Combining Eqs. (3.64) and (3.65) yields T = T0 -

mC0 + GL x k0

(3.66)

When this temperature is superimposed on the curve for equilibrium freezing temperature, as in Fig. 3.25c, it is seen that the temperature and composition fields in the melt lie below the local liquidus temperature.20 This is called constitutional supercooling because it develops by combined solute and temperature distribution in the liquid ahead of an advancing L-S interface, regardless of whether k0 > 1 or k0 < 1. For onset of constitutional supercooling GL C0 1 - k0 £m V DL k0

(3.67a)

or

DT0 GL £m V DL

(3.67b)

or

GL £ mGc

(3.67c)

3.38

CHAPTER THREE

where DT0 = TS - TL = freezing range of the alloy = mC0(1 - k0)/k0 and Gc = C0[(1 - k0)/k0] (V/DL). According to the nature of Eq. (3.67), above a critical value of GL/V, the S-L interface is stable, i.e., planar; below the critical value, instability occurs and the morphology changes. A planar interface under CS conditions becomes unstable because any bulge forming on the interface leads to transverse diffusion of solute atoms, their accumulation, and a localized decrease in the liquid temperature. This, in turn, will break up or result in the formation of cellular or dendritic microstructure.90 Figure 3.26a is a schematic plot of the relation between GL and V, called GLV space, showing its division by lines into different GL/V values to differentiate between planar to cellular instability and cellular to dendritic instability.91 Cellular and dendritic instabilities are shown in Fig. 3.26b. Equation (3.67) is applicable to both the presence and absence of convection because of the formation of laminar layer near the solidifying interface in every case.7 Both theoretical and experimental results support the CS theory of predicting conditions of breakdown of the planar front in metals and semiconductors solidifying with a diffuse interface. 3.6.3.2 Mullins and Sekerka (MS) Instability Theory. The CS criterion has resulted in better understanding of many crystal growth, solidification, and casting processes. However, this concept does not yield any information about the size scale of instability. Also, for some solidification processes such as rapid solidification, the CS criterion can lead to erroneous results because this criterion ignores the temperature gradient in the solid and the S-L interfacial energy. In these situations, the theory of morphological instability was developed and analyzed by Mullins and Sekerka,92 Sekerka,93,94 and others95,96 to fill in these gaps. They assumed a local equilibrium at the S-L interface, isotropy of the S-L surface tension (which is a very good approximation for many metals at low growth velocities), and initial occurrence of a very small amplitude of sinusoidally perturbed S-L interface with isotropic interface plane (x, y) moving in the pure melt in the z direction (Fig. 3.27), which is expressed by

FIGURE 3.26 (a) Diagram of GLV space showing regions of cellular and dendritic growth as a function of temperature gradients and solidification rates. (b) Cellular and dendrite instability.

SOLIDIFICATION

3.39

FIGURE 3.27 Sinusoidally perturbed S-L interface as observed from a reference frame that advances along with the unperturbed interface (z = 0).94 (Reprinted by permission of Pergamon Press, Plc.)

z( x, t ) = d (t ) Cos wx = d (t ) exp(s t + iwx)

(3.68)

where d(t) is the amplitude of a Fourier component of the interface shape, w is the wave number of perturbation = 2p/l, l is the wavelength, and s (= sr + isi) is the growth (or decay) rate of perturbation represented by a complex quantity. If sr is positive or negative, respectively, for all values of w, the interface becomes unstable or stable with respect to perturbation of all Fourier components. Using the appropriate equations and boundary conditions, a relationship between s and w, called the dispersion relation, is required to determine the stability limit of the system and, therefore, derivation of the critical wavelength at the threshold l* = 2p/w*. This is the stability exchange principle which is given by sr(w) < 0; [sr(w)]max. = 0, and si(w) = 0.59 The dispersion relation, calculated by the these authors by assuming the thermal conductivity weighted temperature gradient (G) on the liquid side and a (Sekerka’s) stability function S(w) or S(A, k0), dependent on the capillarity parameter A and k0, is expressed by È K SGS + K LGL DL ˘ 1 S( A, k0 )˙ s (w ) = Í + V Î KS + KL ˚ DT0 2

A=G

Ê k02 ˆ Ê Tm ˆ Ê V ˆ Ê Tm k0 ˆ Ê V ˆ = GÁ ˜ Ë mGc ¯ Ë DL ¯ Ë 1 - k0 ¯ Ë mC0 ¯ Ë DL ¯ G=

K SGS + K LGL KS + KL

(3.69) (3.70) (3.71)

where KS and KL are the thermal conductivities of the solid and liquid, respectively, G is the Gibbs-Thompson parameter, and Tm is the melting point of pure metal. The final analysis of Eq. (3.69) leads to the instability criterion if G < mGc S( A, k0 )

(3.72a)

G > mGc S( A, k0 )

(3.72b)

or the stability criterion if

CHAPTER THREE

3.40

The instability of the planar interface for dilute Al-0.1 wt% Cu alloy has been found to exist over a range between two critical velocities. Beyond this range, the growth (or decay) rate of perturbation s < 0 for all wavelengths l, and the interface remains always stable.91 The upper and lower critical velocities may be approached closely by two different limiting cases: 1. For normal solidification rates, i.e., at low growth rates of solidification (or interface velocity), the values of the capillary forces (or interfacial tensions) are so small that A 1. The temperature gradient term GL is always positive for an undercooled melt because of negative temperature gradient in the liquid. Hence, the driving force for destabilization, which is the difference between the two gradients in Eq. (3.83), representing the solute and thermal effects, is always positive in an undercooled alloy melt. This destabilizing effect is balanced by the stabilizing effect of the capillary term on the right-hand side. For an Ivantsov dendrite which is assumed to be isothermal, the temperature gradient in the solid is zero, so that the effective temperature gradient reduces to G¢ = [(KSGS + KLGL)/(KS + KL)]zL. The values of composition gradient in the liquid at the tip G*c and the thermal gradient at the tip G*L are given from solute and thermal flux balance at the tip interface as G*c = -

V (1 - k0 )Ct DL

(3.85)

G*L = -

2 DH m Ê V ˆ Ê DH m ˆ =Pt Ë a L ¯ Ë cL ¯ R cL

(3.86)

Combining Eqs. (3.85), (3.86), and (3.83), one finds a general tip radius selection criterion in an undercooled alloy melt as Ê DH m ˆ 1 cL ˜ Ê k0 DT0 ˆ Ê Ct ˆ 2Á x + VR Á VR ˜ xL = Ë GDL ¯ Ë C0 ¯ c Ë 2 Ga Lb ¯ s* 2

(3.87)

where b = 0.5 (1 + KS/KL), KL = aLcL, and DT0 = mC0(k0 - 1)/k0. Note that b = 1 when KS = KL. The value of the composition at the tip is given by Eq (3.76b). Equations

CHAPTER THREE

3.48

(3.81) and (3.87) now explain completely the growth problem in an undercooled alloy. Equation (3.87) shows that the two terms on the left-hand side are quite similar. They differ only by the factor 2b due to the one-sided diffusion problem, that is, DS >> k0GLDL, Eq. (3.96) converts to 14

l1 = 2 2 {DL G[ mC0 (k0 - 1)]} GL-1 2V -1 4

(3.97)

127

The Okamoto-Kishitake model correlates l1 with solidification parameters and assumes the secondary arms to be plates which are thickened during the course of solidification, which is given by -1 2 12 l1 = 2e[ - DLC0 m(1 - k0 )] (VGL )

(3.98)

where e is a constant ( Vtr, l1 is given by 14

Ê DL G ˆ -1 2 -1 4 l1 = 4.3 DT 1 2 Á ˜ V GL Ë k0 DT0 ¯

(3.101)

Both the Hunt and Kurz-Fisher models at high growth rates appear to be very similar except for the difference in the constant values. However, at low growth rate, the results obtained by applying these models are different. The Trivedi model130 is a modified Hunt model and assumes a marginal stability criterion which is given by l1 = 2 2GL

V -1 4 ( Ak0 DT0 GDL )

-1 2

14

(3.102)

where A is a constant which depends on the harmonic perturbations and, for dendritic growth, is equal to 28. According to the Hunt-Lu model131 which is based on the numerical modeling of cellular/dendritic array growth, the dimensionless primary spacing l¢ is given by l ¢ = 0.7798 ¥ 10 -1 V ¢(a - 0.75)(V ¢ - G¢)

0.75

G¢ -0.6028

(3.103)

where a = -1.131 - 0.1555 log G¢ - 0.7859 ¥ 10-2 (log G¢)2. This yields a value of a between -0.34 and -0.58. Note that l¢ is the radius rather than the diameter and is independent of k0. Hence, the values obtained from such a model should be multiplied by 2 to 4 for comparing with the experimental data.

3.52

CHAPTER THREE

Among all the above five theoretical models, the Kurz-Fisher and Hunt-Lu models allow one to estimate reasonably l1 as a function of directional solidification parameters.

3.7.5 Secondary Dendrite Arm Spacing A reduction in secondary arm spacing l2 obtained by the higher cooling rate of cast materials results in a reduction in microsegregation and will produce a more easily homogenized casting.1,7,132 Secondary arms form initially very near the dendrite tip, presumably due to the presence of noise at the dendrite tip. The precise distance behind the tip where sidebranches are observed depends on the strength of the noise (due to thermal or hydrodynamic fluctuations in the system) which is yet to be established. For both free and constrained growth, the ratio of the initial secondary arm spacing to dendrite tip radius is approximately 2.5 (actually between 2.0 and 3.0) over a wide range of growth conditions in different systems. Theoretical work by Langer and Mueller-Krumbhaar133 and two-dimensional numerical solution by Saito et al.134 have confirmed this relationship. However, the actual secondary spacing developed in a fully solidified material is coarser than that of the initial spacing. The coarsening of the final secondary arm spacing arises first from melting or dissolution of the smaller arm at the expense of the larger arms and then from the minimization of surface energy during the coming closer of sidebranches to the neighboring dendrites.101 For all except the most dilute alloys, the time-dependent law of coarsening of secondary spacing l2 has been developed for directional solidification conditions, which is given by135 l2 = 5.5(Mt f )

13

(3.104)

where tf is the solidification time and C ln Ê f ˆ Ë C0 ¯ GDL M= k0 DT0 C f -1 C0

(3.105)

where Cf is the final composition of the liquid at the base of the dendrite. For directional solidification, a relation between the local solidification time and the magnitude of the cooling rate |GLV| can be expressed by tf =

DTf GLV

(3.106)

where DTf is the nonequilibrium temperature range of solidification, i.e., the temperature difference between the tip and the base of the dendrite. Bower et al.110 obtained, within the experimental error, Eq. (3.92) to be valid in directionally solidified Al-4.5 wt% Cu alloy. The precise value of the parameter M differs slightly for different models; however, its effect on secondary arm spacing is small due to its relation through the cube root. Recently, Li and Beckermann116a have investigated an evolution of the sidebranch spacings in three-dimensional dendritic growth.

SOLIDIFICATION

3.53

3.7.6 Cell-Dendrite Transition When the degree of morphological instability v becomes substantial at intermediate velocities, the cellular structure undergoes a transition sidebranching instability, thereby transforming to a dendritic morphology. On the basis of a sharp change of the primary spacing with velocity and approximation of the tip shape, Kurz and Fisher129 proposed the condition for low-velocity cell-dendrite transition VC-D as VC -D = v=

or

GLDL k0 DT0

(3.107)

1 k0

(3.108)

However, the correlation between theory and experimental results was very poor, as reported by Tewari and Laxmanan.136 The condition for low-velocity cell-dendrite transition based on the Ivantsov solution can be given by VC-D, using the Ivantsov solution for solute diffusion given in Eq. (3.76b): VC -D =

GLDL [1 - (1 - k0 )Iv( P )] k0 DT0

(3.109)

This equation, at very small Peclet number, converts to the condition of Eq. (3.107) (for k0 < 1), as proposed by Kurz and Fisher on the basis of a sharp change of primary spacing with velocity. In the low-velocity cellular regime, the tip undercooling is mainly given by the term GLDL/V. In reality, the cell-dendrite transition occurs when the solute diffusion length CL are called nonperitectic alloys because they can form the peritectic phase b directly from the melt, and not through peritectic reaction.

3.8.1 Eutectic Solidification The eutectic alloys are of great industrial importance because of their following characteristics: (1) lower melting points than those of the pure components; (2) zero or small freezing ranges which effectively remove the dendritic mushy zone, reduce segregation and shrinkage porosity, and provide excellent fluidity (valuable in casting, welding, and soldering processes);138a and (3) the prospect of forming fine in situ composites139 which possess superior mechanical properties with a spacing which is an order of magnitude finer than the primary and final secondary arm spacing in dendritic alloys.140 The most widely used eutectic alloys include lowtemperature soldering (such as Pb-free, ternary eutectic Sn-3.5Ag-0.9Cu with eutectic temperature of 217°C)140a or brazing materials, wear-resistant alloys, and casting alloys such as cast irons and Al-Si alloys. Binary eutectics are usually classified, according to morphology, into two major classes, called normal or regular and anomalous or irregular eutectics. The factor that determines the formation of a particular type of microstructure in a given alloy system is the faceting tendency of the growing phases. When both phases grow with a nonfaceting S-L interface, the familiar regular eutectic microstructure is formed during solidification. The mechanism of formation of these microstructures is well established. Common examples are Al-

SOLIDIFICATION

3.55

CuAl2 and Pb-Sn eutectics. On the other hand, an irregular eutectic structure is formed when at least one of the phases solidifies with a faceted interface (e.g., binary Sn-Ag and Sn-Cu systems, where Sn is a nonfaceted phase and Ag3Sn and Cu6Sn5 are faceted).140a The formation of these microstructures is not well understood and is sometimes classified, in order of increasing volume fraction of the faceting phase, as broken lamellar, irregular, complex regular, and quasi-regular types. Binary eutectics where both phases solidify with faceted interface have been little studied.141 Generally, normal binary eutectic systems show regular microstructures of lamellar or fibrous and rod form (Fig. 3.34) and phase diagrams with similar melting points of metals. They form by simultaneous movement of the two solid phases in a coupled fashion at a well-defined S-L interface. They exhibit well-defined grain structure and an orientation relationship between the lamellae within the grains. Anomalous eutectic alloys form in systems with metals of different melting temperatures. The anomalous eutectic systems exhibit an irregular and wider range of microstructures in which a single, isothermal growth front does not exist and one phase has the tendency to grow ahead of the other, leading to a more ragged

(a)

(b)

(c)

FIGURE 3.34 Typical solidification microstructure of eutectic 60Sn-40Pb solder: (a) lamellar microstructure formed by furnace cooling and (b) fine equiaxed structure formed by rapid cooling. (c) Directionally solidified eutectic alloys showing a rod structure in the Sn-18 wt% Pb system. (a and b: Courtesy of Z. Mei and J. W. Morris, Lawrence Berkeley National Laboratory, Report LBL 31240, Feb. 1992. c: after Trivedi and Kurz, 1988.)

CHAPTER THREE

3.56

S-L interface.142 That is, in anomalous structure, minor phase is crystallographically dominant, grains cannot be discerned, and preferred orientation relationship is absent.143 3.8.1.1 Regular Lamellar Growth. During eutectic growth, also called cooperative growth, the solute rejected from one phase is laterally transported across the interface to the other as the interface advances forward. This leads to a concentration profile across the eutectic interface and the associated variation in solute undercooling DTS and the capillarity undercooling DTR. Since thermal diffusivities are much higher than the solute diffusivities, the variation in the interface temperature TI along the S-L interface, i.e., the total undercooling DT, is shown in Fig. 3.35. Simple lamellar or fibrous microstructures, also called nonfaceted-nonfaceted (nfnf) eutectics, are produced when both phases have low entropy of melting. In the Jackson-Hunt model and most of the other models, it is assumed that heat flow or thermal diffusivity is very large relative to that of the solute diffusivity, and the simplified solution of the liquid diffusion problem is used for a planar rather than a curved front.

B-rich liquid

A-rich liquid

Composition

a

b

Interface shape l

B

(b)

CE A Interface solute concentration DTR

Temperature

(a)

(c)

TE DTK

DTK

DTS DTR T1

DT

DT DTS Interface undercooling contributions

FIGURE 3.35 (a) Schematic representation of (a) lamellar a-b eutectic interface, (b) concentration profile (of liquid composition B) across an a-b interface, and (c) contribution of the total undercoolings DT present at the S-L interface, comprising undercoolings attributed to solute DTS, solute curvature DTR, and kinetic DTK effects. (After Hunt and Jackson, TMS-AIME, vol. 236, 1966, p. 843.)

SOLIDIFICATION

3.57

FIGURE 3.36 Schematic diagram of lamellar eutectic structure which defines coordinate system, contact angle at the triple junctions, x, z, l, Sa, and Sb.

Let us consider the steady-state growth of a eutectic structure, in which the solute flow occurs in the liquid by diffusion and there is no convection. A schematic diagram of lamellar eutectic, showing the definition of relevant length scales and angles required for the theoretical model, is shown in Fig. 3.36. The steady-state diffusion transformed to coordinates x and z (Fig. 3.36) attached to the S-L interface and moving at a constant velocity V in the z direction is given by — 2C +

V ∂C =0 DL ∂ z

∂ 2C ∂ 2C V ∂ C + + =0 ∂ x 2 ∂ z2 DL ∂ z

or

(3.110a) (3.110b)

where C is the composition. The boundary conditions are C = CE + C• ∂C =0 ∂x

at x = •

at x = 0 and x =

1 l = Sa + Sb 2

(3.111a) (3.111b)

where CE is the eutectic composition, l is the eutectic lamellar spacing, and Sa and Sb are the half-width of the a and b phases, respectively (defined in Fig. 3.36).145 For planar interface, the conservative equations at the boundary become

and

V a Ê ∂C ˆ C0 Á ˜ =Ë ∂ z ¯ z =0 DL

0 < x < Sa

(3.112a)

V b Ê ∂C ˆ C0 Á ˜ =+ Ë ∂ z ¯ z =0 DL

Sa < x < Sa + Sb

(3.112b)

where Ca0 = CLa - CSb and C b0 = CSb - CLb are both positive, are amounts of A and B rejected by the a and b phases as a unit volume solidifies, and depend on the local liquid composition [for a nonplanar interface the velocity and gradient in Eq. (3.112) would be normal to the interface]. At the S-L interface, it is essential to fulfill the undercooling equation as given by DT = TE - TI = DTS + DTR + DTk

(3.113)

CHAPTER THREE

3.58

where DTS = m(CE - C). The Jackson-Hunt model of eutectic growth which relates the eutectic spacing l with the growth rate V for directionally solidified alloys is effective at low velocities. This model assumes that, at low growth rates, DTk is small and DTR is the only major contributor to the total undercooling DT for nonfaceting materials. At very high growth rates, DTk may be substantial and the Ca0 and Cb0 of Eq. (3.112) would be dependent on both the liquid composition and the growth rate. The solution to diffusion Eq. (3.111) under these conditions is •

C = CE + C• + Â BnCos n =0

12 Ê ÏÔ V ÈÊ V ˆ 2 Ê 2 np ˆ 2 ˘ ¸Ô ˆ 2 npx -Í + expÁ - Ì ˙ ˝z˜ l Ë ÔÓ 2DL ÎË 2DL ¯ Ë l ¯ ˚ Ô˛ ¯

(3.114)

Since 2np/l >> V/2DL for n > 0, i.e., when the spacing l is smaller than the diffusion distance DL/V, Eq. (3.114) reduces to 2 npx 2 npz ˆ Ê V 2 ˆ n =• expÊ C = CE + C• + B0 expÁ BnCos ˜+ Ë Ë 2DL ¯ Â l l ¯ n =1

(3.115)

This assumption holds except at the highest velocities (typically > 10 mm/s). The Fourier coefficients Bn and B0 may be evaluated using the continuity of matter equations at the interface and are given by

and

Bn =

lVC0 2 npSa 2 Sin l DL ( np )

(3.116)

B0 =

2(C0a Sa - C0b Sb ) l

(3.117)

where C0 = Ca0 + Cb0 is shown in Fig. 3.37. The average compositions in the liquid Ca and Cb, respectively, at the interface ahead of the a and b phases at z = 0 are given by Ca = CE + C• + B0 +

2(Sa + Sb ) VC0 P Sa DL

(3.118a)

Cb = CE + C• + B0 +

2(Sa + Sb ) VC0 P Sb DL

(3.118b)

2

2

FIGURE 3.37 A eutectic phase diagram showing the quantities used in equations for conservation of matter in two-phase a-b lamellar growth.17 (Courtesy of John Wiley & Sons, Ltd.)

SOLIDIFICATION

3.59

where 3 È npSa ˘ 1 ˆ P = ÍÂ Ê Sin 2 ˙ Ë ¯ n S p a + Sb ˚ Î

(3.119)

Here P depends on Sa/Sb and is usually tabulated. The average undercooling on each phase becomes È 2(S a + S b ) VC0 P ˘ aLa DTa = ma ÍC• + B0 + ˙+ S a DL ˙˚ S a ÍÎ

(3.120a)

È 2(S a + S b ) VC0 P ˘ aLb DTb = mb Í- C• - B0 + ˙+ S b DL ˙˚ S b ÍÎ

(3.120b)

2

2

where ma and mb are the positive slopes of the liquidus lines and aaL and abL are constants, obtained from the Gibbs-Thomson coefficient, and are defined elsewhere.145 By eliminating B0 from Eq. (3.120) and assuming the volume ratio x = Sb /Sa to have its equilibrium value for the given alloy composition, the growth equation relating undercooling, growth velocity, and lamellar spacing, due to Jackson and Hunt,144 is given by

where

DT aL = VlQL + l m

(3.121)

1 1 1 = + m ma mb

(3.122)

2

QL =

P(1 + x ) C0 xD

(3.123)

abL ˆ Ê aaL a L = 2(1 + x )Á L + ˜ Ë ma xmb ¯

(3.124a)

TE aaL = Ê ˆag aL Ë L¯

(3.124b)

where g aL is surface energy (a-liquid). Here the superscript L denotes the lamellar growth. The Trivedi-Magnin and Kurz (TMK) model is a more general one to make theoretical predictions at high growth rate (large Peclet number) regime (under rapid solidification conditions). It consists of (1) the solution of the solute diffusion equation in the liquid with boundary conditions and nonequilibrium effect at the interface and (2) the use of a proper spacing selection criterion for lamellar eutectic for a given growth rate condition.146 Eutectic Spacing Selection. Assuming (1) planar and isothermal lamellar solidifying interface, (2) a coupling between the constitutional supercooling and the Gibbs-Thomson effect, and (3) the diffusion distance ahead of the interface much larger than the interlamellar spacing, Jackson and Hunt144 derived the following expression for the average undercooling DT at the solidifying interface as a function of solidifying rate V and lamellar eutectic spacing l:

CHAPTER THREE

3.60

FIGURE 3.38 A schematic plot of undercooling versus eutectic interlamellar spacing at a given velocity showing the stable and unstable spacings (or regions), as predicted by Jackson-Hunt analysis.

DT = K1lV +

K2 l

(3.125)

where K1 and K2, respectively, are the constant parameters represented by the constitutional supercooling effect and the Gibbs-Thomson effect of the solidifying interface. They used the theoretical values of the minimum and maximum stable spacings lm and lM, respectively, as shown in Fig. 3.38.147 On the basis of this assumption, they were able to derive a simple relationship of lm and lM as a function of growth rate V as l2mV =

and

K2 K1

or

lm µ V -1 2

(3.126)

Ê K1 ˆ 1 2 DT = 2 V = AV 1 2 Ë K2 ¯

(3.127)

lM2 V = K3

(3.128)

where K3 is a constant factor and is dependent on properties of the system only.147 Many experimental results in several alloy systems usually display a narrow band of spacings. On the basis of Zener’s maximum growth rate hypothesis for eutectic growth in an undercooled melt, Tiller proposed an equivalent criterion for directionally solidified eutectics. A more correct criterion should be based on the stability of the interface with respect to small variations in eutectic spacings. Jackson and Hunt concluded that of all the possible eutectic spacings predicted by Eq. (3.125), only a finite range of spacings will be stable with respect to fluctuations in the interface morphology. That is, the eutectic spacings lm will be stable. For the eutectic spacings >> lm, the steady-state interface shape develops a hollow or pocket at the center of the wider (or larger volume fraction) phase, as shown in Fig. 3.39b. At some large spacings, denoted by lM, the hollow interface gets deeper, i.e., the slope of the (hollow or pocket) interface becomes infinity so that all eutectic spacings above lM become unstable. Thus, Jackson and Hunt concluded that of all possible spacings based on diffusional growth considerations, only those spacings will be stable that lie in the range lM < l < lm. Figure 3.40 exhibits

SOLIDIFICATION

3.61

3 2 1

(a)

(b)

(c)

FIGURE 3.39 Schematic illustrations (a) of the instability of a lamella. Interface position 1 shows the originally regular lamella, position 2 the formation of narrower lamellae at the center, and position 3 the change in interface shape with time. (b) For a wider lamella spacing. The shape instability of the interface of one phase that occurs with very wide spacing. A new lamella forms in the hollow pocket. (c) A lamellar fault.145 (a: Courtesy of Elsevier Science, Amsterdam; b and c: Courtesy of McGraw-Hill, New York.)

FIGURE 3.40 Theoretically predicted relationships between the interface undercooling DT and the lamellar spacing l for different growth rates. The experimentally observed ranges of lamellar spacings at different velocities are denoted by the hatched regions, and the average spacings at the different velocities are shown by the solid circles.139 (Reprinted by permission of the Metallurgical Society, Warrendale, Pa.)

the theoretically predicted minimum and maximum stable spacings lm and lM, given by Eqs. (3.126) and (3.128), respectively, and indicated by the pairs of vertical arrows. It is seen that DT is largest for large and small lamellar spacings, because diffusion is difficult in the former and curvature effects are dominant in the latter. Criterion for Growth. In the Jackson-Hunt analysis, the extremum condition was ignored by applying a criterion noted on an observed growth mechanism. It was

CHAPTER THREE

3.62

suggested that the lamellar spacings are controlled by the stability of a lamellar fault (Fig. 3.39c). This figure shows that the average spacing decreases when the fault moves to the right. Stable lamellar growth occurs when the fault remains stationary.144 Extent of Eutectic Range. Lamellar eutectics can grow from a liquid of both eutectic and off-eutectic compositions by solidifying these materials with varying gradients and growth rates. When an alloy having off-eutectic composition solidifies, first dendrites or cells of the primary phase will occur at some undercooling below the liquidus temperature, and then the eutectic phase will appear from the remaining liquid reaching the eutectic composition.148 The extent of the range depends on the temperature gradient and the growth rate. Mollard and Flemings149 and Cline and Livingston150 obtained the eutectic structure in the off-eutectic compositions by solidifying them at varying growth rates and temperature gradients to maintain a planar interface. The two-phase structures are achieved with different volume ratio of the phases from that given by the equilibrium eutectic composition. According to the competitive growth mechanism, cells or dendrites cannot form unless they grow with a tip temperature higher than the eutectic growth temperature. Combining the Burden and Hunt dendritic growth model and the JacksonHunt eutectic model, the dendritic tip undercooling DTt as a function of velocity and temperature gradient can be given by DTt = T0 - Tt =

GLDL +B V V

(3.129)

where T0 is the alloy liquidus temperature and Tt is the tip temperature for eutectic; Eq. (3.127) is DT = TE - TI = A V

(3.130)

where A is a constant. At the critical condition DL = TI, the eutectic composition range on the a side of the eutectic becomes DC = CE - C0 =

T0 - TE GL DL = + [( B - A)ma ] V ma ma V

(3.131)

where B and A are constants for the dendrite and eutectic structures, respectively. At high gradients, the GLDL term dominates and the eutectic range is proportional to GL/V. At sufficiently higher velocities, the (B - A)÷V term dominates and the range is proportional to ÷V. It is thus clear that the eutectic can form and be stable at both very low and very high velocities.151 Figure 3.41 shows a schematic representation of this approach. When the composition is changed from A-rich to the B-rich alloy, the b liquidus temperature moves upward and a liquidus temperature downward with respect to the eutectic in order to change the transition velocities. Figure 3.42 shows the superimposition of results from Fig. 3.41 for different compositions on the phase diagram by plotting the eutectic growth temperature versus composition. This shows the composition-temperature zone in which eutectic microstructures can be found, and this region is usually called the coupled zone of competitive growth. Figure 3.42a shows an example for two different gradients and Fig. 3.42b for skewed eutectic due to very different growth kinetics of the primary phases. A comparison of theoreti-

SOLIDIFICATION

3.63

FIGURE 3.41 A schematic diagram of the a and b dendrite tip temperatures (the dotted lines represent plots for a zero gradient, the continuous lines are for a larger gradient) and the eutectic interface temperature plotted against square root of velocity for a given temperature gradient. It is assumed that the eutectic structure forms without dendrites when the eutectic has the higher growth temperature (i.e., at high and low velocities). Conversion from a-rich to b-rich phase will raise the b but lower the a line, resulting in a variation in the transition velocities.145 (Reprinted by permission of Elsevier Science, B.V., Amsterdam.)

b

a b dendrites

a dendrites eutectic

and eutectic

and eutectic

(a)

b

a a dendrites and eutectic

b dendrites and eutectic eutectic

Composition %B (b)

FIGURE 3.42 (a) The superimposition of results from Fig. 3.41 for different compositions on the phase diagram by plotting eutectic growth temperature against composition. The dashed line exhibits the boundary for a larger gradient. (b) The influence of a steeper b dendrite line on the eutectic range.145 (Reprinted by permission of Elsevier Science, B.V., Amsterdam.)

CHAPTER THREE

3.64

cal and observed eutectic ranges shows a reasonably good agreement. In the case of rapid eutectic solidification, Trivedi and Kurz proposed a schematic plot similar to Fig. 3.41 except that the eutectic line has a maximum velocity and the dendrite line changes to cellular and then planar again at very high velocities (Fig. 3.43). The results again can be condensed by superimposing on the phase diagram (see Fig. 3.44a). Figure 3.44b is an example for calculated results showing metastable phases.139 3.8.1.2 Rod Eutectics. Like lamellar structures, two nonfaceted phases form normal rodlike structures, and rod growth has been analyzed. Figure 3.45 shows a schematic diagram defining the length scales for a rod eutectic. The rod growth can be given by DT aR = VRQR + m R

(3.132)

R abR ˆ 1 2 Ê aa a R = 2(1 + x ) Á + ˜ Ë ma xmb ¯

(3.133)

where R is the rod spacing and

QR =

4(1 + x )C0 M xD

(3.134)

where aaR and abR are surface energy terms and M is a sum of Bessel function terms similar to P, defined and tabulated elsewhere.144 Like lamellar growth, Eq. (3.132) requires an additional condition for a solution. In the absence of a detailed mechanism for variation of rod spacings, the structure can be expected to grow at the extremum spacing. In this situation

FIGURE 3.43 Schematic plot similar to Fig. 3.41 except that the eutectic line has a maximum velocity, as predicted from high-velocity eutectic analysis, and the dendrite line becomes cellular and then planar again at very high velocities.139 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

SOLIDIFICATION

3.65

Temperature

L a+L

b+L b

a Ca

Cb

E Db

Da

Pb

Pa B G Composition (%B)

Temperature (K)

(a)

950 L Al3Fe(b) Al6Fe(g ) 940 TE(ab ) 930 aN + EabEag b + E ab 920 gN + Eab P 910 900 aN + Eag bP + Eag 890 gN + Eag Eag 880 870 860 850 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Concentration (wt. % Fe) (b)

FIGURE 3.44 (a) Schematic phase selection produced from Fig. 3.41 for a given gradient. E: eutectic, P: planar, C: cellular, D: dendritic, B: banded, and G: glass. (b) A calculated phase selection map for Al-Fe system.139 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

FIGURE 3.45 rod eutectic.

Schematic diagram defining a

CHAPTER THREE

3.66

VR 2 =

aR QR

(3.135)

DT 2 = 4 m2 a RQR V

and

(3.136)

According to Walter and Cline, VR2 is a constant for NiAl-Cr rod eutectic. Rodlike eutectics usually form when the volume fraction of one phase is much smaller than the other. Consequently, when the minimum undercooling for rodeutectic formation becomes less than that for the lamellae, the rod eutectic structure will grow preferentially, i.e., when abL ˆ Ê aaL + Á ˜ Ë ma xmb ¯ 4M > 3 2 abR aaR P(1 + x ) + ma xmb

(3.137)

For isotropic surface energies for rods and lamellae, the left-hand side of Eq. (3.137) becomes equal to unity. Table 3.1 shows the value of the right-hand side of this rela-

TABLE 3.1 Comparison of the Stability of Rods and Lamellae†152 1 Sa = Sa + Sb 1 + z 0 0.01 0.02 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

4M Ê 1 ˆ Á ˜ P Ë 1+z ¯

3 2

— (0.8)‡ 0.75 0.74 0.806 0.946 0.994 1.054 1.088 1.116 1.152 1.142 1.374 —

† For isotropic interfacial free energies, rods grow with lower undercooling for values of z such that the number in the right-hand column is less than 1. A lamellar structure grows with lower undercoolings for the cases where it is greater than 1; see Eq. (3.138). ‡ The series defining M and P become difficult to sum for small values of 1/(1 + z). Courtesy of the Institute of Materials, London.

SOLIDIFICATION

3.67

tionship for different values of x. After the evaluation of M and P terms, it is found that the critical volume fraction for the minor phase to form rods is given by152 1 1 Sa = < Sa + Sb 1 + x p

(3.138)

In reality, rodlike structures do not form for volume fractions greater than this value. Lamellar structures, however, are found to occur for volume fractions slightly less than this value, and it is suggested that this results from the generation of low-energy boundaries between the solid (lamellae) phases.145 3.8.1.3 Irregular Eutectic: Faceted-Nonfaceted Eutectics. Faceted-nonfaceted (f-nf), irregular, or abnormal eutectic structures usually form when at least one of the solidified phases is faceted. The best examples of such eutectics are industrially important Al-Si alloys and cast irons which are also discussed in Secs. 3.11.4 and 3.11.6. The main characteristics of irregular eutectics which differentiate them from the regular eutectics are as follows: 1. The growth of two dissimilar phases is cooperative rather than the coupled in which the growing S-L interface is the nonisothermal. 2. The microstructure is irregular due to the nonparallelism of different lamellae arising from growth anisotropy of high entropy faceted phase. 3. It is found in practice that undercoolings and spacings are much larger than those in regular eutectics. This is described by indicating that lamellar eutectic grows, penetrating each other and thereby making the structure coarser. As a result of larger growth undercoolings, the nucleation of new eutectic grains often occurs ahead of the solidifying eutectic interface. These results do not follow the JacksonHunt lamellar theory of growth at the minimum undercooling condition. 4. Unlike regular eutectics, the irregular average eutectic spacing is a function of temperature gradient in the liquid, growth rate, and addition of very small amounts of a third element or impurities (e.g., Na to Al-Si and FeSi and Mg to cast iron). The final spacing is a measure of the ability of the structure to branch or form new phases.59 The growth of each lamella of the faceted phase is determined mostly by its own crystallographic orientation, independent of heat flow direction. Consequently, irregular eutectics solidify with a wide range of spacings growing simultaneously according to the following mechanism (Fig. 3.46). According to Fisher and Kurz,153 Magnin and Kurz,154 and Magnin et al.,155 for solidification of converging lamellae, when l < lmin, the growth of one lamella stops and the surface energy effects hinder further growth. In contrast, for solidification of diverging lamellae, when the local spacing l > critical spacing lbr, the growth becomes unstable and one of them, especially the faceted phase, branches into two diverging lamellae, thereby decreasing the local spacing.155 5. The severity of faceting on the faceted phase is believed to reduce with the increase in growth velocity.27,156 The main characteristic required for the formation of irregular eutectic is the inability of the a-b boundary to change its position smoothly due to the presence of a facet in or adjacent to the boundary. Magnin et al. describe this departure of average eutectic spacing l from the extremum eutectic spacing lex (in the JacksonHunt analysis) and have introduced two dimensionless parameters, average value f and its extent h, to describe the operating range for the irregular eutectic, which are defined as

3.68

CHAPTER THREE

FIGURE 3.46 Growth mechanism of irregular eutectics, as proposed by Fisher and Kurz.153–155 (Courtesy of Acta Metall.)

l = flex h=

lbr - lmin l

(3.139) (3.140)

Since the lamellar spacing varies between lm and lbr, they showed that the average spacing for irregular eutectics will be larger than the average spacing by some operating factor f. For gray cast iron, f = 5.4, and, for unmodified Al-Si alloy, f = 3.2. It appears that no criterion is available as yet to predict the f value.

3.8.2 Monotectic Solidification The monotectic alloys have unique features in that both A and B elements are almost insoluble in each other in the solid state, liquids L1 and L2 have different densities, and the alloy exhibits a liquid miscibility gap which appears to be very wide, as seen in the diagram in Fig. 3.47. Hence, it is very difficult to obtain a homogeneous hypermonotectic alloy melt, and alloys having compositions within the miscibility gap solidify with the associated heavy gravity segregation prior to the start of the monotectic solidification. These problems have restricted the application of monotectic alloys as industrial materials and systematic study on the solidification of monotectic alloys.157 However, a few studies on the solidification of monotectic alloys reveal that the gravity segregation does not occur in alloys with a monotectic composition. It has been reported that in Al-Bi, Al-Pb, Al-In, Cu-Pb, and Bi-Si systems, fibrous or rodlike composite structures form in a similar manner as for eutectic alloys. Monotectic alloys having Al- or Cu-matrix with a fine particulate dispersion and/or fibrous arrangement of the second phases Bi, Pb, and In are considered as important materials for light and soft in situ superconductors, wearresistant materials, etc.

SOLIDIFICATION

FIGURE diagram.

3.69

3.47

Idealized

monotectic

phase

3.8.2.1 Directional Solidification of Monotectic Alloys. Two types of solidification processes have been distinguished during directional solidification of monotectic alloys. In type A solidification (exemplified by Al-In, Al-Pb, and Al-Bi alloys), at low growth velocities (V £ 5 mm/s), well-aligned, regularly arrayed, close-packed fibrous composite structures are observed (Fig. 3.48a). At high growth velocity, the steady-state growth interface is replaced by an interface which is continually perturbed locally, the L2 phase spheroidizing rapidly behind the interface. A transient stage precedes the breakdown or establishment of the steady-state growth in which the monotectic growth interface is prone to oscillatory perturbations over a wide area. This leads to the formation of rows of globular second-phase L2 droplets aligned parallel and transverse to the growth direction in the solid matrix.158 In type B solidification (exemplified by Cu-Pb system), no steady-state growth is noticed at any growth rate. Coarse, interconnected globules of L2 are formed at low velocity, becoming more aligned as growth velocity is increased, wherein loose coupled diffusion occurs between a and L2 (Fig. 3.48b). In this situation, a fibrous structure with heavy branching morphology and termination is noticed above a critical growth rate. The phase spacings in type B growth are approximately an order of magnitude greater than in type A. These morphological transformations of monotectic are strong functions of GL/V ratio, the volume fraction of liquid L2, and the interfacial energies between the solid and two liquids at the monotectic growth front.157 Figure 3.48c depicts the microstructure of a longitudinal section of a monotectic Zn-0.6 at% Bi alloy, showing two columnar grains of different morphology. The central region represents the perfectly aligned morphology, whereas the adjacent grains show cellular morphologies containing more randomly arranged, shorter Bi droplets. These latter morphologies are called monotectic cells, like eutectic cells.158a Cahn extended Chadwick’s hypothesis of relative surface energies of the solid and two liquid phases and pointed out that these types of growth can be related to the height of the miscibility gap. A high TC - Tm value promotes type A growth

(a)

(b)

FIGURE 3.48 (a) Type A monotectic. Steady-state growth in succinonitrile-7.5 wt% glycerol.137,145 (Courtesy A. Hellawell). (b) Type B monotectic. Non-steady-state growth front in succinonitrile 20 wt% ethanol.137,145,153 (Courtesy A. Hellawell). (c) Microstructure of a longitudinal section of monotectic Zn-0.6 at % Bi alloy (grown at a temperature gradient of 30 ¥ 103 K/m and velocity of 17 ¥ 10-6 m/s) showing two columnar grains of different morphology. The central grain denotes the perfectly aligned morphology, while the other grain shows monotectic cells. The selected area diffraction pattern from the aligned grains at zero tilt reveals [0001] zone axis (inset).158a (Courtesy of TMS, Warrendale Pa.)

SOLIDIFICATION

3.71

FIGURE 3.49 Variation with temperature of interfacial energies in monotectic system. (Courtesy of Elsevier Science, Amsterdam.)

FIGURE 3.50 Three phase profiles at monotectic interface. (a) Balanced wetting of a by L1 and L2 (J. W. Cahn, Met. Trans., vol. 10A, 1979, pp. 119–121). (b) Perfect wetting of a by L1.131 (Courtesy of TMS, Warrendale, Pa.)

whereas a low value promotes type B growth. These findings can be described by Cahn’s findings in terms of the relative surface energies between L1 and L2 and a third phase (a) having different temperature dependencies. Figure 3.49 illustrates this point. At TC, gL1L2 = 0 and gaL1 = gaL2. As the temperature falls, gL1L2 increases (as nearly as DT1.2) and gaL1 and gaL2 diverge (as nearly as DT0.35). The stability of a three-phase junction, in terms of the parameter, can be given by Dg = g aL2 - (g aL1 + g L1L2 )

(3.141)

Assuming a critical (or transition) wetting temperature TW (above which the liquid L1 will wet the third phase a) for which Dg = 0, the two situations can be explained as follows. For TM < TW < TC (high-temperature region), a high miscibility gap occurs and L2 perfectly wets a; this results in a stable three-phase junction representing type A solidification. For TW < TM < TC (low-temperature regime), a lower miscibility gap (i.e., large ratios of TM/TC) occurs, and with Dg < 0, L1 preferentially wets a which leads to the separation of L2 away from the growth front (Fig. 3.50). This concept was supported by experimental studies by Grugel and Hellawell159 and Grugel et al.160 who showed that the transition from type A to type B composite structure occurred when the critical value of the TM/TC ratio was around 0.9 (Table 3.2). The fiber spacing in both types of growth was found to follow

CHAPTER THREE

3.72

TABLE 3.2 Summary of Directionally Solidified Monotectic Alloys Arranged in Order of Increasing TM/TC*

System Ga-Pb Sb-Sb2S3 Al-Bi Al-In (CH2CN)2-H2O(S-W) (CH2CN)2-C3H5(OH)3 (S-G) Cu-16 wt% Pb-3 wt% Al (CH2CN)2-7.5 wt% E-6.9 wt% G (CH2CN)2-C2H5OH (S-E) Cu-Pb Cd-Ga C7F14-C3H7OH

Ê Tm ˆ K Ë Tc ¯ ⬃0.5 ⬃0.5 0.59 0.75 0.887 0.896 ⬃0.9 (est.) 0.937 0.94 0.97 0.98 0.857

Type Aligned “A” Aligned “A” Aligned “A” Aligned “A” Aligned “A” Aligned “A” Transition Transition Irregular “B” Irregular “B” Irregular “B” Surface wetting (Schmidt and Moldover)

* A low ratio represents a high miscibility gap and vice versa.160 Courtesy of the TMS, Warrendale, Pa.

the relationship l2V = constant, where l is the interfiber spacing. Since the fiber spacing is about 10 times greater for type B solidification, the constant for this type of growth is about 2 orders of magnitude greater than that for type A growth. Furthermore, the constant for type A growth lies between 3 and 10 times that for comparable eutectic systems.137 There are many common features between these two types of alloys with respect to the morphological transition of the S-L interface and the incorporation of L2 phase into the monotectic solidification front. First, in unidirectional solidification of planar growth front at high GL/V conditions, the liquid separated through the monotectic reaction is not incorporated into the growth front, which leads to the formation of an Al single-phase region without the L2 perhaps in the Al-based alloys and the formation of L2 phase banded segregation in the Cu-Pb alloy. Second, as the growth rate increases, the morphology of the monotectic growth front in both Al-based and Cu-Pb alloys becomes cellular, and the separated L2 phases are incorporated into the intercellular grooves and composite growth of the solid and liquid L2 advances.157

3.8.3 Peritectic Solidification Many binary-phase diagrams display peritectic reactions. Perhaps the most wellknown example is the Fe-C system. Other important examples include binary FeNi, Cu-Zn, Cu-Sn, Cu-Al, and Ti-Al systems and ternary Fe-Ni-Cr systems, tool steels, magnetic materials such as Co-Sm-Cu and Nd-Fe-B alloys, and various inorganic materials such as yBa2Cu3Oy ceramics (YBCO) which are produced via peritectic reactions.138 Figure 3.51a resembles the Fe-C phase diagram and many Cuand Ti-based systems, with the only exception in the latter cases of the formation of intermetallic peritectic phases. In these cases, k0 < 1 in both the primary and peritectic phases. Hypo- and hyperperitectic alloys are those compositions in both the

SOLIDIFICATION

3.73

FIGURE 3.51 Schematic peritectic phase diagrams illustrating compositions of liquid CL and primary phase Ca, which react to form peritectic b phase with composition CP at peritectic temperature TP. (a) k0 < 1 and (b) k0 > 1. (Reprinted by permission of Elsevier Science, Amsterdam.)

TABLE 3.3 Comparison of k0 < 1 and k0 > 1 Peritectic Systems Definition Nonperitectic Peritectic alloys Hypoperitectic

k0 < 1

k0 > 1

Co > CL Ca < Co < CL Ca < Co < CP

Co < CL CL < Co < Ca CL < Co < CP

phase diagrams, at TP, which lie between Ca and CP and CP and CL, respectively. They constitute a and b phases below TP under equilibrium conditions. In contrast, many Al-based systems, notably Al-Ti, exhibit peritectic reactions in both phases for k0 > 1, as shown in Fig. 3.51b. In these systems, the hypoperitectic term for peritectic alloys with composition < CP implies the range from CL to CP. These alloys constitute the liquid and b phases below TP, in contrast to the a and b phases in hypoperitectic alloys with k0 < 1 systems. Similarly, nonperitectic alloys with k0 > 1 systems, which should form the peritectic phase directly from the liquid, are those compositions C0 < CL. These differences in peritectic systems between k0 < 1 and k0 > 1 are given in Table 3.3. Despite their importance, limited studies have been done to comprehend the transformation behavior of peritectic alloys. Three short reviews161–163 and one detailed review138 are described elsewhere. Peritectic growth consists of three stages: the peritectic reaction, peritectic transformation (solid-state diffusion for thickening of b phase), and direct solidification of the peritectic phase (b on the previously formed b layer). These are discussed in this section. 3.8.3.1 Peritectic Reaction. Depending on the surface tension conditions, two types of peritectic reactions can take place: nucleation and growth of b phase in the liquid with or without contacting a phase. In the first case, which is most prevalent, secondary b phase nucleates at the interface between the primary a phase and the liquid. A lateral growth of the b phase over the a phase would require both dissolution and some resolidification of the a phase (Fig. 3.52a).

CHAPTER THREE

3.74

FIGURE 3.52 (a) Peritectic reaction a + L ∫ b with all three phases in contact at the triple junction, showing the growth of the secondary solid phase b along the surface of the primary solid phase a by diffusion through the liquid. (b) Peritectic phase diagram exhibiting liquid compositions which would be in equilibrium with each of a and b phases at undercooling DTP. (c) Peritectic transformation involving diffusion of B atoms through the already formed solid phase b. (d) Schematic representation of the three stages of peritectic transformation: the reaction, the transformation, and direct solidification of b on primary a.

In the second type of reaction, unhindered growth of the secondary phase in the liquid and the simultaneous dissolution of primary phase occur. This type of peritectic reaction has been found in the Al-Mn system. There has also been a tendency for the growth of a secondary phase around the primary phase at larger cooling rates, especially in Ni-Zn and Al-U systems.163 Based on the assumptions that (1) the diffusion of B in the liquid occurs in front of the advancing b phase due to the compositional difference CLb - CLa, (2) the b phase is a platelike surface layer with a tip radius of curvature R which grows at its maximum velocity to provide a thickness 2R, and (3) application of the Bosze and Trivedi model,164 Fredericksson and Nylen165 derived the velocity of plate tip as a function of supersaturation W. For low undercoolings, the b growth velocity is expressed by V=

where

9 DL Ê W Á 2 W W2 8p R Á Ë 1 - p - 2p W=

CLb - CLa CLb - CbL

ˆ ˜ ˜ ¯

2

(3.142)

(3.143)

SOLIDIFICATION

3.75

and CbL is the composition of b in equilibrium with the liquid phase at the growth temperature (Fig. 3.52b). According to Bosze and Trivedi, assuming maximum growth rate and small Peclet number (VR/2DL), the ratio of critical radius for nucleation R* to the actual radius R is given by R* 3 = W R 32

(3.144)

For peritectic reaction, Fredericksson and Nylen predicted the critical thickness of the plate (perhaps twice the critical radius) as R* =

g Vm DGm

(3.145)

where g = gbL + gab - gaL and gbL, gab, and gaL are the surface energies of the b-L, ab, and a-L interfaces, respectively; DGm is the driving force; and Vm is the molar volume of the liquid. This model was employed to compare the theoretical and experimentally found b-layer thickness due to the L-S peritectic reaction for unidirectionally solidified Ag-Sn and Cu-Sn alloys at various rates, which showed only limited success, partly due to difficulties experienced in separating it from other stages.138 3.8.3.2 Peritectic Transformation. The peritectic transformation involves solidstate diffusion and the movement of the a-b interface during cooling (Fig. 3.52c). As shown schematically in Fig. 3.52b and c, the solute diffusion in the solid state is driven by the compositional difference CbL - Cba through the b phase. Using a onedimensional analysis, Hillert161 and St. John and Hogan166 estimated the thickness D of the b layer, for isotropic conditions, represented by the equation

(CbL - Cba )(CLb - CaL ) D2 = Db 2t (CLb - Cb )(Cb - Cab )

(3.146)

where t is the isothermal annealing time; Db is the average interdiffusion coefficient in the b phase; Cba and Cab are the compositions of b phase in equilibrium with the a phase and that of a phase in equilibrium with the b phase, respectively; and Cb is the average composition of the b phase. The peritectic transformation is of great significance when the solute diffusivity is large. For example, the most rapid conventional peritectic transformation in the Fe-C system allows a rapid interstitial carbon diffusion and consequent completion of transformation just a few degrees below TP.167 Based on the isothermal experiments, the time dependence of the thickening of austenite in Fe-C systems follows n = 1/2, consistent with Eq. (3.146),168 and is eclipsed by peritectic transformation. Peritectic reaction in Fe-C system leads to some undesirable effects such as creation of tensile stress (leading particularly to the surface cracking of continuously cast slabs), precipitation of inclusions, segregation of alloying elements, and so forth.168a 3.8.3.3 Direct Solidification of Peritectic Phase. As the temperature decreases below TP, the driving force for both the solid-state transformation of a to b and the solidification of b from the liquid increases. Hence, calculations of the total thickness or percentage of b formed below TP consist of contributions from three stages

3.76

CHAPTER THREE

of peritectic growth. However, Fredricksson and Nylen165 have focused on the contribution of direct solidification in their studies on Ag-Sn and Cu-Sn systems and suggested the following equation for the formation of solid b fraction by direct solidification, which uses the Scheil equation and assumption of complete mixing in the liquid and no solid-state diffusion. CLb = CL0fLk0-1

(3.147) 0 L

where k0 is the distribution coefficient of B in the b phase and C is the average composition of the liquid at the onset of direct solidification, which is nearly equal to the peritectic liquid composition CL. According to their model, the temperature at which b forms depends on the surface energy factor g in Eq. (3.145). A discussion of the agreement between theoretical and experimental values of b phase thickness in Cu-20 wt% Sn and Ag-Sn systems is given elsewhere.138 3.8.3.4 Solidification Growth Kinetics. The breakdown of a steady-state planar front growth of initial composition C0 can be predicted by the constitutional supercooling criterion as GL mC0 (1 - k0 ) DT0 DTplanar = = ≥ V k0 DL DL DL

(3.148)

In peritectic systems, the equilibrium liquidus-solidus temperature interval DT0 for composition C0 can be very small in order to achieve planar front growth. In these cases, alloys with composition C0 < CPa grow as single-phase a without the formation of the b phase. Alloys with compositions C0 > CPb can grow as single b phase without any a phase, if b-phase nuclei form. Initially a phase will be formed, followed by the decrease of interface temperature below TP near the steady state; and finally, if b phase nucleates, it will grow very fast over the a phase. In reality, planar front growth becomes unstable between the limits of constitutional supercooling, i.e., between Vc =

GLDL DT0

(3.149)

DT0v DL kv G

(3.150)

and the limit of absolute stability Va =

where DT0v is the nonequilibrium liquidus-solidus interval, G is the Gibbs-Thomson coefficient, and kv is the nonequilibrium distribution coefficient. The critical velocity for nonequilibrium liquidus slope (mv) and kv to change drastically with velocity is given by the diffusional velocity (or diffusion rate) across the interface VD (of the order of ms-1).138 3.8.3.5 Layered Structure Formation. Several theoretical and experimental studies have been made to gain greater understanding of the formation of layered or banded structure in peritectic systems, directionally solidified at low velocities in which two phases form alternate layers (or bands) that are aligned parallel to the interface. Boettinger169 and Brody and David170 have plotted the conditions of layer formation on the GL/V versus C0 plot, assuming the steady-state solidification condition. A recent model developed by Trivedi171,172 takes into consideration the inter-

SOLIDIFICATION

3.77

action between nucleation and growth for the two phases and non-steady-state conditions for the two phases for the formation of layered structure. By using the equation developed by Tiller et al.61,173 for the initial transient at a given velocity, the widths of each phase layer can be given by lip =

DL L i ln V ki

(3.151)

where lip is the spacing of the ith peritectic phase in the band (i = a or b), ki is the equilibrium distribution coefficient for phase i, and Li is a function of phase diagram parameters and the nucleation temperatures of the a and b phases. Note that, for a specific alloy system, the functions Li are a function of composition only.171 The total band spacing l = lap + lbp, and l=

DL 1k ln L1aka + ln L b b V

(

)

(3.152)

or l µ DL/V, that is, l is proportional to the diffusion length lD. From this model, some important restrictions for banding can be specified. The requisite criteria for the formation of non-layered morphology are given below:138,172 1. Sufficient convection is present because it leads to complete mixing in the liquid, destabilizes the layer formation, and prevents the interface to buildup the necessary boundary layer. 2. C0 < kaCIM or DTNb required for the nucleation of the b phase is large, i.e., below the a solidus temperature for the composition C0. In the latter situation, the plane front a reaches C0 and will grow in a steady-state manner without b forming in the steady-state regime. 3. In the situation that C0 > kbCIM only one band forms, followed by steady-state growth of the plane front b. Hence, there is a narrow range of composition for the formation of peritectic bands (Fig. 3.53), given by the condition kaCIM < C0 < kbCIM

(3.153)

This condition imposed on the phase diagram provides the following necessary criterion for band formation: ka DTMa b - DTNa < kb DTMa b + DTNb

(3.154)

where DTMa/b is the difference of the melting points of a and b. Phase diagrams which do not satisfy this condition will cease to form bands. Note that Eq. (3.154) is based on the nucleation undercoolings for the discrete phases. Hence, it depends on both the system and the nucleation state of the melt in contact with both phases, with the crucible and probable effects of presence of any inclusions or precipitates and convection. This helps to describe the variation in results reported for different systems, as discussed earlier. Clearly, if both nucleation undercoolings are zero, the above inequality gives rise to ka < kb, which is always the case in peritectics. In this situation, bands would not form.138,172

CHAPTER THREE

3.78

FIGURE 3.53 Peritectic phase diagram with banding cycle and concentration window for banding (kaCIM < C0 < kb CIm) which depends strongly on nucleation undercooling.138,172 (Courtesy of International Material Reviews.)

3.9 SEGREGATION Segregation occurs in solidified metals and alloys at two distinct length scales, macroscopic (macrosegregation) and microscopic (microsegregation). In general, segregation is the result of solute rejection at the solidification interface during solidification of a casting. Macrosegregation develops on the scale of entire casting product whereas microsegregation develops on the scale of dendrite. Macrosegregation depends on the type of morphology of the S-L interface: planar or columnar dendrite. Complete elimination of macrosegregation is very difficult.

3.9.1 Macrosegregation Many early attempts to model macrosegregation have centered on the shrinkage effects as the only source of interdendritic liquid flow. Using a continuum model, Xu and Lu examined the shrinkage phenomenon in Cu-Al alloy and proposed that a large pressure gradient is required deep in the mushy zone to replenish liquid to feed associated volume changes. Consequently, shrinkage-induced flows might be

SOLIDIFICATION

3.79

predicted to dominate in the regions next to the solidus. They demonstrated macrosegregation patterns for the completely solidified ingot. Mostly, macrosegregation in a solidified ingot is caused by a mechanism involving physical movement of liquid or solid phases through the interdendritic channels or regions in the L-S zone driven by solutal buoyancy and solidification contraction.174 However, the settling of precipitated phase early in solidification or the deformation of solid skeleton of the mushy zone can also modify solute redistribution on the macroscopic scale. The motion of free liquid, which partially penetrates into the mushy region and presumably gives rise to the formation of segregated channels (freckles), has been simulated by the Boussinesq approximation method.175 Figure 3.54 is a schematic representation of solidification of an ingot. The macrosegregation (or solute redistribution) equation has been derived based on the following assumptions: (1) liquid composition is uniform in a sufficiently small volume element, which is considered as differentiable; (2) to perform a mass balance, solute enters or leaves the volume element only by liquid flow to feed shrinkage; (3) the liquid and solid phases have different densities; (4) there is complete diffusion of solute in the liquid, but no solid diffusion; (5) there is no solid movement; and (6) there is an absence of voids. The character of (longitudinal) macrosegregation can be expressed quantitatively in three manners using necessary thermal relations and the Scheil equation [Eq. (3.55) or (3.56)].7,176 The L-S region (or mushy zone) is treated as permeable (or porous media) to interdendritic liquid flow, as shown by the schematic volume element in Fig. 3.55. Using D’Arcy’s law, the interdendritic liquid flow through the element with a fluid velocity v is given by v=-

Kp (—P + r L g ) hfL

(3.155)

where Kp is the permeability of a mushy zone; h, fL, and rL are the viscosity, volume fraction, and density, respectively, of the interdendritic liquid; —P is the pressure gradient; and g is the gravitational acceleration.

FIGURE 3.54 Sketch of an ingot solidifying in a metal mold (shaded) with a refractory “hot top.” The center of the ingot (dotted) is fully liquid; the outer portion (white) is fully solid; and a semisolid region exists between the two.73 (Reprinted by permission of VCH, Weinheim.)

3.80

CHAPTER THREE

FIGURE 3.55 Fluid flow through a solidifying “volume element.”73 (Reprinted by permission of VCH, Weinheim.)

The solute flow may be gravity-driven, acting on a fluid of varying density within the L-S zone, solidification shrinkage-driven, convection-driven in the bulk liquid, or electromagnetic or centrifugal acceleration-driven. It may also be solid movement- or deformation-driven, as in bulging in continuous casting.73 Since the liquid composition varies spatially within the L-S zone, any interdendritic flow except parallel to the isotherm must change the composition of the volume element. Conservation of solute leads to the modification of the Scheil equation into a new local solute redistribution equation, given by175–177 ∂ fL 1-b Ê v—T ˆ Ê fL ˆ =1+ ∂ CL e ¯ Ë CL ¯ 1 - k0 Ë

(3.156)

where k0 is the distribution coefficient, fL is the volume fraction of liquid, b = (rS rL)/rS = solidification shrinkage (or contraction), v is the velocity of interdendritic liquid flow, —T is the local temperature gradient vector, and e is the local rate of temperature change. This expression considers steady-state solidification with planar isotherm moving with velocity V in the x direction, which becomes ∂ fL 1-b Ê vx Ê f L ˆ =1+ ˆ ∂ CL V ¯ Ë CL ¯ 1 - k0 Ë

(3.157)

where vx is the isotherm velocity normal to isotherms. Equation (3.157) converts to the Scheil equation when both vx and b equal zero. It also reduces to the Scheil equation (written in terms of volume fraction) if the interdendritic flow velocity just equals the flow required to feed solidification shrinkage vx = -

b V 1-b

(3.158)

SOLIDIFICATION

3.81

In the steady-state solidification, macrosegregation is not observed. When flow is in the same direction and greater than (down the temperature gradient form) that of Eq. (3.158), it causes a negative segregation. When the speed is lower and in the opposite direction, it clearly produces a positive macrosegregation (for normal alloys where b > 0 and k0 < 1). This is usually found in ingots and is called inverse segregation,178 since it is reversed from what we would predict based on the initial transient of plane front growth.4 Figure 3.56 shows a schematic illustration of the application of the above principles to continuous casting. When interdendritic flow lines remain all vertical, no segregation results. When flow lines resemble those of Fig. 3.56b and c, respectively, negative segregation and positive segregation result. In the latter case, greater downward flow toward centerline promotes centerline segregation73 (see also the next section for segregation in continuous casting). Normal Segregation. Normal segregation, also called positive segregation, results in the rejection of a high concentration of the low-melting-point composition at the S-L interface and its accumulation (for k < 1) or lack (in the case of k > 1) at the last or central portion of the ingot or casting (Fig. 3.57a). Normal segregation is

FIGURE 3.56 Interdendritic fluid flow in a continuous casting: (a) no segregation, (b) negative segregation at midradius, and (c) positive segregation at centerline.7 (Reprinted by permission of McGraw-Hill, New York.)

FIGURE 3.57 Normal segregation and inverse segregation of an alloy ingot.

3.82

CHAPTER THREE

often found in steel ingots, where higher concentrations of P and S are observed in the central region than in the outer one. Average concentrations are used to explain normal segregation, and microsegregation effects are assumed to be negligible and are ignored.63 The solute distribution for k < 1 is predicted to conform broadly to one or the other of the types shown in Fig. 3.19b and c. The amount of segregation will be greater in the case of an alloy with a larger segregation coefficient value; however, normal segregation tends to lessen if the separation of crystals from the mold wall takes place by thermal convection or other crystal detachment mechanisms.9 The principles of normal segregation are used in the zone refining of impure metals, which has been discussed previously.69,82 Inverse Segregation. In the inverse segregation, lower-melting-point composition is usually observed in the outer portion of the ingot, and the higher-melting-point region lies in the central region of the ingot (Fig. 3.57b). Alloys with a wide freezing range are especially prone to this type of segregation. A typical example is the zone of low-concentration negative segregation of impurities that often exists in the lower central portion of the steel ingots. In Al-Cu alloy castings, the outer surface contains low-melting (eutectic) alloy, and the central one the primary crystals of high melting point, which constitute the inverse segregation. Hence it is customary to remove about a 1-mm-thick surface by machining prior to subjecting to the subsequent plastic deformation. In Cu-10% Sn alloy casting, exudations of 20 to 25% Sn are often observed; these exudations are known as tin sweat. The occurrence of these exudations or so-called inverse segregation results from the entrapping of low-melting-point composition of the interdendritic liquid during solidification. Its extent is a function of solidification time and the amount of contraction-stimulated flow.63 As solidification progresses, the solid shell contracts and pulls away from the mold wall. Then the enriched low-melting interdendritic liquid in the center of the casting is captured by the roots of the crystals on the mold wall and is forced to the contracted region, where it begins to solidify.9,179 Banding Segregation. Banded microstructures consist of alternate structures or phases which develop mostly parallel to the transformation front in the unidirectional solidified alloy ingot (in contrast to the traditional solidified structures such as dendrites and eutectics which form perpendicular to the S-L interface). They form when the solute-rich liquid concentrated at the advancing S-L interface becomes unable to escape and is trapped among dendritic side arms of the crystals (because of the presence of greater undercooling by the rejected solutes). Usually they appear as two different bands which alternate in time. Different types and mechanisms of banding segregation are schematically represented in Fig. 3.58.172 Banding phenomenon has been found (1) at relatively low speeds, in welded or laser-treated materials; (2) in directionally solidified peritectic Ag-Zn, Sn-Sb, Sn-Cd, Zn-Cu,Ti-Al, and Ni-Al alloys; (3) in off-eutectic or off-monotectic compositions;138a and (4) in rapidly solidified Ag-Cu,Al-Cu,Al-Fe,Al-Pd, and Al-Zr alloys.172 In eutectic cast irons, such a banding segregation occurs after passing gas bubbles through the melt. This is, however, eliminated by preventing any source of gas in the mold or within the flux used (see also Sec. 3.8.3.5). Banding segregation in centrifugal castings occurs only with larger wall thickness (in excess of 2 to 3 in., or 50 to 75 mm) and not in thinner wall castings. They are characterized by a hard demarcation line at the outer edge of the band that usually disappears into the base metal of the casting. Most alloys susceptible to banding have wider solidification range and greater solidification shrinkage. It is believed to result from vibration, variations in gravitational force between the top

SOLIDIFICATION

3.83

FIGURE 3.58 Schematic representation of different types and mechanisms of banding segregation: (a) different scales of same structure and phase a; (b) fluctuations of composition in plane front growth of same phase (usually do not form visible bands); (c) different microstructures of same phase; (d) different phases (a, b) with plane front growth morphology; and (e) different phases and microstructures (c = cells, p = plane front).172 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

and bottom of the mold, and irregularities in the flow of the liquid metal during the entrance into the rotating mold.180 Gravity Segregation. Usually this occurs during early stages of solidification, before, or just after the initial growth nuclei have formed. It is caused by the density variation between the solid and liquid phases or between two nonmixing liquid phases (exhibiting miscibility gap) which leads to differential movements within the

3.84

CHAPTER THREE

liquid. For example, Cu and Pb exist as two layers in the molten state, and unless they are stirred thoroughly during freezing, the lower-density Cu-rich layer remains on top of a Cu-Pb portion. Similarly, during the zone leveling of Ga-Si alloys, the Ga-rich portion settles at the bottom of a horizontal ingot whereas the Si-rich portion is at the top. Gravity segregation in polyphase systems is of even greater significance and has been widely found in bearing alloys. During the solidification of the Sn-base bearing alloys containing Cu and Sb, two primary intermetallic phases form; for example, in isolation, cuboids of SbSn (lower density) float on the surface and needles of Cu6Sn5 (higher density) sink. In combination, they form an entangled network and remain unsegregated. Negative cone segregation often occurring in steel ingots has been presumed to be formed by the settling of free primary dendrites or melted-off dendrites into the bottom of a steel casting, because of its density higher than the liquid (see the next section). When an Al-Si alloy with hypoeutectic composition is cooled rapidly, only hypoeutectic primary crystals form throughout the entire casting; but when it is cooled slowly, both hypo- and hypereutectic sides of the alloy appear separately in the upper and lower part of the casting. The extent of gravity segregation depends on the amount of Fe, Mn, and Si and the thermal condition of the melt. Gravity segregation disappears when a melt with the recommended Fe, Mn, and Cr compositions is thoroughly stirred and mixed after heating to at least 720°C.181 The reasons for this segregation have been attributed to (1) flotation of the primary Si particles to the top of the melt, (2) the nucleation of primary Si, especially at the sidewalls and the bottom of the molds, and (3) the localized growth of such primary crystals at the expense of the nucleation and growth of the primary Si in the bulk of the melt.182 Figure 3.59 shows the segregation after the solidification of Sn-Pb alloy. Likewise, during solidification of hypereutectic Al-Si alloys, clusters of primary Si particles have been found: (1) in slowly solidified castings, including block and bar, sand castings, and step castings in sand and graphite molds; (2) by rapid solidification such as in wedge castings, and in laser-treated surfaces; and (3) during isothermal holding in the freezing range. The sink rate is controlled by a form of Stokes’ equation which states that a Stokes or terminal velocity is the maximum velocity achieved by a solid particle of radius r falling through a convection-free liquid and corresponds to the point at which the relative weight of the particle just balances the viscous drag by the fluid; this is given by the relation V=

gr 2 ( r S - r L ) 9h

(3.159)

where rS and rL are the densities of the solid and liquid, respectively. It is influenced by the shape of the particle, forced convective stirring, and centrifugal effect. If lighter solids such as nonmetallic inclusions and kish or spheroidal graphite are formed early in solidification, they can float to the upper part of the casting, resulting in positive segregation areas. In centrifugal casting, centrifugal force stimulates gravity and can result in the compositional variations between the internal and external parts of the casting. It is pointed out that gravity segregation does not occur in single-phase liquid. Channel Segregation. Examples of channel segregation include “inverse-V” segregation; “A” segregate channels in large-scale steel billets and ingots with mostly

SOLIDIFICATION

FIGURE 3.59

3.85

Gravity segregation after the solidification of Sn-Pb alloy.179 (Courtesy of A. Ohno.)

horizontal heat flow; and “freckles” formation in unidirectional solidified Al-Mg and Al-Mg-Cu, Pb-Sn, Pb-Sb, or Pb-Sn-Sb alloys and superalloys. This is caused by convection due to the density variation between the bulk liquid and solute-rich interdendritic liquid. Channel-type segregation occurs when the solidification rate is below a critical value Vc, which is given by Vc = c

Ê Dr L ˆ Ë DTL ¯

0.6

(Sin q )1.3

(3.160)

where c is a constant (= 1.8 ¥ 10-4 for Al alloys and 0.5 ¥ 10-4 for steel ingots), DrL is the density variation in the interdendritic liquid for a temperature change DTL, and q is the inclination angle of the solidification interface to the horizontal plane.183

3.86

CHAPTER THREE

Freckles are formed mostly on the casting surface in single-crystal or directionally solidified castings; however, some may be found inside the castings. Metallurgical examination reveals that freckles consist of chains of small equiaxed grains and eutectic constituents.183a Particularly, in directionally solidified single-crystal superalloy parts, freckles are usually a reason for rejection.183b The Lutwig-Soret Effect. This type of segregation is an interesting behavior in which homogeneous melt of, say, Zn-Sn, Cu-Sn, and Pb-Sn alloys of uniform composition, held in a temperature gradient, becomes nonuniform in composition34 because solute transport related to temperature occurs in one direction and chemical diffusion in the opposite direction. The Soret coefficient is equal to D¢/D, where D¢ and D are thermal and chemical diffusion coefficients, respectively.17 That is, Sn always migrates to the higher-temperature end. A large concentration gradient could be generated in this manner, e.g., for a 36% Pb in Sn alloy the excess Pb content at the cooler end was 5.28%. Again, convection must result in a remarkable divergence from ideal behavior.63

3.9.2 Segregation Patterns in Steel Ingots Ideally, an ingot with a homogeneous composition is desired, but during the solidification of steel, solute elements such as C, Mn, S, and P become concentrated in the liquid ahead of the advancing dendrites, resulting in microsegregation between dendrites, called interdendritic microsegregation (due to settling of free-floating solid or flow of solute-rich liquid into or out of the volume elements during solidification) and macrosegregation for dendritically solidified materials over large dimensions typically ranging from millimeters to the size of the entire ingot casting. Macrosegregation is of great concern because it may carry through the defect into the final product. In large killed steel ingots, three major types of macrosegregation such as A segregation, V segregation, and negative segregation are commonly found.184 Figure 3.60 is a schematic representation of structure and segregation pattern in a killed big-end-up ingot. The factors that are responsible for the formation of vertical macroscopic segregate channels called A segregates are the extent of solute enrichment, the dependence of steel density on solute concentration, and the morphology of the dendritic network. They are formed in the same way as freckles. V bands, probably occurring during disturbances or fluid flow in the interdendritic region, may also be observed running parallel to the solidification front. Since the V segregation forms in the center of the ingot and during the final stage of solidification process, it possesses the largest degree of solute (such as S, P, and C) enrichment. The inverse V segregation in steel is usually observed as string-shaped.185 The negative cone of segregation (implying lower concentration) normally appears as a zone in the bottom third of the ingot, and is associated with the pile of dendritic debris. This region also contains a relatively high concentration of inclusions, probably trapped by the equiaxed crystals in the core bottom of the ingot. Other zones of positive segregation (implying a higher alloying content) also occur near the centerline, primarily at the top of the ingot. In fact, all three types of segregation arise from the unsolidified pool and are influenced by the ingot structure (i.e., the relative amount of the columnar and equiaxed zones). Both positive and negative segregates can form by various mechanisms, the most common being interdendritic flow due to solidification shrinkage or density difference, generating a natural convection during the solidification process.185 It is pointed out that any

SOLIDIFICATION

3.87

Shrinkage cavity

A-segregates V-segregates Bands Equiaxed Zone Columnar Zone Chill Zone Bottom Cone

Structure (Detail of equiaxed zone not shown)

Cone of Negative Segregation

Macrosegregation

(a)

(b)

FIGURE 3.60 Schematic representation of (a) structure and (b) segregation pattern in a killed, big-end-up ingot. (Reprinted by permission of Pergamon Press, Oxford.)

factors that influence structure and fluid flow must also affect the segregation in ingots. More segregation is observed in rimmed ingots than in killed ingots. The rim zone displays negative segregation whereas positive segregation is measured at the center.184

3.9.3 Segregation Patterns in Continuous Steel Casting Continuously cast billets solidify mainly in a dendritic structure that can lead to heavy centerline segregation and cavities. However, axial segregation, center unsoundness, and V segregates are less predominant in continuously cast steel. The axial segregation (Fig. 3.61a) and center unsoundness in billets are often found periodically along the strand that may pose serious quality problems. The factors responsible for severe axial segregation and porosity are minimal equiaxed structure, large superheat, predominant columnar structure, no electromagnetic stirring (EMS) application in the upper part of the strands, increased section size, higher

CHAPTER THREE

3.88

(a)

(b)

FIGURE 3.61 Sulfur prints in continuous casting showing axial or longitudinal segregation in (a) a billet and (b) a slab. Notice that segregation in the billet is periodic. (After Alberny and Birat, 1976, The Metals Society, London. Reproduced with permission.)

aspect ratio, and carbon content. The carbon content influences both the width of the central segregation zone and the size of the axial voids: 0.3% < 0.1% < 0.6% C.186 For very high carbon contents, and in the case of Cr-bearing grades, the centerline consists of a tubular segregated region. The severity of V segregates increases with carbon content. The mechanism of formation of V segregates involves the settling of dendrites, volume shrinkage, and interdendritic flow in the very long mushy zone. The axial segregation typically observed in slabs is shown in Fig. 3.61b. The main factors responsible for segregation are the stability of the strand, bulging of the slab, and superheat of the steel. 3.9.4 Microsegregation Microsegregation is the segregation of solute elements over distances on the order of cellular spacing, dendritic arm spacing, and grain boundaries. Microsegregation due to the nonequilibrium solidification and consequent solute redistribution causes nonequilibrium second phases, porosity, and crack formation.187 Microsegregation is observed when the solute diffusivity in the solid alloy is too low to homogenize this second phase during its growth.188 For melt alloy systems with k0 < 1, the severity of segregation increases with the decrease in the k0 value.189 The extent of microsegregates in the alloy structure is determined experimentally by measuring (1) the amount of nonequilibrium eutectic, (2) the amount of nonequilibrium second phases, (3) minimum solid composition, (4) ratio of minimum to maximum composition of primary phase, or (5) composition versus fraction solid profile.190 Microsegregation, due to time-dependent variations of the interface, convection velocity, or unsteady convection in the melt, is often the major problem for the crystal growth.191 For example, rotation of the sample in an asymmetrical thermal field, pulling device instability or vibration, or temperature fluctuation causes timedependent variations of the interface, thereby leading to striations. Another possible mechanism for microsegregation is associated with the unsteady convection flow in the melt.192 It can serve either directly on the solutal field, as in the case of g-jitter during solidification in microgravity, or indirectly when the temperature fluctuations

SOLIDIFICATION

3.89

linked with the flow modify the growth velocity and thereby the composition of the alloy.193 Microsegregation varies significantly with the history of the growth of the solid. For example, microsegregation frequently increases with the cooling rate for equiaxed dendritic solidification; however, it decreases for unidirectional solidification. Microsegregation can also change considerably if a phase transformation occurs during solidification due to the variation of k0 values with phase.194 In binary Al-4.5 wt% Cu alloy, the intercellular fluid flow has a small, but perceptible effect on microsegregation and cell morphology.188 During solidification of the melt when the uniform solute distribution is usually lost, it gives rise to a nonuniform cored solute distribution over distances of dendrite arm spacing, a spacing dependent on cooling rate. An extreme case of this segregation is the formation of insoluble inclusions with similar spacing. Although some nonuniformity can be eliminated by heat treatment after solidification, much segregation still persists and may cause a reduction in mechanical properties.189 The reason is that, in the former case, the liquid composition is uniform in the interdendritic liquid whereas in the latter case there is a solute built up on the dendrite tip. Cellular Microsegregation. During the early stages of growth of cellular solidification in single-phase alloys, low degrees of constitutional supercooling (CS) develop; the liquid near the advancing interface is richer in solute for k0 < 1. This difference in solute concentration from the cell center to the cell boundaries is called cellular segregation, which is taken as the result of the cell thickening process and extends over distances on the order of the cell size of ⬃5 ¥ 10-3 cm. Once the rounded cell projection is stabilized, solute will be rejected (k0 < 1) from the sides of the projections as well as from the top, and the accumulation of solute will occur in the cell boundaries. The most severe segregation will be found at the junction points (nodes) in the hexagonal array. The situation then appears as shown in Fig. 3.62. For k0 > 1, the cell boundary regions are depleted of solute.

VOLUME ELEMENT

VOLUME ELEMENT

x

SOLID

y

LIQUID

S

ᐉ C*

CE

L

DCL (MAX.)

CL

–

CL

–

kC*L

Ct CO xE

xt

x (a)

y¢

y (b)

FIGURE 3.62 Cellular redistribution (microsegregation) during cellular growth. (a) Cellular growth and solute distribution in the growth direction. (b) An enlarged “volume element” and solute distribution transverse to the growth direction.195 (Reprinted by permission of Pergamon Press, Oxford.)

CHAPTER THREE

3.90

Quantitative studies have shown that the solute concentrations at the cell boundary nodes could be up to two orders of magnitude greater than the melt average concentration. Annealing times of the order of 1 hr should be adequate to remove cellular segregation, assuming the solid-state diffusivity of ⬃10-8 cm2/s and a cell size of ⬃5 ¥ 10-3 cm. Ma and Sahm195 have extended the Bower et al. model by taking into consideration the effects of cell geometry on the solute rejection in both longitudinal and lateral directions and, therefore, on the final segregation profile. It is suggested that the forward solute rejection at the advancing S-L interface could result in a reduction of solute enrichment in the intercellular region, and, thereby, a flatter segregation profile of alloying elements is anticipated. The application of the modified model holds well with the experimental data.195 Dendritic Microsegregation. In the interdendritic region, first solid to form has a concentration C0. Figure 3.63 shows the microsegregation in Sn-10 wt% Pb alloys according to equation

(CL - CS ) df S* = (1 - fS ) dC

(3.161)

k0-1

where CL [= C0(1 - fS) ] and CS are concentrations in the liquid and solid at the interface, respectively, in weight percent and fS is the weight fraction of solid. When CL = CE and 1 - fS = fE, the liquid solidifies to eutectic mixture with a concentration CE.195 The microsegregation resulting from solute redistribution during dendritic solidification leads to coring in the primary phase, which is a variation in the solute concentration between the center and the outside of dendrite arms (Fig. 3.64). In extreme instances the accumulation of solute between the growing dendrite arms can result in the formation of second phases in the interdendritic region in amounts significantly higher than those predicted from the equilibrium diagram.195a Brody and Flemings first considered the diffusion in the solid during dendritic solidification using the Scheil nonequilibrium solidification equation. In the simplest form, the equation for solute redistribution at the S-L interface during linear dendritic growth is fS ˆ C*S = k0C0 Ê 1 Ë 1 + ak ¯

k 0 -1

(3.162)

FIGURE 3.63 Dendritic microsegregation in Sn-10 wt% Pb alloys.195 (Reprinted by permission of Pergamon Press, Oxford.)

SOLIDIFICATION

3.91

400 mm (b)

100 mm (c)

(a)

FIGURE 3.64 (a) Coring or microsegregation during directionally solidified Ni-60% Cu alloy. The centers of the dendrite are deficient in copper which segregates to the dark etching regions between the dendrite arms that are the last portions to solidify. (b, c) Dendritic microsegregation at the midradius location of the longitudinal cross section of the as-cast 718 (Ni-18.5Fe-18Cr-5.3Nb-3Mo-0.9Ti0.5Al-0.03C) electroslag remelting (ESR) ingot. (b) The low-magnification microstructure reveals a dendritic solidification pattern with numerous primary dendrite spines and secondary arms. (c) The high-magnification microstructure reveals the enriched (in Nb, Mo, and Ti) interdendritic regions, which reflect the last metal to solidify.195a (a: Reprinted by permission of Pergamon Press, Oxford, Plc; b, c: Reprinted by permission of TMS, Warrendale Pa.)

and for parabolic growth of dendrite (k0 -1) (1- 2 ak0 )

C*S = k0C0 [1 - (1 - 2ak0 ) fS ]

where CS* is the solute concentration in the solid at the S-L interface, k0 the equilibrium partition coefficient, C0 the alloy composition, fS the weight fraction of solid, and a the diffusion Fourier number, defined as DStf/l2, where DS is the solute diffusivity in the solid, tf the local solidification time, and l one-half of the characteristic dendritic arm spacing (DAS).91,196 When ak0 > 1, there is almost complete homogenization, for example, for the carbon distribution in Fe-C alloys, where very fast interstitial diffusion of carbon occurs. Figure 3.65 illustrates the complex nature of isoconcentration profiles for a low-alloy steel columnar dendrite after complete solidification. Table 3.4 lists the microsegregation ratio k0 (i.e., ratio of maximum solute concentration to the minimum solute concentration after solidification) found at 1.7, 2.5, and 5.75 in. from the chill, for Mn and Ni. However, the real microsegregation ratio is not easily estimated. Note that the microsegregation analysis also predicts the amount of nonequilibrium eutectic or other secondary phases.4 Both these parameters are required to evaluate the extent of microsegregation. Microsegregation is more severe across and between primary dendrite arm spacings than secondary dendrite arm spacings. The microsegregation ratio is largest at low alloy contents, although there is no basic difference in the solidification mode. Both columnar and equiaxed dendrites display essentially the same microsegregation feature.135 The DAS defines the range of dendritic microsegregation. The smaller the DAS, the easier to homogenization by subsequent heat-treatment. For a given alloy heattreated at a fixed temperature, the homogenization time varies as the square of the DAS. For coarse DAS (⬃10-2 cm) in steels, heat treatment of ⬃300 hr at 1200°C is required (in the absence of simultaneous working) to produce any appreciable

CHAPTER THREE

3.92

FIGURE 3.65 Isoconcentration profiles for a low-alloy steel columnar dendrite.135 (Reprinted by permission of McGraw-Hill, New York.)

TABLE 3.4 Electron-Microprobe Measurements of Microsegregation135

Element Mn Mn Mn Ni Ni Ni

Distance from chill, in.

Solute concentration (max)

Solute concentration (min)

ko

1.7 2.5 5.75 1.7 2.5 5.75

0.68 0.88 0.83 2.33 2.11 2.16

0.45 0.62 0.46 1.92 1.85 1.61

1.52 1.39 1.79 1.21 1.14 1.34

reduction in dendritic microsegregation. Simultaneous working improves the situation only slightly. Refining of DAS by increasing cooling rate can therefore have extremely advantageous effects. Complete removal of dendritic microsegregation gives rise to improved mechanical properties. This is especially difficult if second-phase particles are formed. For example, Turkdogan and Grange197 observed the formation of second-phase sulfide inclusions in the solute-rich interdendritic regions in steel toward the final stage of freezing. These inclusions were extremely stable, inhibited grain refinement, and along with the solute microsegregation were believed to be responsible for subsequent banding of the wrought steel. The importance of dendritic microsegregation to the occurrence of banding or fibering had been previously discussed by Flemings.198

SOLIDIFICATION

3.93

Grain Boundary Segregation. Grain boundary (GB) segregation during solidification originates from two sources. First, if the GB lies parallel to the growth direction, surface energy requirements lead to a GB groove, where the boundary meets the interface. The groove is typically about 10-3 cm deep.199,200 In the presence of CS, conditions are energetically favorable201 for considerable segregation to GB groove. The second type of GB segregation occurs due to the impingement of two interfaces moving with a growth component perpendicular to each other which may be regarded as a form of microsegregation.73

3.10 SOLIDIFICATION PROCESSES AND CAST STRUCTURES This section deals with three traditional solidification processes: ingot casting, continuous casting, and welding; however, the last two processes hold a key importance in today’s technology. Almost entire molten metal is cast into solid state by either the traditional ingot casting or relatively modern continuous casting. The development of the ingot and continuously cast structure as well as fusion welding structure and weld cracking are briefly described.59

3.10.1 Ingot Casting Ingots are cast shapes produced with a cross section that is suitable for rolling, extrusion, and forging operations. They are commonly used in steel and copper industries. In ingot casting, steel is poured into a top or bottom of cast iron molds. The molds are often equipped with “hot tops” comprising insulating boards and exothermic compounds to greatly reduce the depth of the shrinkage cavity formed during the solidification of ingot (Fig. 3.54). After a required period, the molds are removed from ingots and charged into soaking pits for rolling into finished products.101 Unlike continuously cast semifinished products, ingots have larger thickness in at least one transverse direction, the depth of liquid core is small, solidification occurs in a stationary space, and time is measured from the viewpoint of a fixed observer.184 3.10.1.1 Classification of Steel Ingot. Steel ingots can be classified into four types according to the deoxidation practice used or, alternatively, by the amount of gas evolved during solidification (Fig. 3.66). The gas is generated primarily by the reaction between carbon and oxygen dissolved in the steel, which is favored thermodynamically at lower temperatures.184 These types are called killed, semikilled, capped, and rimmed steels. If practically no gas is evolved, the steel is termed killed because it lies quietly in the molds. Increasing extents of gas evolution result in semikilled, capped, or rimmed steels.202,203 Killed steel ingots are fully deoxidized so that there is only a slight or practically no evolution of gas during solidification. These ingots are normally cast in “hottopped, big-end-up molds” to reduce the depth of the shrinkage cavity. The top of the ingot solidifies much faster than in semikilled or rimmed steel. Killed steel is characterized by a homogeneous structure (free of blowholes and segregation), even distribution of chemical composition and properties, and formation of pipe in the upper central portion of the ingot, which is later cut off and discarded. All steels having more than 0.3% carbon are killed. Alloy steels, forging steels, and carburiz-

CHAPTER THREE

3.94

1

2

Killed

Semikilled

3

4

5

Capped

6

7

8

Rimmed

FIGURE 3.66 Classification of steel ingot according to the deoxidation practice used. These commercial ingots are cast in identical bottle-top molds, where the degree of suppression of gas evolution ranges from maximum for completely killed ingot (No. 1) to that of a minimum for violently rimmed ingot (No. 8).204

ing grades of steels are usually killed, where the essential emphasis is placed on soundness and homogeneity of structure.184,204–206 Killed steel is produced by various steel melting practices involving the use of certain deoxidizing elements which act with varying intensities. The most common deoxidizing elements are Al and ferroalloys of Mn and Si; however, calcium silicide and other special strong deoxidizers such as V, Ti, and Zr are sometimes used. Deoxidation practices in the manufacture of killed steels are generally left to the discretion of the producer. Semikilled steel ingots are partially deoxidized so that gas evolution is not completely suppressed during solidification by deoxidizing additions. There is a greater degree of gas evolution than in the killed steel, but less than in the capped or rimmed steel. An ingot skin of considerable thickness is formed prior to the beginning of gas evolution. A correctly deoxidized semikilled steel ingot does not have a pipe, but does have well-scattered large blowholes in the top-center half of the ingot; however, they weld shut during rolling of the ingot. Semikilled steels, generally, have a carbon content in the range of 0.15 to 0.30%; they are commonly used in structural shapes, plates, merchant bars, skelp, and pipe applications. The main features of semikilled steels are (1) variable degrees of uniformity in composition, which are intermediate between those of killed and rimmed steels, and less segregation than rimmed steel and (2) pronounced tendency for positive chemical segregation at the top center of the ingot (Fig. 3.66). Rimmed steel ingots are characterized by only a small amount of deoxidation and a greater degree of gas evolution during solidification in the mold and a marked difference in chemical composition across the section and from top to bottom of the ingot (Fig. 3.66). This results in the formation of an outer ingot skin or rim of relatively pure iron (hence the name rimming steel) and an inner liquid (core) portion of the ingot with higher segregation/concentration of alloying elements, especially carbon, nitrogen, sulfur, and phosphorus, having lower melting temperature. The high-purity zone in the surface is preserved during rolling.21 Most low-carbon steels (0.4%) steels.231 This nonuniformity, which restricts heat transfer to the mold, is most likely connected with the d Æ g phase transformations. Better roll containment, decrease in casting speed, or increase of spray water also overcomes these problems. In continuous casting, the production of good-quality billets for low-carbon steels depends on a very low oxygen activity in the mold in order to avoid pinhole occurrence, and low total oxygen contents to minimize the amount of inclusions. This last requirement becomes important in the case of the billet casters with small radius where the inclusions accumulate near the upper side of the billets.232 Alloys made of Ca, Mg, and rare earth elements are used to remove, at least partly, nonmetallic trace elements in steels and to control the composition and morphology of residual inclusions.233 Inclusion modification by CaSi injection (1) prevents the formation of Al2O3 inclusions and the resulting nozzle clogging and (2) produces cleaner steel with very good surface, improved mechanical properties, and anisotropy.234 In the Al-killed continuous-cast steels, Ca treatment provides a steel with good fluidity if a critical ratio of Ca/Al is achieved; prevention of nozzle clogging; possibility of longer sequence; reduction in product rejects; and effective control of oxides and sulfides, particularly in low-silicon steel and in construction steel grades such as electroslag refining (ESR) quality.235 Selection of appropriate mold powder such as TIS ULC flux containing 46% (SiO2 + Al2O3), 40.2% (CaO + MgO + SrO), 1.2% (Na2O + K2O + Li2O), 2.3% (MnO + Fe2O3), 8.5% F, and 1.6% C (with higher surface tension and viscosity and low solidification temperature) has been reported to have considerable promise in reducing the entrainment and entrapment of the in-mold slag and achieving a higher surface quality of the cast slab and lower sliver occurrence in the coil.236

SOLIDIFICATION

3.101

In a horizontally cast strand, crystal formations sink down whereas the upper part of the strand still solidifies with a dendritic structure. To avoid this, special stirring has to be applied to increase the initial homogeneity of the billet and to prevent deviation of the metallurgical center. Through the use of special stirring, a highquality strand is attainable. Thus both surface and internal quality improvements are attainable with an optimum horizontal continuous casting process design.237 3.10.2.2 Continuous Casting of Nonferrous Alloys. In continuous casting of nonferrous alloys, the molten metal is continuously poured within formed molds (rolls, belts, or a wheel and belt) which are constantly in motion and solidify as it moves along the water-cooled copper molds (Fig. 3.69b). The advantages of continuous casting include improved surface quality of the cast bar and more uniform structure. Rolls and belts are employed mainly in the casting of sheets, plates, foils, and strips for various thicknesses. In general, nonferrous bars and rods for the production of wire are cast using wheel belt, dip forming, and upcasting methods.226 The main advantages of a moldless vertical continuous casting process for the production of Al, Al-Cu, and Al-Si rods are (1) near-net shape material with small and complex cross-sectional shape; (2) full automation preventing breakout of molten metal and allowing easy start-up and easy shutdown; (3) unidirectionally solidified cast structures; and (4) geometries with variable cross-sectional configuration along their axes. An attractive casting process using a heated rather than a cooled mold has been developed by Ohno for single-crystal growth or unidirectional solidification of ingots, rods, and wires of unlimited length.4 Ohno Continuous Casting (OCC) Process. Figure 3.70 shows the schematic sketch of the integrated Ohno continuous casting (OCC) process for crystal growth and casting of alloy wires ⬃1.7 mm in diameter. Essentially, it consists of stainless steel or graphite crucible, a graphite mold, a cooling device, and pinch rolls for withdrawal of the cast product.238 The process is based on the fact that the mold is heated just above solidification temperature of the metals to be cast to prevent the formation of new crystals on the mold surface. The cooling device is confined to a short distance away from the mold exit where solidification starts just before entering the cooling water, producing unidirectional cast structure. The process variables required in single-crystal growth are the mold exit temperature, the casting speed, and the cooling condition. The diameter of the single crystal or alloy wire decreases

FIGURE 3.70 Schematic illustration of the Ohno continuous casting equipment.239 (Courtesy of H. Soda, A. McLean, Z. Wang, and G. Motoyasu, J. Mater. Sci., vol. 30, 1995, p. 5438.)

CHAPTER THREE

3.102

with the increased mold exit temperature. Single crystals of small diameter can be easily produced with excellent surface quality for use in vapor deposition applications.239,239a Another feature of this process is that the heated mold can be placed in a vertically upward, a vertically downward, or a horizontal position. Ohno has obtained encouraging results with Al, Pb, Sn, Cu, and their alloys.238 More recently, Kim and Kou238 and Soda et al.239 investigated the experimental variables of the OCC process and accomplished numerical modeling of heat and fluid flow. Tada and Ohno extended the OCC principles for the production of Al strips using an open horizontal, heated mold and patented as the Ohno strip casting (OSC) process. This technology has the potential to offer an alternative route to traditional methods such as rolling and extruding for materials that are otherwise difficult to fabricate due to limitations over other physical properties.239a Like most solidification processes, attempts have been made to mathematically model the freezing of a continuous cast nonferrous alloy ingot. For aluminum alloys, a short review has been made by Shercliff et al.241 Hot-Top Casting. The hot-top level-pour system has been developed for casting extrusion billets or slabs of normal-purity aluminum and aluminum alloys (containing higher Mg content) in order to achieve a superior cast surface quality, higher casting speed (due to a reduction in effective mold length), and a more uniform and finer subsurface structure that can be exploited to advantage in certain applications requiring a uniform surface finishing response.241 The metal feeding into the mold occurs with spout and float (Fig. 3.71a).241a

Launder

Float

Hot top

Melt

From furnace

Aluminum mold Air gap

Primary water chamber

Mold

Slab

Bottom cover

(a) Level control valve on floating metal distributor

Secondary water chamber Mold body

Graphite bore liner

Spout

Cooling water

Top cover

(c) Constant metal level in trough Liquid

Liquid metal

Liquid Semisolid

Starter block

(b)

To next mold Water box

Trough Screen Inductor Water spray

Solid

Water spray Water curtain

Ingot Bottom block Withdrawal ram

(d) 241a

FIGURE 3.71 (a) Schematic view of VAW hot-top mold for casting slab. (b) The vertical direct chill (DC) casting process for Al and Cu alloy casting process.225 (c) Schematic cross section of LHC mold.245a (d) The electromagnetic casting (EMC) process.249 (a and c: Courtesy of Light Metal Age. b: Courtesy of Academic Press, Boston. d: Courtesy of TMS, Warrendale, Pa.)

SOLIDIFICATION

3.103

To avoid molten metal bleedouts at start-up, due to thermal stresses and consequent curling developed in the first formed ingot, the metal level in the mold is established by start-up requirements. Other modifications have been introduced such as Isocast and Alcoa 729 processes that reduce the intensity of water cooling by promoting film boiling. This is achieved by passing CO2 gas into the cooling water stream, thereby minimizing curl and allowing the ingot to be started with a low head.242,243 Similarly, increasing casting speed will increase ingot surface temperature and promote film boiling.243a The Wagstaff Turbo process uses air bubbles while the Alcan pulsed water process uses a special rotary valve to turn the water on or off. Direct Chill (DC) Casting. Important variations of direct chill casting process include vertical direct chill (VDC) casting, horizontal direct chill (HDC) casting, low-head composite (LHC) casting, electromagnetic casting (EMC), and so forth.243b More recently the trend has been in favor of VDC casting, which allows a far greater number of ingot or slab strands to be cast simultaneously and which permits the more advanced EMC and hot-top mold technologies to be applied.244 The VDC casting process (Fig. 3.71b) is used widely to prepare rectangular rolling ingots or slabs and cylindrical billets of Al, Cu, Zn, and Mg and their alloys. The VDC casting process can directly produce billets for extrusion, blocks or ingots for rolling, and sheet for fabrication, thereby removing the intermediate mechanical processes by casting near-net shapes.245 This offers the ability to cast the full range of alloys up to the highest aircraft alloys, but requires an expensive hot rolling. This allows a far greater number of ingot strands to be cast simultaneously and the application of more advanced EMC and hot-top mold technologies. At the start-up, the open-end of the mold is closed off with the bottom block. As the mold begins to fill with degassed and filtered molten metal, the bottom block is lowered at such a rate that the molten metal level is controlled at a particular distance above the lower end of the mold, called the metal head. Thus, during casting, the mold exactly equals the metal exiting. As shown, solidification starts with the formation of a solid shell in the water-cooled mold. This shell shrinks away from the mold, and a gap forms which restricts the heat transfer to the mold. Most of the heat of solidification is eliminated from the ingot as it comes out from the mold into a water curtain which impinges directly onto the surface. There are three factors that affect the separation of the ingot shell from the mold: (1) shrinkage at the ingot shell itself, (2) thermal strain within the ingot shell, and (3) the shrinkage in the block section under the mold and associated mechanical strains in the shell. The primary and secondary water cooling systems usually influence them and thereby the ingot structure, primarily at the ingot surface. The air gap formed when the solid shell contracts away from the mold can give rise to a rough surface due to a surface liquation phenomenon. This shell shrinks away from the mold, a gap forms which diminishes the heat removal to the mold, and the solidified metal (or skin) reheats, causing surface and subsurface defects such as microand macrosegregation. The shell then reheats the point in which a mushy zone extends to the outer surface, and droplets of interdendritic liquid bleed through that surface, to produce a rough surface (“liquation beads”) on that surface as further contact is made with the mold. Reheating produces micro- and macrosegregation exudations, runouts, retardation of the solidification in the subsurface zone, and variations in the cell/dendrite size of the outer surface of ingots. Zones of coarse dendritic substructure may extend 2 to 3 cm below the surface. Additionally, large particles of intermetallic constituents are formed by eutectic reactions, which may be exposed by surface machining that is usually performed before fabrication.245 Several methods have been

3.104

CHAPTER THREE

proposed to reduce surface defects (such as cold shuts or laps, liquation, butt curl, butt swell, folds, and major and minor bleed-outs) and associated hot tearing propensity. The most successful ones are those that reduce the heat extraction at the mold through the control of the microgeometry of the mold surface, e.g., by machining fine grooves in the face of the mold. The molten Al does not fill the grooves due to surface tension. Highly automated DC casting is equipped with a programmable controller to control all or some of the parameters such as cast length, casting speed, cooling water, metal temperature, metal flow, and mold lubrication. Additional control capacity addition in the DC mold covers further requirements such as grain refiners wire rod feed rate, hot-top mold gas injection rate, and emergency shutdown due to loss of power.245a Low-Head Composite Casting Technology. The low-head composite (LHC) casting technology is similar to traditional DC casting technology in several ways. This arises from casting with a low metallostatic head in a permeable graphite-lined aluminum mold surface with a strong direct chill. Figure 3.71c is a schematic cross section of an LHC mold.245a The additional advantages of LHC mold are:245b (1) reduction of scrap rates (up to 2% increase in overall mill recovery between casting and cold rolling); (2) increase of casting speeds by 25 to 40%; (3) reduction of oil usage to 5 gal per month (versus the previous use of 500 gal per month); (4) reduction of shell zones†245c or scalping to 50% (with respect to a traditional DC mold); (5) very good ingot geometry; (6) smooth surface, usually free from exudation; and (7) unique method of heat removal during the run portion of the cast. Electromagnetic Casting. More recently, moldless electromagnetic casting (EMC), which has been developed by Getselev et al.,246 is finding applications in the production of sheet ingots (i.e., of rectangular cross section) because of the improved metal surface quality. It is a semicontinuous casting operation in which solidification of metal occurs without contact between the liquid metal and the mold wall (Fig. 3.71d). For highly conductive metals, electromagnetic forces are concentrated only close to the surface, extending over only a few millimeters, that supports the liquid metal away from the mold and against gravity at the periphery of the pool and causes stirring.247 Although this method has been gaining increasing acceptance since the 1980s, it is confined to non-heat-treatable alloys, because of the difficulty in crack removal associated with the more crack-prone heat-treatable alloys.247 Advantages of EMC over DC castings include (1) smooth and segregation-free surface, thereby allowing hot rolling without scalping operation; (2) solidification of the cast slab that is more uniform; (3) edge trimming that is reduced or avoided; and (4) finer than usual cell size in the subsurface area, resulting in reduced or no edge cracking.223 Disadvantages of EMC over DC casting are as follows: (1) Higher capital, licensing, and operating costs of EMC do not seem to outweigh improved recoveries. (2) Additional energy consumption is due to dissipation (10 to 30 kWh/ton) of electric power as heat into the metal, screen, or inductor. (3) One of the horizontal (surface) defects observed on EMC ingots is a surface waviness with ⬃1-cm wavelength, presumably because of the variation of the metal head above the solidification line. This results in leaning of the meniscus outward or inward from its normal position to maintain a balance between electromagnetic and metallostatic pressures, with a †

Shell zone is considered as a combination of inverse segregation and depletion.

SOLIDIFICATION

3.105

slight increase or decrease in ingot section. According to Dobatkin, the alloys exhibiting high hot brittleness or longer solidification time have susceptibility to develop surface cracks and coarse-grained structure in EMC. That is, EMC yields coarse-grained microstructure, leading to surface cracks.248 Developments of EMC led to improvements in conventional DC casting such as control of metal level and development of automation systems.243a There are two similarities between EMC and DC castings.249 (1) Both are semicontinuous processes in that an ingot, supported on a descending bottom block, is withdrawn continuously from a molten metal pool (the sump) into a casting pit, at a speed of about 1 mm/s, whereas a fresh molten metal flows in the pool from above. (2) In DC casting, the solidifying alloy is in contact with the mold; in EMC, the liquid pool is constrained on the side by electromagnetic forces (see Fig. 3.71d) while it is chilled. These forces are moderated by a band of conducting material, usually stainless steel, partially interposed between the inductor and liquid pool and termed the screen or shield. Alternatively, water cooling of the screen, inductor, and metal is provided, with the last being by sprays that strike the periphery of ingot just below the solidification line.

3.10.3 Solidification and Structure of Fusion Welds Weld metal solidification behavior controls the size and shape of grains, the extent of segregation, the distribution of inclusions, the extent of defects such as hot cracking and porosity, and the properties of weld metal. In the last three decades, several excellent reviews have been published emphasizing various aspects of weld solidification and weld microstructures.4,250–256 Figure 3.72 is a schematic diagram that describes an autogeneous welding process, exhibiting three distinct zones of a fusion weld. They are the fusion zone

FIGURE 3.72

Schematic diagram showing three zones of a fusion weldment.

3.106

CHAPTER THREE

(FZ), also called the weld metal; the unmelted heat-affected zone (HAZ) near the FZ; and the unaffected base metal (BM). The characteristics of the FZ depend, to a great extent, on the solidification behavior of the weld pool. However, according to close metallographic evaluation, the FZ can be further divided into three subzones: the composite zone (CZ), the unmixed zone (UZ), and a partially melted zone (PMZ), present between the FZ and the HAZ257 (Fig. 3.73). The UZ occurs in welds with filler metal additions and consists of molten base metal and a resolidified zone without mixing with filler metal additions during the movement of the weld pool. This zone can act as initiation sites for microcracking as well as corrosion susceptibility in stainless steel. In the PMZ region, the peak temperatures developed lie between the liquidus and solidus which leads to melting of lowmelting-point inclusions and segregated zones. After cooling, these areas may serve as potential sites of microcracks.258 In this section, weld pool solidification, shape, macrostructure, microstructure, and cracking are briefly discussed. 3.10.3.1 Characteristics of Weld Pool Solidification. Inherent to the welding process is the formation of a molten weld pool directly beneath the heat source which contains impurities. The weld metal shape is influenced by both the resultant heat and fluid (or metal) flow. A significant turbulence, i.e., good mixing, takes place in the molten metal. The heat input determines the volume of the molten metal and, therefore, dilution, weld metal composition, and the thermal condition. The molten metal volume is small when compared to the size of the base metal. The composition of molten metal is very similar to that of the base metal. There are large temperature gradients across the melt. Since the heat source moves, weld solidification is considered as a dynamic process comprising very rapid localized melting and freezing. Crystal growth rate is geometrically related to weld travel speed and weld pool shape. Hence, weld pool shape, weld pool composition, cooling rate, and growth rate are all factors interrelated to heat input that will ultimately affect the solidification microstructure. In the spectrum of solidification processes, cooling rates in conventional weld metal solidification lie between cooling rates for most castings (ranging from 10-2 to 102 K/s) and those for rapid solidification process (ranging from 104 to 107 K/s). For conventional welding process such as shielded metal arc, gas tungsten arc (GTA), submerged arc, and electroslag welding, the cooling rates may vary from 10 to 103 °C/s. For modern high-energy beam processes such as electron beam (EB) and laser welding (LW), cooling rate is typically of the order of 103 to 106 K/s within the welding pool. Weld pool solidification, therefore, involves features of both extremes of solidification, i.e., conventional casting as well as rapid solidification.

FIGURE 3.73 Schematic diagram of the different subzones of a fusion zone.257 (Courtesy of Welding J.)

SOLIDIFICATION

3.107

The local solidification conditions and cooling rates vary significantly within the weld pool.250,254 3.10.3.2 Weld Shape and Macrostructure. In contrast to a casting, solidification in a weld pool is more transient, there is no nucleation barrier, and chill zone is absent. In fact, in Al- and Fe-base alloys, it has been established that the initial weld pool solidification occurs by epitaxial grain growth at the FZ-HAZ interface (i.e., partially melted solid grains).251,258,259 The development of weld pool geometry is influenced by the amount of heat transfer from the heat source to the workpiece, the welding speed, the nature of fluid flow in the weld pool, and the rate at which heat can be removed at the L-S interface. As shown in Fig. 3.74a, at low heat inputs and welding speeds, an elliptically shaped weld pool forms and the columnar grains curve along the welding direction. At high heat inputs and welding speeds, weld pool becomes teardrop-shaped and the columnar grains are straight (Fig. 3.74b). In both situations, grain growth begins from the substrate at the fusion boundary and advances toward the weld centerline. In this manner, the weld metal grain structures adjacent to the fusion boundary are eclipsed by the epitaxial growth process. An elliptical or spherical pool is usually observed in weldments of a higher thermal diffusivity (e.g., Al), whereas a tear-shaped weld pool is more likely in the weldments of a low thermal diffusivity

FIGURE 3.74 Schematic showing effect of welding parameters on a grain structure: (a) low heat input and low welding speed producing elliptical weld pool; (b) high heat input and high welding speed, in the absence of nucleation in the bulk weld metal; and (c) high heat input and welding speed, in the presence of nucleation in the bulk weld metal.261 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

3.108

CHAPTER THREE

(e.g., austenitic stainless steels). In higher-carbon steels or in Al alloys such as 5083 Al alloy, the columnar grains have the tendency to multiply near the weld centerline with a corresponding increase in dendritic arm branching. This provides the formation of stray grain structure.256 In most commercial Al alloys, low-carbon steels, and 304 stainless steels (at still higher weld speed, say, >30 to 40 cm/min), equiaxed grains are found near the weld centerline (Fig. 3.74c). In this case, the competitive growth changes from epitaxial columnar grains to nucleation and growth of equiaxed grains in the bulk weld metal, similar to that noticed in ingots and castings.251,260,261 In addition to the factors mentioned above, the weld pool geometry is influenced by convectional and the resultant heat and fluid flow behavior due to three driving forces such as buoyancy effects, electromagnetic forces, and surface tension gradient forces. The interactions among these driving forces have been modeled by Wang and Kou,262 illustrating the effects on the shape and weld penetration in GTA aluminum welds, and by Zacharia et al.263 in GTA welding of type 304 stainless steel. It is important to mention that impurities in the weld metal, acting as surface active elements in the base metal, can change the surface tension of the liquid metal and its temperature dependence. In the absence of surface active elements such as O2 and S (5 wt%). Sr addition produced a significant grain refinement effect in pure Mg and low-Al containing alloys but a very small grain refinement effect in Mg-9Al alloy. The addition of a small amount of Zr, Si, and Ca to pure Mg produced an effective grain refinement effect. The grain refinement is mainly attributed to the growth restriction effects caused by constitutional undercooling, during solidification; but the influence of nucleant particles, either introduced with the alloying additions or as secondary phases formed due to these additions, may enhance the grain refinement. Readers are referred to an article by Lee et al. for more details on this topic.300b

3.11.4 Eutectic Modification of Al-Si Alloys Among the most important foundry alloys are those on the Al-Si and Fe-C systems. The mechanical properties of these metal-nonmetal (nonfaceted-faceted) eutectics are mainly dependent on the morphologies where the nonmetals solidify. The Al-Si system forms a simple eutectic with a composition of 12.5 wt% Si at 577°C (850°F) between an Al solid solution containing 1.65 wt% Si and virtually pure Si. Alloys with Si as the major alloying additions are very important Al casting

3.126

CHAPTER THREE

alloys primarily due to their high fluidity, low casting shrinkage, high corrosion resistance, low coefficient of thermal expansion (CTE), good welding, and easy brazing. In the eutectic Al-Si system, the Al phase is nonfaceted and in hypoeutectic alloy it is dendritic. The minor Si phase, in pure binary alloys, freezes as faceted flakes, either as primary phase or as the finer eutectic constituent; it is usually able to grow in only certain crystallographic directions, which plays a major role in modification. The modification is essentially a method of changing the shape/structure (morphology) in Al-Si alloys during solidification. In normal, unmodified Al-Si alloys, the Si crystals grow in a faceted manner, i.e., on the close-packed flat (111) faces of the diamond cubic structure, usually combined with a few twins across the ·111Ò planes (Fig. 3.86). Modification of eutectic phase is very desirable, particularly for slower rate solidification processes such as sand or permanent mold casting. However, recent work has shown that in pressure die casting, Sr modification also produces an improvement in feeding characteristics.284b The modification is carried out by treating the melt with certain elements,† called modifiers, or by subjecting the melt to a fast cooling rate. The modification of eutectic structure in Al-Si casting alloys is obtained by the use of Na, Sr, and Sb, Ca, K, P, Li, B, Ti, and selected rare earth elements. Of all these, Na, Sr, and Sb are the most popular and effective modifiers at low concentration levels, typically 0.01 to 0.15 wt%.301 The effectiveness of various elements as modifiers is clearly a function of the number of twins formed in the modified Si fibers. Careful studies of TEM have revealed that modified Si fibers contain orders of magnitude more twins than do unmodified Si plates. The significant increase in twin density is caused by the addition of only a fraction of 1 wt% of modifier which becomes concentrated in the Si, and not in the Al phase.295 The surface of the fibers is microfaceted and rough due to intersection of myriad twin planes with it. The modification features of silicon fibers in Al-Si alloys by Na addition are summarized in Fig. 3.87 and Table 3.6.302 The modification of Si in Al-Si alloys can improve its mechanical properties by changing the Si structure of the alloy (Table 3.7). Since Si is a major constituent of these hypoeutectic alloys—typically ranging between 5 and 12%—it plays a great role in the processing and final properties of the casting. The modification of the structure of these alloys and the resultant effect on the mechanical properties drive the foundrymen to manipulate the microstructure based on the casting alloys.295 The microstructural change from acicular to fibrous silicon is not a sharp one. Modification with Sr is often less uniform than with Na; and Sb will, of course, produce only a lamellar and never a fibrous structure. The formation of exact microstructure depends on five parameters: type and amount of modifier used, impurities present in the melt, Si content of the alloy, and freezing/solidification rate. Type and Amount of Chemical Modification by Na, Sr, or Sb. Sodium is the most efficient modifier, although the effect fades rapidly by evaporation and oxidation during holding in the liquid state. Additionally the use of Na modifier decreases the castability of Al-Si alloys. The addition of Na is associated with a violent reaction

† They are surface active agents with favored crystallographic orientations which facilitate nucleation of silicon grains.284b

3.127 (a)

(b)

FIGURE 3.86 Microstructure of Si flakes in unmodified Al-Si eutectic alloy. (a) Optical microstructure showing irregular growth front; (b) SEM, deep-etched removing metal matrix; and (c) TEM showing a few {111} twins and growth orientation.301 (Reprinted by permission of TMS, Warrendale, Pa.)

(c)

3.128 (a)

(b)

FIGURE 3.87 Microstructure of Si fibers in Na-modified Al-Si eutectic alloy. (a) Optical micrograph, showing a near planar growth front; (b) scanning electron micrograph of deep-etched removing metal matrix; and (c) transmission electron micrograph showing a high density of multiple twinning and growth orientation. (Reprinted by permission of TMS, Warrendale, Pa.)

(c)

SOLIDIFICATION

3.129

TABLE 3.6 Summary of Experimental Evidence due to Modification301

Unmodified

Modified minor additions

Chill

Microstructural shape

Faceted flakes

Microfaceted fibers

Smooth fibers (nonfaceted)

Internal structure (twin spacing)

Very few twins (0.4–1 mm)

Heavily twinned (0.005–0.1 mm)

Few twins (very large)

Growth undercooling

Relatively small

Large

Very large

Growth interface

Irregular

Near planar

Uncertain

Distribution of modifier

Not applicable

Within Si phase

Not applicable

Reprinted by permission of TMS, Warrendale, Pa.

TABLE 3.7 Tensile Properties of A356 Alloy Unmodified and Treated with Different Modifiers Heat treatment†

As-cast Modifier None Sb Na Sr

Silicon structure

UTS, MPa

El., %

Q, MPa

UTS, MPa

El., %

Q, MPa

Acicular Lamellar Fibrous Fibrous

180 201 195 196

6.8 11.9 16.4 15.9

305 362 377 376

304 293 292 301

11.8 16.5 15.1 14.4

465 476 469 475

†

Solution treatment at 450°C for 10 hr and aging at 160°C for 6 hr. Source: Modern Casting, January 1990, pp. 24–27.

and results in the increase in H2 levels. Sodium in excess of ⬃0.015% causes overmodification bands where the duplex eutectic front is briefly arrested and overgrown by aluminum film. The eutectic growth front is no more jagged and irregular but becomes more or less planar. Na Modifier. Sodium has easy dissolution above 700°C, but poor and somewhat unpredictable recovery. Sodium is the more powerful modifier, easy to use, and the most beneficial to the eutectic structure (by producing more uniformly modified structures at low concentrations) and the mechanical properties; but it fades rapidly by evaporation and oxidizes during holding of the molten alloy. Prolonged holding (>>30 to 40 min) can require renewal of the treatment. The Na-modified melts have the tendency to gas pickup and increased porosity. Oxidation of melt reduces castability and attacks aggressively against mold coatings or electrical resistance. Sodium interacts with P, if present in the alloy, necessitating increased additions of Na with the increased P content and decreased solidification rate. Typical retained Na levels sufficient to modify the Al-Si system are in the range of 0.005 to 0.015%. Remodification is accomplished, if required, to maintain the desired modification level. Since sand castings cool more slowly than permanent mold castings, they will need higher Na content; the time delay for pouring is more critical in sand than in permanent mold casting.295

3.130

CHAPTER THREE

Sodium is added as a flux, but many modern applications employ vacuum prepacked Na metal in small aluminum cans to minimize its oxidation and hydrogenation. Dissolution of Na is usually complete within 5 min. Sr Modifier. Strontium has a similar effect as Na on the eutectic structure. The current trend is to use Sr as a modifier, which is introduced into Al-Si alloy melt as pure solid metal, salt, or master alloys such as Al-3.5Sr, Al-10Sr, Al-10Sr-14Si, Al10Sr-15Si, and 90Sr-10Al, in the form of waffles, piglets, or rods. The rod form is reported to improve production-floor flexibility and effectiveness. This is attributed to a finer microscopic size of the contained Sr compound.302 However, the problem of adding elemental Sr and its salts to liquid metal, in addition to its high cost, limits the commercial use of Sr as a modifier. Recently, less expensive master alloys have been put in market, making Sr addition practical from the cost standpoint. Additionally, Sr has less effect on the oxidizing behavior of the alloy; its action is more durable than that of Na, having a semipermanent effect. The Sr treatment does not introduce hydrogen into the liquid-metal state, which causes greater porosity in the casting. The rate of loss of Sr is definitely less than that in Na, and the use of premodified alloy ingots is simple. Its sensitivity to gas pickup and porosity formation in part areas not well fed or cooled may be a drawback in day-to-day foundry practice and for mechanical properties in highly stressed parts. It is, however, easy for many types of castings. A range of 0.015 to 0.05% Sr is a standard industrial practice. Remodification through Sr additions may be required, although less frequently done than for Na. Its amounts of 0.02% are adequate to modify an A356 (Al-7SiMg) alloy, but up to 0.04% is required for a eutectic alloy such as A413. Strontium additions can be made with ease in the form of Al-15Si-10Sr master alloy with optimum additions in the range of 0.04 to 0.1 wt%. The improvement in mechanical properties is similar to that found by Na modification without some of the undesirable features. Partial modification is obtained with additions of Ce, La, and Na with La, the last being the most effective in increasing the UTS by ⬃250%. An addition of 0.1% Ce also yields an improvement in machinability. A 0.2% addition of all three elements in a fluoride mixture is an effective modifier, and an addition of 1% mischmetal produces complete modification with a corresponding improvement in mechanical properties. The addition of 150 to 300 ppm of Sr to the wrought 6061 alloy improves the formation of ductile intermetallic alpha phase (AlFeSi) which is more potent than the less desirable brittle intermetallic beta phase (AlFeSi) which forms during solidification of the alloy.302 Sb Modifier. The Sb modifier has some benefits with respect to its capacity as a permanent-type modifier, primarily through its extremely low oxidizability. However, Sb has some limitations. It can only be used in die casting, because the lower speed of solidification results in segregation. The degree of modification with Sb is less than that achieved by Na or Sr modification. Moreover, the addition of Sb itself into Al-Si alloy is associated with various problems due to the very long time needed for the Sb to be dissolved in the molten metal, even with continuous agitation of the melt. Antimony yields a lamellar eutectic structure which is particularly sensitive to the freezing rate. In practice, Sb is used in the range of 0.06 to 0.50%. Its permanent effect is linked with a tendency to have less gas pickup and less porosity formation, both being beneficial for permanent mold castings that are subjected to high stresses and require consistent properties. Antimony-treated alloy is purchased as pre-modified ingot from primary Al suppliers and is simply remelted and cast. As

SOLIDIFICATION

3.131

Sb is very stable in the melt, losses are virtually nil, and no extra additions are needed.295 Antimony treatment is not recommended for sand casting, because of the occurrence of nonuniform lamellar structure due to the slow freezing rate. It is usually confined to permanent mold applications and is used mostly in Europe and Japan.303 Impurities and Contamination Present in the Melt. The presence of minor or impurity elements in Al-Si alloys may be either beneficial or detrimental to certain mechanical properties. Iron impurity leads to detrimental effect on the ductility and corrosion resistance of the alloy. Presence of Pb in Al-Mg-Si alloy yields a lowductility intergranular fracture. However, mechanical properties of Al-Si alloys, especially elongation, depend on both the alloy composition and the eutectic-silicon size and morphology. It is pointed out that the presence of phosphorus makes modification difficult; that is, low P content in the alloy simplifies the modification. Since Sb interacts with both Na and Sr in a negative manner, Sb-containing melts need very high levels of either modifier to produce the desired structures. It is reported by some researchers that Mg makes modification easier and by others, more difficult.295 Silicon Content of the Alloy. Larger modifier content (up to 50%) is required to produce complete modification in higher (11%) Si concentration than in Al-7% Si alloy. Solidification Rate and Modifying Efficiency. It is well recognized that Al-Si eutectic microstructures may be altered significantly by both the solidification conditions (cooling/growth rates) and minor additions of certain modifiers, which result in enhanced modification and finer structures. In the presence of an impurity- (or chemical-) modified eutectic Al-Si alloy, the Al phase is not markedly affected; but the modified Si phases/fibers become very heavily twinned and are actually microfaceted, and the angle of branching increases with solidification rate. In modified alloys these impurities are linked with the Si phase, and it is found that they are adsorbed upon the Si liquid growth front.304 In contrast, quench-modified Si is essentially twin-free and nonfaceted. This observation is in agreement with the previous finding of twin formation in Si, grown epitaxially from Al solutions in the presence of Na.305 The impurity-induced modification is accomplished at much slower rates by minor additions of Na (⬃0.01 wt%), Sr (⬃0.1 wt%), and less certainly with other alkali, alkaline earth, and rare earth metals.306 It is also noted that chemical modifiers are more effective at higher freezing (solidification) rates as in chill casting rather than in a heavy section sand casting. This structure is considered as a very fine scale of the unmodified eutectic. It is of little practical consequence, since commercial casting processes, with the possible exception of die casting, do not involve very high solidification rates to cause quench modification. Like the Al-Si system, Si modification may be observed in other diamond cubic phases such as Ga and III-V compounds.304 Quench Modifier Fibers. These occur at growth rates of ≥1 mm/s and are quite smooth on external surfaces and more often twin-free than otherwise. The quenchmodified Si fibers are characterized by isotropic growth, nonfaceted in appearance, very much finer than the slowly grown flakes, and also finer than the impuritymodified fibers which form at relatively low growth rates.

3.132

CHAPTER THREE

Impurity-Modified Fibers. Detailed examination of the Na-modified fibers shows them to be externally rough or microfaceted and that they contain a very high twin density on up to four {111} systems (Fig. 3.87); the preferred growth axis is then in a ·100Ò direction, with symmetric branching in a ·211Ò direction. In some cases, the average twin spacing may be as low as 5 nm, and when compared to the prevailing growth rates, this corresponds to up to 104 twinning events per second. The structure is therefore considerably imperfect crystallographically. Like quenchmodified fibers, the external surfaces of impurity-modified fibers are not smooth, but are rough or microfaceted on a scale determined by twin density, and the occurrence of general fibrous morphology is a consequence of growth in many directions on that fine scale. TEM studies of Sr-modified materials show a similar fibrous morphology although less heavily twinned or faceted for a comparable analyzed level of impurity concentration. Higher solidification rate accelerates the modification process; hence, lower modifier content is required in permanent mold casting than in heavy section sand castings. Additions of 0.02 to 0.03% Sr to die casting 380.0 alloy lead to a noticeably finer microstructure, thereby improving machining properties. Also, doubly modified alloys that are inoculated with Na as well as quenched showed a fine fibrous structure with a very high multiple-twin density, comparable with Na modification at slower growth rates; that is, the promotion of twinning by impurity is independent of the growth rate for medium to high rates.304 A combination of grain refining and impurity modification seems to improve the mechanical properties, especially at low cooling rates.307 Overmodification. Overmodification with Na in excess of 0.018 to 0.02% produces coarsening of Si (due to the formation of AlSiNa compound) together with bands of primary aluminum in the final cast product. An overmodification with Sr is less critical than an overmodification with Na from the standpoint of mechanical properties. However, an overmodification may influence castability. Strontium overmodification leads to coarsening of the Si structure and the reversion of the fine fibrous Si to an interconnected platelike nature. The reasons for its occurrence are still unknown. Another evidence of Sr overmodification is the formation of Sr containing intermetallic phases in the microstructure such as Al4SrSi2 phase. It is evident that both of these effects will yield reduced properties of the alloy to the levels more typical of untreated alloy. Surprisingly, these two effects do not seem to occur simultaneously; that is, Al4SrSi2 can take place without remarkable Si coarsening, and vice versa. Modifier Fading. Exact fading rates of Na depend on the circumstances of addition. Large melts are less susceptible to fading than small ones due to the lower ratio of melt surface area to melt volume. Stirring increases fading rapidly, and thus degassing, even with an inert gas, is not recommended after Na treatment. Usually, Sr fades considerably more slowly than Na; therefore, it is considered as a semipermanent modifier. The major Sr losses from the melt are from oxidation because of the formation of stable strontium oxide. The Effect of Phosphorus. As mentioned earlier, P interferes with modification by either Na, Sr, or Sb and alloys containing high P contents require higher retained modifier concentration in order to produce an acceptable cast structure. An Al-Si casting alloy with lower (9°C), pyramidal instabilities occur on the faces of graphite crystal. At undercoolings of 29 to 35°C (50 to 65°F), instabilities occur on the (101¯0) faces of the pyramid, and the theory predicts the occurrence of spherulite graphite at these undercoolings. Finally, at large undercoolings of 40°C (70°F), the growth form observed is a pyramidal one, bounded by (101¯0) faces. These pyramidal crystals are part of the series of imperfect forms noticed especially in thick-wall SG iron castings.314 More recently, it is shown that the solidification mechanism of SG iron is more complicated and that g dendrites play an important role in eutectic solidification.314 Coral Graphite. This is an intermediate morphology between lamellar and spherulitic graphite and exists as cylindrical rods (Fig. 3.94)320 in high-purity Fe-CSi alloys with S < 0.001% and high Si.321

FIGURE 3.94 Cylindrical-shaped crystals called coral graphite in cast iron.320 (Reprinted by permission of Pergamon Press, Plc.)

SOLIDIFICATION

3.141

As the name suggests, the graphite occurs as fibers with roughly round cross section. It is an irregularly branched rod-shaped structure (in contrast to the branched flake structure of type D flake) and connected to form a highly convoluted and interconnected network in a manner similar to that of fibrous Al-Si eutectic. TEM studies suggest that the internal structure of fibers comprises sheets of graphite wrapped around the fiber axis to provide irregular scrolls.322 It is reported that coral graphite does not occur in commercial irons because of the reduction of interface undercooling and neutralization of the influence of silicon as a result of adsorbed sulfur.143 Compacted (Vermicular) Graphite Eutectic. Compacted graphite grows with at least one branch in direct contact with the liquid by an instability of flake surfaces leading to thick (chunky or compacted) graphite crystals as interconnected segments within an austenitic matrix. The pyramidal instabilities form on primarily a flake-type growth and may become rounded as recalescence occurs and the interface undercooling is decreased. Alternatively, the pyramidal promontories may flatten.323 The scanning electron micrographs of chunky graphite extracted from an iron alloy are shown in Fig. 1.17b. Compacted graphite occurs very frequently in heavy section spheroidal iron castings. It has been reported that even with the presence of sufficient spheroidizing agent, the very slow cooling rate can result in the formation of incomplete shell around spheroids and that thermal currents can dissolve a portion of the g shell, exposing the graphite to the liquid. In reality, a balance between flake-promoting elements such as S and O, spheroidizing elements such as Mg, Ce, and La, and antispheroidizing elements such as Ti and Al is necessary for a successful growth of compacted graphite structure.

3.12 NEW SOLIDIFICATION PROCESSES In this section we describe rapid solidification, squeeze casting, semisolid metal forming, metal-matrix composite fabrication, the Cosworth process, and improved low-pressure casting processes.

3.12.1 Rapid Solidification Process (RSP) Rapid solidification is defined as the rapid cooling (in excess of 103 K/s) of the molten metal through the solidification temperature range which is well above the range achieved in conventional ingot metallurgy. In RSP the significant undercooling produces metastable effects which can be grouped into constitutional and microstructural categories.324 Constitutional changes result in extension of solid solubility limits, formation of nonequilibrium or metastable crystalline or quasi-crystalline intermediate phases, production of metallic glasses, and retention of disordered crystalline structures in normally ordered materials and intermetallic compounds. The microstructural effects comprise morphology changes and refinement of scales of microstructural features (such as grain size, DAS, shape, and location of phases present) and a large reduction in scales of solute segregation effects.324 Figure 3.95 shows the combination of molten metal stream and cooling medium used in the major rapid solidification processes.325

3.142

CHAPTER THREE

FIGURE 3.95 Combinations of molten metal stream and cooling used in the major rapid solidification processes.325 (Courtesy of VCH, Weinheim, Germany.)

Important advantages of RSP are the chemical homogeneity of material (i.e., reduction in chemical and dendritic or microsegregation), microstructural refinement (i.e., small grain size, fine dendritic arm spacing, elimination of massive phases, and fine-scale dispersion of precipitates and dispersoids), large extension of solidsolubility limit of alloying elements (i.e., alloying flexibility), achievement of highdensity point defects, formation of unique crystalline (amorphous or metallic glassy) metastable (or nonequilibrium) phases, and increased tolerance of tramp elements,325–327 which have a significant bearing on the properties and structural engineering applications of alloys. Other advantages include better fabricability, the manufacture of net or near-net shapes, and the removal of highly textured products, especially in hcp Ti materials.324 A widespread application of RSP has been to develop new alloy systems (Al-, Mg-, Cu-, Fe-, Ni-, and Ti-based alloy systems) with greatly superior strength properties, improved corrosion resistance, and a highly desirable combination of magnetic properties which cannot be obtained in conventional materials.328,329 Typical examples are ferromagnetic metallic glasses based on amorphous Fe-B-Si alloys for transformer applications; crystalline soft magnetic Fe-B, Fe-Si-Al, Fe-Al, and Fe-BSi-Al alloys;330 hard magnetic alloys based on crystalline Fe-Nd-B alloy; and Be-Cu thin strip for applications in electrical and electronic connectors, switches, relays,

SOLIDIFICATION

3.143

diaphragms, corrugated bellows, and miniaturization of many electronic devices.331 In the last decade the RSP has been extended to the production of ceramic powders directed to the circumvention of toughness problems. Rapid solidification methods for the production of metal powders may be classified into conduction and convection processes. Chill block melt spinning (CBMS), free-flight melt spinning (FFJM), free-jet melt spinning (FJMS), planar flow casting (PFC), crucible melt extraction (CME), and melt overflow (CMO) processes follow the conductive cooling. Spray deposition (or forming) and ultrasonic atomization follow the convective cooling. RSP involves mostly rapid removal of latent heat of fusion of the melt by conductive and convective heat-transfer mechanisms. In both cases the solidification rate depends primarily on the heat-transfer coefficient of the liquid metal. Typically the cooling rate of conduction processes ranges between 106 and 108 °C/s (1.8 ¥ 106 and 1.8 ¥ 108 °F/s) whereas the convection processes may be limited to the range 104 to 106 °C/s (1.8 ¥ 104 to 1.8 ¥ 106 °F/s).

3.12.1.1 Conduction Processes Chill Block Melt Spinning Process. The CBMS, or simply melt spinning, process has become the most widely used technique to produce long and continuous ribbons. A wide variety of materials such as steel, Al, Cu, Ti-base alloys, superalloys, Fe- and Ni-based metallic glass alloys (for numerous applications including soft ferromagnetic lamination for power distribution transformers, cutting and forming tool materials, and wear- and corrosion-resistant hard facing coatings) have been successfully melt spun as filaments.332,333 The CBMS process has been modified to manufacture helical glassy alloy ribbons,334 composite alloys, and multilayer deposits.335 The thickness d (mm) of ribbon filament in melt spinning is proportional to the jet diameter d and inversely proportional to the velocity v (m/s) of the spinning disk surface: d=

kd v

(3.165)

In the CBMS process, a jet of liquid metal is ejected through a round orifice onto a rapidly moving highly thermally conductive, chill wheel substrate surface (Fig. 3.96a). Continuous, long, and uniform ribbons, typically about 5 mm wide and 15 to 100 mm thick, are formed. The important processing parameters involved in CBMS are selection of appropriate crucibles and wheels chamber atmosphere, substrate velocity, molten metal pressure, nozzle design, and nozzle-substrate distance.324,336 Free-Flight Melt Spinning Process. Also called melt extrusion, this process consists of ejecting the liquid through an orifice and subsequently solidifying a stable liquid metal jet through a gaseous or liquid quenching medium (Fig. 3.96b). Typically, circular liquid jets ranging from about 50 to 1250 mm are used, and the pressure required to eject the molten metal from the orifice increases with the decrease of orifice area. The quenchants generally used to both stabilize the molten metal jet and accelerate its cooling rate include ambient air, air-water fog (for cast iron), liquid quenchant, liquid N2, inert gases, and brine. The main advantage of this process is its simplicity and production of continuous, round filaments for wire-type applications such as Fe-base tire cord.325

CHAPTER THREE

3.144

(a)

(b)

FIGURE 3.96 Schematic illustration of (a) chill block melt spinning (or melt spinning) and (b) free-flight melt spinning technique.325 (Courtesy of VCH, Weinheim, Germany.)

Melt crucible Heating system Free melt jet Molten alloy puddle Ribbon

Rotating substrate wheel

Melt

Glass Substrate

Free jet melt-spinning (FJMS) (a)

(b)

FIGURE 3.97 Schematic representation of (a) free jet melt spinning (FJMS) and (b) planar flow casting (PFC) processes.337,339

Free Jet Melt Spinning (or Jet Casting) Process. In the continuous FJMS process, a thin molten alloy is ejected under pressure through an orifice to form a free melt jet which strikes an external surface of a water-cooled copper wheel ⬃100 mm in diameter rotating with a speed of ⬃30 m/s where the jet forms a melt puddle (of thickness about equal to and length about double that of the jet) and solidified into ribbons, which is expelled from the surface of the wheel as shown in Fig. 3.97a. This process is used to produce a wide range of amorphous alloys, but the meltpuddle does not remain considerably stable. As a result, the process is restricted to the production of ribbons 5-cm) deposit without sacrificing quality.348 Ultrasonic Gas Atomization (USGA) Process. An ultrasonic gas atomizer, also called a Hartmann whistle acoustic atomizer, involves the production of ultrafine and more uniform droplets from the molten metal stream and solidification into more uniform ultrafine powders (⬃20 mm) by impingement of ultrasonicfrequency (of 80 to 100 kHz), supersonic gas jets (often attaining 1.7 to 2.5 Mach speed of gas) by accelerating high-pressure gas (such as N, Ar, or He) through the resonant cavities (as the Hartmann tube).343,346,351,352 Figure 3.100b shows a convergent-divergent nozzle to produce an ultrasonic-frequency (of 80 to 100 kHz), supersonic gas jet. It is claimed that droplet formation in the USGA is a one-step process compared to the three-stage mechanism of conventional gas and water atomization. The advantages of the traditional USGA process include the more efficient breakup of molten stream, high relative velocity between gas and melt droplets leading to high average quench rates, finer and more uniform powder particles, and gas chilling factor due to expansion of high-pressure gas. The yields are >90%, and this process has been successfully used for the large-scale commercial production of low-melting alloys such as Al-Li alloys and for small-scale production of highmelting-temperature alloys such as Ti alloys, stainless steels, and Ni- and Co-based superalloys.329 Because of the large specific areas, care must be taken in the safe handling of these powders.

3.12.2 Solidification Processing of Metal-Matrix Composites Metal-matrix composites are defined as artificial or advanced material comprising any combination of fibers, whiskers, particles, and metal wires embedded in a metallic matrix.353 They possess attractive properties such as enhanced specific strength,

3.150

CHAPTER THREE

specific modulus, and damping properties; low specific gravity and coefficient of thermal expansion (CTE); superior wear and abrasion resistance and frictional properties; better elevated-temperature strengths; increased creep-rupture and fatigue properties; good microcreep performance;353a moderate to high toughness; improved corrosion resistance; high electrical and thermal conductivities; and in several cases, relatively moderate composite fabrication cost. MMCs also offer the opportunities to develop new materials with a unique set of properties that cannot be realized with conventional monolithic materials.354–356 However, the main drawback with these materials is that they often suffer from poor tensile ductility, insufficient fracture toughness, and poor fracture resistance compared to the reinforced matrix alloy.353a Recently there has been a widespread use of reinforced MMCs as materials of construction for high-performance structural, aerospace, automotive, marine, electronic, and sporting goods applications. A detailed review of the processes for producing MMCs has been available covering numerous metallic matrices such as light alloy matrices, high-temperature alloy matrices, high-thermal-conductivity matrices, and reinforcements in the form of a diversity of particulates, whiskers, and continuous fibers or short fibers.1,357–361 Table 3.8 lists some of the potential applications of MMCs in automobile components and justification for their use.357 Processing methods may be classified into (1) solidification, casting, or liquid-state processing; (2) solid-state processes; (3) deposition processes; and (4) deformation processes. Among these, solidification processing in which the molten metal matrix is combined with the reinforcing phase in the form of the particles, whiskers, or fibers in the final composite material is attractive and finding greater recognition in the production of complex-shaped components at high production rates due to simplicity, cost-effectiveness, and ease of handling liquid metal around the reinforcing phases. It also offers a wide selection of materials and processing conditions. Good wetting condition is a prime requirement for the formation of a satisfactory bond between solid ceramic phase and liquid-metal matrix during casting of composites.353 Based on the mechanism for combination of reinforcement and molten-metal matrix, liquid-state processing of MMCs can be broadly grouped into four major categories: infiltration, dispersion, spraying, and in situ fabrication.363 Infiltration Casting Process. This involves injecting molten metal into a preheated preform of porous ceramic-reinforced skeleton in a metal mold in which open porosity is entirely filled, and a composite is produced with enhanced mechanical and thermal properties over conventional engineering materials. This is often accomplished with short alumina fibers such as Saffil or with very fine SiC whisker fibers. Figure 3.101 shows two examples of infiltration processes. The main parameters common to infiltration processes are the morphology, initial composition, volume fraction, and temperature of the reinforcement; the initial composition and temperature of the infiltrating molten metal; and the nature and magnitude of the external force applied to the liquid metal, if any. Accordingly, infiltration processes are categorized into pressureless infiltration, vacuum-assisted infiltration, vibration-assisted infiltration, pressure infiltration, and electromagnetic body force-driven infiltration.361,362 In the Lanxide pressureless infiltration, the metal or alloy ingot is placed on the ceramic preform in a graphite mold, and the assembly is heated in a nitrogen atmosphere above the liquid temperature, so that the molten metal spontaneously infiltrates the ceramic preform. The process variables are infiltration temperature, particle size, alloy composition, and nature of atmosphere. Here, infiltration occurs spontaneously, and thus the wettability of the dispersoid by the alloy is very important. In pressure infiltration the hydraulic pressure

TABLE 3.8 Some of Potential Applications of MMCs in Automobile Components356 Composite† Al-SiC (p)

Components Piston Brake rotor, caliper, cylinder liner Propeller shaft

Benefits Reduced weight, high strength and wear resistance Higher wear resistance and reduced weight

Manufacturers Duralcan, Toyota, Martin Marietta, Lanxide Duralcan, Lanxide

Reduction of weight and high specific stiffness

GKN, Duralcan

Connecting rods

Reduced reciprocating mass, high specific strength, stiffness and low CTE

Nissan

Mg-SiC (p)

Sprockets, pulleys, and covers

Reduced weight, high strength, and stiffness

Dow Chemical

Al-Al2O3 (sf)

Piston ring Piston crown (combustion bowl)

Wear resistance, high running temperature Reduced reciprocating mass, high creep and fatigue resistance Increase of driveshaft length for a constant cross-sectional area

Toyota T&N, JPL, Mahle, etc.

Al-SiC (w)

3.151

Driveshaft

Duralcan

Al-Al2O3 (lf)

Connecting rod

Reduced reciprocating mass, improved strength and stiffness

Dupont Chrysler

Cu-graphite

Electric contact strips, electronics packaging bearings

Low friction and wear, low CTE

Hitachi Ltd.

Al-graphite

Cylinder, liner piston, bearings

Low resistance, reduced friction, wear and weight

Associated Eng., CSIR

Al-TiC (p)

Piston, connecting rod

Reduced weight and wear

Martin Marietta

Al-fiber flax

Piston

Reduced weight and wear

Zollner

Al-Al2O3 (f) -C (f)

Engine block and cylinder liner

Reduced weight, improved strength and wear resistance

Honda

†

p, particle; w, whiskers; sf, short fibers; lf, long fibers. Reprinted by permission of Pergamon Press, Oxford.

CHAPTER THREE

3.152

Ram Fiber preform

Liquid metal

Infiltrated composite Liquid metal (a)

Infiltrated composite Fiber preform (b)

FIGURE 3.101 Schematic illustration of (a) pressureless (or spontaneous) infiltration and (b) pressure-driven infiltration.365 (Reprinted by permission of Butterworth-Heinemann, Boston, Mass.)

Motor Particles

Liquid metal

FIGURE 3.102 Schematic illustration of a dispersion process.365 (Reprinted by permission of Butterworth-Heinemann, Boston, Mass.)

of squeeze casting is replaced by gas pressure. The process variables are similar to those in squeeze casting.363 Dispersion Process. In the dispersion process, shown schematically in Fig. 3.102, the reinforcement phase is introduced in loose form into molten or semisolid metal

SOLIDIFICATION

3.153

matrix and vigorously stirred. This method is now the most inexpensive one to produce particulate-reinforced MMCs, which can be further processed by casting or extrusion. This process is applied to mixing of SiC particulates in molten Al under vacuum, using a specially designed impeller which reduces impurities, oxides, or gases as a result of vacuum and limited vortexing. Other methods include the bottom mixing process, where a rotating blade is gradually lowered into an evacuated bed of particles covered with molten Al and the injection of particles is carried out below the melt surface using a carrier gas. The important parameters in dispersion processes are the reinforcement particle size, volume fractional content of the reinforcement, dispersion uniformity (by agitation of liquid slurry), the temperature and composition of the metal, and the applied shear rate. Spray Casting Process. In this process (discussed earlier), droplets of liquid metal are sprayed together with the reinforcing phase and collected on a substrate, where they solidify completely. Alternatively, the reinforcement may be placed on the substrate, and liquid metal may be sprayed onto it. The important parameters in the spray processing are the initial temperature, size distribution, and velocity of metal drops; the velocity, temperature, and feeding rate of the reinforcement (provided it is simultaneously injected); and the position, nature, and temperature of the substrate collecting the material. According to Alcan International Limited, the particles may be injected within the droplet stream or between the liquid stream and the atomizing gas as in the Osprey process. Examples include SiC-, Al2O3-, or graphitereinforced Al alloys. Advantages of this technique include the matrix microstructure containing fine grain size and low segregation and minimal interfacial reaction, thereby allowing the production of thermodynamically metastable two-phase materials such as Fe particles in Al alloys.364 Disadvantages include the production of only simple forms such as ingots and tubes, extent of residual porosity, further processing requirements, and reduced economy compared to dispersion or infiltration processes, due to the high cost of gases used and the large amount of powders wasted.365 Reactive Processing (in situ MMCs). Recently, in situ processes of MMCs for nonferrous and intermetallic systems have been developed to produce a new class of naturally stable composites for advanced structural and wear applications due to their low production cost.366 They result from the directional solidification of eutectic alloys where a binary alloy melt with sufficiently low volume fraction of one of the phases (such as carbides, nitrides, or borides) is allowed to solidify in the form of fibers (rather than the more commonly observed lamellae of the eutectic composition). The critical volume fraction depends on solute diffusion characteristics and interfacial energy but is typically about 5%. Additionally, by choosing the growth conditions, the fiber diameter and spacing can be monitored to some extent.367 Some examples include eutectic systems based on Cr and Ta; Fe-TiC composites produced from solidification of Fe-Ti-C alloy melts; TiB rods in TiAl matrices produced from solidification of melts containing g-TiAl, Ta, and B; and TiC/Ti composites from mixtures of Ti, C with Al additions. The XD process is another in situ process which is also referred to as the selfpropagating high-temperature synthesis (SHS) process. This involves reaction between liquid metal and reacting constituents or compounds to produce the required reinforced metallic alloy or suitable composites such as TiB2-reinforced Al alloys, where Ti, B, and Al powders heated to 800°C react to form TiB2; TiB whiskers obtained after laser melting of Ti and ZrB2 powders; or TiAl matrix obtained after

CHAPTER THREE

3.154

squeeze casting of molten aluminum into TiO2 powders or short fibers. Reaction of molten metal with a gas also produces in situ composites such as Al2O3/Al composites by directional oxidation of molten Al and TiC particle reinforced Al-Cu alloys by injecting CH4 and Ar gas through a melt of Al-Cu-Ti alloy (Fig. 3.103).368 The main advantages of this process are the homogeneous distribution of reinforcing phase and the control of spacing or size of the reinforcement. However, the choice of systems and the orientation of the reinforcements are restricted, and the kinetics of the processes (in the case of reactions) and the shape of the reinforcing phases are sometimes difficult to control.365 Metal-matrix composite solidification constituting the former two processes consists of three stages. The first stage corresponds to the interaction of the reinforcement material and the liquid matrix, which, in turn, is controlled by their wetting and adhesion or chemical bonding characteristics. In most instances, external positive pressure is provided to reach a favorable wetting. However, there are alternative methods to improve the wettability which include reinforcement pretreatment, alloy modification of matrix, and reinforcement coating. To accomplish this, a comprehensive understanding of the surface of metal matrix, the reinforcement surface, and the interface chemistry, influence of alloying additions, and reactive wettings is needed.1 The second stage involves fluid flow, heat-transfer, and solidification phenomena that take place during the infiltration and prior to complete solidification; they dictate to a large extent the microstructure of the cast composites.361 The infiltration mechanism, thermal and solidification effects, and processing of MMC slurries (rheology and particle migration) must be given thorough consideration. The third stage denotes the completion of the solidification process. The reinforcement particle size, its fractional content, and dispersion uniformity have a pronounced influence on nucleation, coarsening, microsegregation, and grain size during the solidification of composites, which, in turn, are very important in improving its mechanical properties.1

Gas injection

Heating elements Liquid metal

Gas bubble Reaction product Liquid metal

Cooling coils Composite

Gas bubbler (a)

(b)

FIGURE 3.103 Schematic illustration of in situ processes. (a) Reaction between an injected gas and a liquid metal. (b) Directional solidification of a eutectic alloy.365 (a: From Koczak and Sahoo, 1991. b: Reprinted by permission of Butterworth-Heinemann, Boston, Mass.)

SOLIDIFICATION

3.155

3.12.3 Squeeze Casting Squeeze casting (SC) is a casting process with a slow filling rate, minimum turbulence, and high pressure throughout the solidification to consistently produce highintegrity castings capable of solution heat treatment.369 The applied pressure (ranging from 17.5 to 175 MPa, or 2.5 to 25 ksi) and the immediate contact of the liquid metal with the die surface produce a rapid heat-transfer condition that yields nearly defect (or pore) -free, fine-grain castings with improved mechanical properties. SC can be easily automated to produce near-net to net shape high-quality components with isotropic properties.370 The process was introduced in the United States in 1960 and has since gained widespread recognition within the nonferrous casting industry. The Al-, Mg-, and Cu-based alloy components are readily produced using this process. Several ferrous components with relatively simple shapes have also been produced by this process. The examples include aluminum automotive wheels, pistons, and dome; gear blanks made from brass and bronze; ductile iron morton shell; steel bevel gear; stainless steel blades; nickel hard-crusher wheel inserts; and superalloy disks. The process has been adapted to manufacture a wide range of MMCs using porous ceramic preform or fibers of SiC and Al2O3 at strategic site. Examples include automotive and diesel aluminum alloy matrix composite pistons and engine blocks that require increased mechanical properties, elevated temperature strength, and wear resistance. There are two basic types of squeeze casting, direct (Fig. 3.104a) and indirect (Fig. 3.104b), each with its own history and development.369 In direct squeeze casting method, a metered amount of liquid metal is poured into a preheated, lubricated mold, and the pressure (from a hydraulically activated source) is applied directly to the entire surface of liquid metal during solidification. In the indirect method, liquid metal is first poured into a shot sleeve and then injected vertically into the die cavity by a small-diameter piston, and pressure is applied through a runner system (during solidification). Of the two methods, the direct process is the more commonly used practice for the production of fully integrated castings such as automotive wheels, cylinder liners, suspension parts, hubs, hardware, drive train and steering components, and MMC components. Direct squeeze casting is mainly used for round, concentric shapes, but is not limited to these applications, and the use of inserts and slides of other materials is common. The major advantages of SC over other casting processes are as follows:369–380 1. High-quality cast components without gas and shrinkage porosities are produced. 2. No feeders or risers are required, and, therefore, no metal wastage occurs. The cast components are produced with near-net shape (close tolerances), requiring only small machining and finishing allowances as well as offering exceptionally high strength-to-weight ratio of the product. 3. The high pressure involved in SC makes the process well suited to casting inserts in place and cast-in preforms that must be infiltered by the cast alloy to produce near-net-shape MMC components for engineering applications. 4. Both common (heat-treatable and non-heat-treatable) and presumably wrought alloys can be squeeze-cast to finished shape. 5. Because of the presence of very fine grain size and small primary and secondary dendritic arm spacing, and the absence of gas or shrinkage porosity, the mechanical and physical properties of squeeze cast parts are superior to those of conventional permanent mold castings.

CHAPTER THREE

3.156

PUNCH LADLE

DIE

➀

➁

EJECTOR

➂

➃ (a)

➀

➁

➂ (b)

FIGURE 3.104 Schematic representation of (a) direct and (b) indirect squeeze casting processes. (Courtesy of ELM Int., Inc., Mich.)

6. Since squeeze casting produces sound castings, costly postsolidification examination by NDT may not be required. The parameters that require close attention are the casting variables such as pouring temperature (6 to 55°C, or 10 to 100°F, above the liquidus temperature), die temperatures (between 190 and 315°C, or 375 and 600°F), pressure levels (between 50 and 175 MPa, or 7.5 and 25 ksi), pressure duration (30 to 120 s), lubrication (colloidal graphite spray for Al-, Mg-, and Cu-base alloys and ceramic-type coating for ferrous castings). Usually the optimization of process parameters is required for quality and reproducibility of each squeeze cast component. Failure to do so results in one or more of the following defects:370

SOLIDIFICATION

3.157

1. Oxide inclusions can be minimized by cleaning melt handling and melt transfer systems, including filter in the melt-transfer system, filling the die cavity without turbulence of liquid metal, and preventing foreign objects from entering open dies. 2. Porosity and voids are removed usually by applying a sufficient pressure of at least 70 MPa or 10 ksi. 3. Extrusion segregation is avoided by increasing the die temperature, minimizing the die closure time, or selecting an alternate alloy. Blistering is avoided by degassing the melt and preheating the handling transfer equipment, using a slower die closing speed, increasing the die and punch venting, and reducing the pouring temperature. 4. Cold laps are alleviated by increasing the pouring temperature or die temperature and reducing the die closure time. 5. Hot tearing is overcome by reducing the pouring temperature and die temperature, increasing both the pressurizing time and draft angles on the casting. 6. Case debonding and extrusion debonding are eliminated by increasing the tooling or pouring temperature and decreasing the die closure time.

3.12.4 Semisolid Metal Forming Processes Semisolid metal (SSM) processing is a relatively new process of forming metals and alloys in the semisolid state to near-net-shape products. It depends on the behavior of semisolid slurries in which stirring creates strong shear forces that break off original dendritic structure into a globular or spheroidal structure and produce very fine grain size without using grain-refining additions (Fig. 3.105).381–383 This process of stirring semisolid metallic alloys during solidification to produce nondendritic solid within a slurry and subsequently injecting this slurry directly into

FIGURE 3.105 London.)

Semisolid metal forming processes.381 (Courtesy of The Institute of Materials,

3.158

CHAPTER THREE

dies, as in die casting, was initially called rheocasting. Alternatively, the slurries can be cast as a high-quality cylindrical billet, precisely cut to proper weight and sized slugs, and reheated rapidly to the desired semisolid (partially molten) condition, introduced into the diecasting machine or open die forging machine to make thixocasting or thixoforging.1 Using this method, all types of casting can be produced, including continuous rheocasting with or without electromagnetic stirring (EMS). The lower heat content of the slurry, thixotropic characteristic, and much-reduced solidification shrinkage and cracking lead to the following advantages of SSM forming.381–385 1. There is smooth filling of the die with essentially laminar flow, no air entrapment, and low shrinkage porosity, producing parts of high quality or integrity to nearnet shape rapidly and efficiently (Table 3.9). 2. There are substantial savings in risers and gating, closer tolerances and improved surface finish, pressure-tight components, reduced tendency toward hot tearing, and increased alloy range. 3. It is an energy-efficient process with the potential of easy automation, weight savings, minimal scrap, and increased die life and productivity. 4. There is ease of formation of composite materials by the addition of fibers or other solid particulates into the feedstock compocasting. 5. It is a cost-effective process due to finer and more uniform microstructure and reduction in component costs due to improved design. 6. There is faster formation of more intricate shapes, use of smaller presses, lower finishing costs, and significant reduction in forming stresses. 7. Capital investment and operating costs are lower than for conventional casting methods because of the containment of the entire process within one machine, easy maintenance of foundry cleanliness, and additional benefits, as in item 3. 8. Since the microstructure does not depend on the cooling rate, the mechanical properties were found to be independent of local sections. Semisolid Metal Casting. Its additional advantages include (1) control of metal quality at the source using filter, (2) no melt loss at the casting site, (3) elimination of liquid handling problems such as oxide and dross losses, (4) simple automation of reheating process because of the self-supporting slugs, and (5) energy saving because of the elimination of liquid metal-holding baths and melting of no more than 50% of the alloy at the SSM casting site. The drawbacks of the process are (1) runners, overflows, etc., must be recycled through a complete melting cycle and cannot be reclaimed by semisolid processing, and (2) the slug heating furnace is more expensive than a liquid-metal furnace. Unlike the liquid-metal casting process, SSM material flows under the action of applied pressure and does not follow gravity. This provides a degree of flexibility in die design not available elsewhere where liquid metal must always flow uphill and gating systems must adapt to this requirement. The unique rheological properties of SSM materials require careful control of the injection cycle. A balance must be exercised between filling a cavity fast enough to avoid premature freezing and maintaining laminar flow and thereby excluding air entrapment. Semisolid Al alloys exhibit only about one-half of the volumetric (solidification) shrinkage when compared to liquid alloys (3 versus 6%), and for some applications only moderate pressure is needed to produce acceptable parts. Since the parts

SOLIDIFICATION

3.159

do not contain any gas due to the laminar filling behavior, good surface and heattreatable parts can be produced readily. For pressure-tight parts such as brake parts for automobiles, a final pressurization stage is essential to feed liquid metal from the biscuit to the shrinkage centers, and care must be taken in gating and risering design to ensure an adequate feeding path. Raw Materials. For SSM forming, part production needs the special microstructure discussed above. When semisolid, this structure consists of spheroids or globule-shaped solid particles suspended in a matrix of low-melting alloy liquid. Preservation of this microstructure in materials heated from the solid state requires the retention of some residual microsegregation to provide differential melting between solid and liquid phases.386 SSM Components and Properties. SSM components are finding a wide range of applications in the automotive, aerospace, and electronic fields. Typical examples are multilink rear axle components for European automobile, fluid handling systems; Ford air conditioner front and rear compressor heads; and suspension parts, bicycle cranks, fuel system, electrical connectors, valve bodies, and threaded brass plumbing fittings. It is typically a competitor with wrought, permanent mold cast, or sometimes investment cast components. Mostly all these parts are being SSM cast from the well-established A356 and A357 (Al-7% Si-0.5% Mg) alloys and AZ91D Mg alloy. The mechanical and functional performance of SSM cast parts typically competes well with that of permanent mold cast parts and, in some instances, forged or machined parts. A major advantage of SSM casting is the ability to provide near-net shape with minimum machining. Table 3.10 compares an SSM cast automotive master brake cylinder part with its permanent mold-cast counterpart which shows that the SSM part requires less machining.381 Slurry casting or compocasting is the simplest and most economically attractive method of MMC production, in which the liquid metal is mixed with solid ceramic particles and then the mixture is allowed to solidify. Being a variation of rheocasting, this is conducted through a rapid temperature increase (up to liquidus point) just before pouring.382 It involves the development and production of MMCs containing nonmetallic particles, taking the benefits of rheological behavior and structure of partially solidified and agitated matrix. The particulate or fibrous nonmetals are introduced to the partially solid alloy slurry. The high viscosity of the slurry and the existence of a high volume fraction of primary solid in the alloy slurry prevent the floating, settling, or agglomeration of nonmetallic particles. As the mixing time, after addition, is increased, bonding increases due to the interaction between particles and the alloy matrix. The composites are then heated to the semisolid state in a second induction furnace and forged into shape with hydraulic press. Very promising wear-resistant alloys have been obtained by Sato and Mehrabian in Al alloys containing particulate additions of Al2O3 and SiC.387 Matsumiya and Flemings388 extended the application of SSM to strip casting, and the basic technology of SSM provides a potential method of metal purification.1,389

3.12.5 Cosworth Process Conventional methods of aluminum castings exhibit turbulent liquid-metal transfer; dispersed, concentrated, and connected porosity; poor and inconsistent mechanical strength; dimensional inaccuracies (especially when extremely cored); and

TABLE 3.9 Comparison of Semisolid Forging (Thixoforging) and Permanent Mold Casting for Production of Aluminum Automobile Wheels

Wt. from die/mold, kg

Finished part wt., kg

Production rate per die/mold, pieces/hr

Semisolid forging

7.5

6.1

Permanent mold castings

11.1

8.6

Process

Aluminum alloy

Heat treatment

Tensile strength, MPa

Yield strength, MPa

Elongation, %

90

357 (Al-7Si-0.3Mg)

T5

290

214

10

12

356 (Al-7Si-0.5Mg)

T6

221

152

8

3.160

Source: M. P. Kenney et al., ASM Metals Handbook, 9th ed., vol. 15, Casting, ASM International Metals, Park, Ohio, 1988, pp. 327–338.

TABLE 3.10 Comparison of Semisolid Cast Aluminum Automotive Master Brake Cylinder with Its Permanent Mold Cast Equivalent (Al-7Si-0.5Mg)381

Process

Cast wt., kg

Finished wt., kg

Production rate, pieces/hr

Heat treatment

Tensile strength, MPa

Yield strength, MPa

Elongation, %

Semisolid cast

0.45

0.39

150

T5

303

228

8

Permanent mold casting

0.76

0.45

24

T6

290

214

8

Courtesy of International Materials Reviews, London.

SOLIDIFICATION

3.161

dimensional instability during machining or service.390 In addition, they may display considerable entrained oxide films; considerable contraction on the order of 7% by volume during solidification; blowholes from chills, cores, and adhesives; inaccurately-located cores or mold halves; metallurgical insufficiencies (especially poor hardness or strength); and considerable fettling required to remove large feeder heads and extensive gates. Cosworth process is a unique low-pressure sand casting method, capable of producing a wide range of accurate, precise, completely reproducible, sound (free from porosity, oxide inclusions, and gas), high-integrity, pressure-tight, and qualitycontrolled components, including thin-walled section (0.16 in. or 4 mm). This process utilizes controlled melting, holding, and preparation of metal as well as liquid-metal transport through a programmable electromagnetic pump from the center of the bulk (or holding bath) to the final filling of the zircon sand mold cavity (located above the furnace) in a smooth, quiescent, nonturbulent fashion (Fig. 3.106). This approach enables one to dispense with fluxing, degassing, grain refinement, and modification procedures and reduce the subsequent heat treatment cycle to a very short time span compared to conventional practice. Extended holding (or dwell time) between the melting and casting under controlled atmosphere allows oxides and inclusions to separate by floating or sinking. The mold is permeable to allow air to escape from the cavity.390 It is equally beneficial when applied to permanent molds with or without cores. The Cosworth castings exhibit superior strength, ductility, and dimensional accuracy to gravity diecasting due to well-designed, high-quality tooling, a thermally stable high-purity zircon sand as mold material, programmable mold filling, consistent and predictable contraction, low casting temperature, high rate of solidification, positive metal feeding during solidification, need of minimal machining, and sound cast structure with consistent response to heat treatment. Additional advantages of the Cosworth process include rollover, weight savings, cost-effectiveness, recycling of zircon sand, excellent machining properties, and low tool wear rates.391,392 Applications include complex cylinder heads, engines of Formula 1 Grand Prix cars, MBA engines, cylinder blocks and Indianapolis CART racers, and highpressure flight-refueling manifolds. Weight of Al castings ranges between 0.2 and 55 kg (0.44 and 121 lb). The maximum mold size used in the Cosworth process has been reported to be 36 ¥ 24 in. (915 ¥ 610 mm). Castings for aerospace and defense include gas-turbine front-end components; fuel system pumps and controls; flightrefueling manifolds; weapon mountings; lightweight undercarriage components, transmission housings, and manifolds and ductings. Automotive castings include high-performance cylinder heads, air-cooled cylinder heads, marine cylinder heads, engine blocks, engine sumps, and transmission cases for racing engines and passenger cars, rally cars, and upmarket domestic cars.

3.12.6 Improved Low-Pressure (ILP) Casting Process The ILP process, developed in Australia, is a precision casting process that utilizes transfer of molten metal vertically through a riser tube to the bottom of the mold cavity. Degassed and filtered metal is delivered to the casting furnace which employs a pressurized N2 atmosphere to enable nonturbulent, computer-controlled filling of the mold. Once filled, the mold is sealed and immediately removed so that solidification takes place away from the casting section. This permits another preassembled mold to be placed in the station which facilitates high productivity with cycle times of around one minute.

Clean metal

(a)

(c)

FIGURE 3.106 (a) The Cosworth technology process involving (b) metal preparation and (c) metal transport. (Courtesy of Cosworth Technology Ltd., Northampton, England.)

3.162

SOLIDIFICATION

(a)

3.163

(b)

FIGURE 3.107 Schematic representation of ILP process: (a) cast and (b) solidify.393 (Courtesy of Comalco Aluminum Ltd.)

As shown in Fig. 3.107, a special feature of the ILP process is the use of a combination of resin-bonded silica sand for the mold and metal cores, which promotes rapid unidirectional solidification in those areas of castings requiring optimum properties. Moreover, the mold can be inverted to facilitate this controlled solidification, which may yield DAS > ru. This fact is supported by the experimental observation that the dislocation density immediately after the lower yield point is much greater than that at the upper yield stress.26

4.3.4 Yield Point Elongation and Lüder’s Band Formation Many metals, especially low-carbon steel, show a localized, heterogeneous type of transition from elastic to plastic deformation that produces a yield point. The load at which the sudden drop occurs in the load-elongation curve is called the upper yield point. The constant load is referred to as the lower yield point, and the elongation or stretching that occurs for a while without any increase in its flow stress (i.e., constant load and before the load starts to increase monotonically) is called the yield point elongation YPE, or ey (Fig. 4.6).27 The extent of YPE depends on the

PLASTIC DEFORMATION

4.17

FIGURE 4.6 elongation.27

Schematic showing yield point

FIGURE 4.7 Schematic loadtime curve showing two variations of Lüder’s band.28 (Reprinted by permission of Pergamon Press, Plc.)

grain size, strain rate, and temperature. The plastic deformation that occurs during YPE is inhomogeneous, discontinuous, small, and localized and is oriented approximately 55° to the tensile axis. This is called Lüder’s band, Lüder’s strain, or stretcher strain. Successive generation of Lüder’s band continues over the entire length of the specimen at a comparatively low and constant applied stress (i.e., at the lower yield point) (Fig. 4.6). Lüder’s band formation does not produce strain hardening until it is complete (i.e., when the YPE is complete). That is, when YPE zone or Lüder’s zone ends, work hardening is registered on the stress-strain curve. Lüder’s zone occurs in steels that are prone to strain aging, that is, strong interaction between solutes (e.g., C or N in iron and steel and Mg in Al), and dislocations that produce solute segregation to, and immobilization (pinning) of, dislocations. There are two variations of distinguishable Lüder’s band: (1) serrated yielding, that is, occurrence of plastic flow, where the leading edge of a single Lüder’s band propagates intermittently along the entire gauge lengths—this is characterized by a very uniform sawtooth stress-time plot (Fig. 4.7)—and (2) jerky flow, that is, occurrence of plastic flow at random locations along the specimen. When load serrations stop, a single Lüder’s band front passes from one end to the other end of the specimen. It is characterized by a randomly fluctuating stress-time plot (Fig. 4.7).28

4.18

CHAPTER FOUR

Discontinuous or inhomogeneous yielding and flow by Lüder’s band formation and propagation has long been a nuisance and is sometimes an unacceptable problem in the fabrication of complex shapes such as automobile doors and bumpers from low-carbon steels because of the surface appearance marred by Lüder’s lines. However, discontinuous yielding and strain aging have sometimes been proved to be advantageous, as in machining and paint-baking of low-carbon steels. With coarsening of grain size, the difference in upper and lower yield stress is lowered and the yield drop and Lüder’s band fronts become less pronounced (also mentioned earlier) and finally become diffuse and difficult to observe. In this case, ey decreases according to29 ey = YPE(%) =

k y d -1 2 a

(4.31)

where d is the grain size and ky and a are constants depending on the material and tensile testing conditions, respectively. Usually, both the yield stress and the YPE are increased by the presence of fine precipitates.30

4.4 FLOW STRESS The flow stress is defined as the stress required to maintain deformation (or cause the metal to flow plastically) at any given strain. Flow stress of many metals in the region of uniform plastic deformation can be described by the nth-power hardening equation (Ludwik-Hollomon) s = Ke n

(4.32)

where s is the true flow stress; e is the true plastic strain; and K and n are the strength coefficient and strain-hardening exponent (or index), respectively, and both are independent of temperature at sufficiently lower temperatures. A log s versus log e plot for the true stress–true strain curve up to maximum load provides the n value, represented by the slope of a straight line for most polycrystalline materials, as shown in Fig. 4.8.8 Tables 4.1 and 4.4 list the n values for several materials. Typically, the value of n, for most metals, varies between 0.1 and 0.56.

FIGURE 4.8 A log s versus log e plot for the true stress–true strain curve.8 (Reprinted by permission of McGraw-Hill, New York.)

PLASTIC DEFORMATION

4.19

TABLE 4.4 Values of n and K for Metals at Room Temperature1 K Metal 0.05% Carbon steel SAE 4340 Steel 0.6% Carbon steel 0.6% Carbon steel Copper 70/30 Brass

Condition

MPa

ksi

0.26 0.15

530 641

77 93

0.10

1572

228

0.19 0.54 0.49

1227 320 896

178 46.4 130

n

Annealed Annealed Quenched and tempered at 540°C (1000°F) Quenched and tempered at 705°C (1300°F) Annealed Annealed

Reprinted by permission of ASM International, Materials Park, Ohio.

The alternative equation suggested for another group of metals including some steels with a temperature-dependent yield stress is expressed as s = s0 + Ke n

(4.33)

where s0 is the yield stress and K and n are the same constants as in Eq. (4.32). It can be shown that in a tensile test, the limit of true uniform strain is equal to the strain-hardening exponent, that is, eu = n. The specimen will begin to neck at a point of maximum load (stress) on the engineering stress-strain curve. Hence, the specimen will neck down and fail when eu ≥ n. The true stress–true strain curve of fcc metals having low stacking fault energy, such as austenitic stainless steel, can be expressed by Ludwigson as s = Ke n + exp (K1) exp (n1e)

(4.34)

where K1 and n1 have been introduced as additional constants, exp K1 is nearly equal to the proportional limit, and n1 is the slope of the deviation from Eq. (4.32) plotted against e. The true strain term in Eqs. (4.32) and (4.33) precisely should be the plastic strain e p = e total - e E = e total -

s E

(4.35)

4.4.1 Strain-Hardening Exponent Wok hardening (or strain hardening) is a phenomenon whereby, during the deformation of metals at lower temperatures, the yield stress (required to continue) increases with increasing strain. Strain hardening is caused by the storage of dislocations within a metal and the resistance they offer to the passage of other dislocations. Strain hardening influences the cold formability of sheet. As a result of strain hardening, in many instances, an annealing treatment is required after each forming operation, to enhance the formability and obtain the desired deformation.31 The strain-hardening behaviors of fcc, bcc, NaCl with an abundance of equivalent

4.20

CHAPTER FOUR

FIGURE 4.9 Relationship between work- (or strain-) hardening exponent n and the yield strength. (National Steel Corporation.)

slip systems are different from those of hcp metals that do not undergo intersecting slip until very late in their deformation.15 The strain-hardening exponent (n value) is a measure of the ability of a material to resist localized straining and thereby increase uniform elongation. It is a very significant parameter in sheet metal forming. The final strength of a cold-worked part can be determined from the n value. Figure 4.9 can be used to approximate the n value of low-carbon steel sheet. The amount of eu, the level of the forming limit diagram (FLD), the strain distribution, and many other forming variables are directly related to the n value. The n value is not influenced by texture, but is reduced by the addition of solid solution elements in steel and by the refinement of ferrite grain size.32 The n values of aluminum alloy sheets decrease sharply with an increase in the tensile strain and are lower than those of steel sheets. The n value of commercial HSLA steels decreases with increasing strength, and its low n value makes it less formable than mild steel.33 There are different strain-hardening behaviors for mild steels and high-strength steels. The strain distribution ability of steels increases with the increase in overall n value. The peak n value at a low strain level increases the strain distribution ability of the steel. Aging of rimmed steels causes the n value to decrease with time. Additionally, excessive temper rolling beyond that required to eliminate YPE (ey) will also reduce the n value. For some materials such as dual-phase steels, some aluminum alloys, and so forth, the n value is not constant, as given by Eq. (4.32). In such cases, two or three

PLASTIC DEFORMATION

4.21

n values may be required to be determined for initial (low), intermediate, and terminal (high) strain regions. The initial n value relates to the low-deformation region, where springback is often a problem. The terminal n value relates to the high-deformation region, where fracture may occur.3 Recrystallized structures produce low yield strength and high hardening capacities and are ideal candidates for forming applications. In addition, low stacking fault energy, such as in brass, is advantageous because the difficulty of cross-slip leads to a high hardening rate. In certain stainless and high-strength steels, the decomposition of metastable austenite to martensite during deformation results in a very high n value. Because most engineering materials need high strength and good formability, a desirable means of strengthening them is by dispersion of hard, spheroidal phases in a soft, ductile matrix. This is also the basis for the dual-phase, ferrite-martensite (high-strength) steels used in the automotive industry. This type of microstructure provides a high initial hardening rate and allows improved shape fixability and less springback compared to high-yield-strength steels.34

4.4.2 Work Hardening and Dislocation Density It has usually been agreed upon by many investigators that the increase of flow stress is proportional to the square root of total dislocation density r .35–38 Since the density of fresh dislocations is much smaller than that of the immobile dislocations, we can write r = rimmobile. Many work-hardening theories, therefore, predict the following interrelationship between initial flow stress and dislocation density produced (in the cell walls), which has been verified by several experiments in metals with varying grain sizes and temperatures.39–41 s f = s 0 + aGb r

(4.36)

where sf is the applied flow stress at a given strain rate, s0 is the friction stress on dislocations arising from all sources except dislocation-dislocation interaction (e.g., grain boundary strengthening, solid solution strengthening), a is a numerical constant measuring the efficiency of dislocation strengthening (usually between 0.2 and 0.4), G is the shear modulus, and b is the Burgers vector. The flow stress, therefore, increases with increasing density of dislocations. The work-hardening rate depends on the rate at which dislocations increase with strain.40 Using Eq. (4.24) to represent the relationship between the density of immobile dislocations building up in the cell structure (to be discussed in the next section) of dislocation and plastic strain for the early stage, we finally get s f = s 0 + aGb r0 + Ke pm

(4.37)

Equations (4.36) and (4.37) predict a parabolic strain-hardening or stress-strain curve for a polycrystalline metal.

4.4.3 TEM Study of Dislocation Tangles and Cells after Work Hardening Heavily deformed (or cold-worked) metal at room temperature raises the dislocation density from 107 to 1012 dislocations/cm2. Both x-ray diffraction data and TEM studies suggest that the distribution of dislocation is not uniform throughout the deformed or cold-worked structure; but this leads to the formation of

4.22

CHAPTER FOUR

FIGURE 4.10 Cell structure of iron rolled at room temperature to a strain of (a) 9% and (b) 70%,25 6850X. (Reprinted by permission of John Wiley & Sons, New York.)

cell walls containing dense or tangled (immobile) dislocations (which represent the boundaries of a subgrain structure) that are surrounded by low-dislocationdensity regions called cells or subgrains (Fig. 4.10). The misorientations across these boundaries increase with the extent of deformation (or cold work); values of 2 to 6° are typical. The cell size in pure metals usually varies between 1 and 3 mm. The dislocation structure comprising cell walls and cells varies from metal to metal depending on their stacking fault energy, the amount, the type, and the rate of plastic strain (deformation) and temperature of deformation. As the temperature of deformation decreases below room temperature, the dislocations develop an increasingly random distribution, and subgrains are not formed. Moreover, dense, tangled dislocations are not formed in the dislocation structure below a certain plastic strain (i.e., below 3%). In polycrystalline iron the cell size decreases with increasing deformation up to about 10% strain. Then they maintain a stable cell size; further increase in plastic strain or flow stress increases the immobile dislocation density in the cell walls (Fig. 4.10),25,42 which makes these walls or tangles more effective obstacles to fresh dislocation motions. This increased flow stress causes the observed strain hardening and can be related to the average dislocation density by Eq. (4.36).43 We, however, find slip lines, deformation (transition or micro-) bands, and overall structural features at low magnification by using conventional microscopy. It is pointed out that the deformation bands are linked with particularly well-developed subgrain structures and marked misorientations across the bands.

4.4.4 Work Hardening and Grain Size The relatively large average grain size dependence of lower yield stress sl for polycrystalline metal is described by the classical Hall-Petch relationship as s l = s 0 + kd -1 2

(4.38)

PLASTIC DEFORMATION

4.23

where s0 and k are constants for the metal (the latter is often termed the Hall-Petch slope and is material-dependent) and d is the average grain size (typically, 1 mm or larger). This equation has wide applicability and is valid for bcc metals (including iron) as well as fcc and hcp metals. Figure 4.11a and b illustrates such results for various alloys of iron, aluminum, copper, and brass.44–46 This equation is also applicable to cellular substructures (i.e., subgrain boundary structures), which are normally produced by severe cold working. It may equally apply for small strains showing no yield and for strains well beyond the yield or beyond Lüder’s extension, if present. In that case, this equation can be related to dislocation density, and it can also be extended to higher strains in analogous form, in which sl is replaced by the flow stress s f = s 0¢ + k ¢d -1 2

(4.39)

where s0¢ is the frictional force required to move dislocations through the crystal and k¢ reflects the pinning effect of the grain boundaries.47 A wide variation in slope of Fig. 4.11b is due to factors such as solute pinning of dislocations (for increased slope) and cross-slip at the head of the pile-up (for decreased slope).48 For very small grain sizes, Coble creep is expected to be active and the s versus d relationship is given by sc =

A + Bd 3 d

(4.40)

where B is both temperature- and strain-rate-dependent. The additional (threshold) term A/d seems to be large if d is in the nanometer scale. For intermediate grain sizes, both mechanisms might be operative for the specimen containing a range of grain size distribution.49 A threshold of the form A/d has been suggested by S. M. Sastri.50 Note that a dislocation theory of the Hall-Petch effect only provides a linear relationship of s with d -1/2 when a large number of dislocations in a pile-up are present and plasticity is not source-limited. At sufficiently small grain sizes, the Hall-Petch model based upon dislocations may not be functional. In this case, Choksi et al.51 have suggested room-temperature Coble creep as the mechanism to explain their results. If it is assumed that a grain size d* prevails at which value the classical HallPetch mechanism converts to the Coble creep mechanism, sth = sc at d = d*, as shown schematically in Fig. 4.12.49 4.4.5 Relationship between Dislocation Density and Grain Size For a particular strain, the relationship between the dislocation density and grain size was first developed by Meakin and Petch52 based on the assumptions that the average slip distance Ls is approximately equal to grain diameter d and that the strain in tension e equals rbLs. Thus, by inserting the dislocation density in the hardening equation (4.36), we get46 s f = s 0 + aGb1 2 d -1 2

(4.41)

This equation is different from the Hall-Petch equation, (4.38), in the sense that here s0 is strain-independent. Ashby53,54 has proposed that when metals or alloys are deformed, a deformed grain consists of a uniform deformation (by slip on one system) and a local nonuni-

Lower yield stress 24 22 20

300

18

tons in–2

MNm–2

16

200

14 12

4

10

3 5 2

8 1

100

6

6

4 0

1

2

3

4 5 d –1/2 (mm–2)

6

7

8

9

(a)

300 RT

σ - TRUE STRESS (MPa)

Fe

200

70/30 Brass 100

Cu AI 0

0

2

4

6

8

10

12

14

d –1/2 - (GRAIN SIZE)–1/2 (mm–1/2) (b)

FIGURE 4.11 (a) Dependence of lower yield stress on grain size for (1) annealed mild steel (En2); (2) En2, nitrided; (3) En2, quenched from 650°C; (4) En2, quenched and aged for 1 hr at 150°C; (5) En2, quenched and aged for 100 hr at 200°C; (6) annealed Swedish iron.44 (En2 steel contains 0.115% C and Swedish iron, 0.02% C.) (b) Yield stress—grain size relationship at room temperature for Al, Cu, 70-30 brass, and Fe.46 [(a) Reprinted by permission of Pergamon Press, Plc. (b) Reprinted by permission of The Metallurgical Society, Warrendale, Pa.]

PLASTIC DEFORMATION

4.25

FIGURE 4.12 Combined Hall-Petch response and Coble creep. The critical d* where transition occurs is represented by dashed line.49

form deformation in the grain boundary area. The uniform deformation causes the work hardening within the grain due to accumulation of dislocation density rs of statistically stored dislocations. The accumulation of geometrically necessary dislocation density rg causes the grain-size-dependent (part of stress-strain curve) work hardening (near the grain boundary). Accordingly, Ashby developed the following equations:53,54 rtotal = rs + rg

(4.42)

rs =

C1e bLs

(4.43)

rg =

C 2e bd

(4.44)

where C1 and C2 in Eqs. (4.43) and (4.44), respectively, are constants. Equation (4.43) is independent of grain size contribution, while rg is dependent on the grain size. The flow stress thus depends on the total dislocation density in the following form: s f = s 0 + aG e

C1b C2 b + Ls d

(4.45)

If grain boundary strengthening is predominant, we have s f = s 0 + aG e

C2 b d

(4.46)

If the grain boundary contribution is small (e.g., in coarse-grained materials), Eq. (4.45) reduces to

CHAPTER FOUR

4.26

s f = s 0 + aG e

C2 b Ls 1 2 C1b d

(4.47)

Equations (4.46) and (4.47) illustrate that the flow stress at a constant strain may be a function of d-1/2 or d-1 according to the magnitude of the grain size contribution.46

4.4.6 Effect of Temperature Temperature has a significant effect on the stress-strain curve of polycrystalline iron when compared with the stress-strain curve obtained at room temperature. When the temperature of deformation is decreased, the upper yield point rises considerably; the yield drop and yield point elongation region both become progressively larger. This has also been established in other bcc metals (e.g., Mo, Nb, and Ta).55 At low temperatures the grown-in dislocations are firmly locked by interstitial impurities; consequently a quite high stress is required to produce new and mobile dislocations. Hence the temperature dependence results mainly from the frictional stress. It has also been experimentally evidenced that frictional stress increases substantially with interstitial solute (C, N) concentrations in the range of 0.001 to 0.3 wt%. In contrast, when the temperature of deformation is raised above room temperature, the upper and lower yield points as well as the yield point elongation region slowly disappear and are replaced by serrated curves (Fig. 4.13) (see Sec. 4.6.1 for more details).56

Stress

105 °C 138 °C 142 °C

(a)

168 °C

(b) (c)

201 °C

10 kg.mm–2

100 MN.m–2

(d)

226 °C 236 °C

(e) (f) (g) Strain

1%

FIGURE 4.13 Stress-strain curve for polycrystalline mild steel at increasing temperature. (Reprinted by permission of The Institute of Metals, England.)

PLASTIC DEFORMATION

4.27

4.5 COLD WORKING Cold working plays an important role in the latter stages of manufacture of steel products, as it offers greater control of dimensions, surface finish, and properties. The terms cold working and annealing are closely linked because the subtle sequence of these operations allows material properties to be closely controlled. At the microscopic level, the original ferrite grains are retained during cold working but are prone to drastic shape changes, representing the macroscopic shape changes linked with the working operation. Thus, in cold rolling the grains assume elongated pancake shape and result in a deformation texture or preferred orientation (see Sec. 4.8). The slip mechanism associated with the plastic deformation of the ferrite crystals in steel generates crystal defects in excess of the thermodynamic equilibrium concentration. The progressive increase of these defects, particularly dislocation density, differentiates cold working from hot working, because in the latter the dislocations are removed spontaneously during the operation. Deformation twinning also occurs during plastic deformation at low temperature and high strain rates, which is discussed in the next section. The dislocations created during cold working at room temperature and above form tangled networks, which represent the boundaries of a subgrain structure (Fig. 4.10). The process of cold working is associated with the phenomenon of work hardening. As a result of cold working of iron and steel, the tensile strength, yield strength, hardness, and electrical resistivity, as well as the susceptibility to intergranular corrosion, are increased while the ductility, density, and magnetic permeability are decreased.

4.5.1 Stored Energy of Cold-Worked Materials When a metal is plastically deformed by cold working, most of the mechanical energy is released as heat, but a small fraction of this energy is retained as stored energy inside the metal grains, thereby raising its internal energy. Thus the stored energy is the change in internal energy produced by plastic deformation.57 The stored energy is mainly in the form of elastic energy in the strain fields of lattice defects (e.g., interstitials, vacancies, dislocations, and stacking faults). That is, this stored energy exists in the metallic grains (or crystals) as point defects, dislocations, and stacking faults. Thus, a heavily deformed metal, being in a stage of higher energy, is thermodynamically unstable. It has been indicated by many workers that the ratio of stored energy Es to the total energy of deformation Ew varies from 2 to 10%. In general, the amount of stored energy and the structure of a deformed material depend on many variables: impurity content or composition of the metal or alloy; type or process of deformation; deformation temperature; grain size; deformation (or strain) rate; and other factors.56 An increase in impurity content hinders the dislocation motion, which enhances the dislocation density (or multiplication) and the stored energy for the same amount of strain. The complex deformation process (e.g., extrusion and wire drawing) usually activates many slip planes when compared to the simple process (as in tension); with pronounced dislocation intersection, this leads to cross-slip and high dislocation density. Lower temperature of deformation or increase in strain rate produces an increased amount of stored energy and dislocation density. Fine-

4.28

CHAPTER FOUR

grained materials cause high levels of stored energy and dislocation density for a given state of strain.

4.5.2 Effect of Variables on the Amount of Stored Energy The amount of stored energy depends on purity, extent, rate, type and temperature of deformation, and grain size, as described below:57 1. Purity. The increased impurity content in a metal increases the amount of stored energy. Moreover, it causes a significant change in the kinetics of its release during recovery and usually raises the recrystallization temperature.57 These effects apparently arise from the pinning of dislocations by impurity atoms and subsequent dislocation multiplication. 2. Extent, rate, type, and temperature of deformation. The amount of stored energy increases with the extent of plastic deformation, for a particular metal and deformation process. It has been suggested that more energy is stored in the fast primary deformation process than in the slow one. This is attributed to a change in heat losses arising from varying time required for deformation. Simple deformation processes of tension, compression, and torsion produce lower stored energy. As the deformation becomes more complicated (e.g., wire drawing, explosive shock loading, etc.), the level of stored energy increases. This is presumably due to (enhanced) slip activity on all possible slip planes. The stored energy increases with decreasing temperature of deformation as a result of the formation of smaller cell diameter and larger dislocation density. That is, the rate of work hardening is larger at a lower temperature.57 3. Grain size. The fine-grained material stores more energy than the coarsegrained material. This is due to the increased grain boundary–dislocation interaction and multiplication of dislocations in the deformed fine-grained material.

4.6 STRAIN AGING Strain aging is the term used to characterize the time-dependent strengthening process in plastically deformed metals and alloys by (1) strong elastic interactions (and high diffusivities) of interstitial solute atoms (e.g., C and N) with mobile dislocations and point defects and (2) allowing solute atmosphere (or cloud) formation around dislocations due to their high diffusivities at temperatures as low as ambient during or after straining or plastic deformation. Aging reactions that occur after plastic deformation (or press-forming operations) are grouped as static strain aging (where plastic deformation and aging are separated in time) and dynamic strain aging (where both plastic deformation and aging processes occur concurrently).58 Static strain aging can take place when the concentration of solute atoms is low and deformation temperature is adequate to allow low-range diffusion of solute atoms. After aging, higher stress levels are needed to produce further straining of the material, either to nucleate fresh dislocations or to pull the dislocations free from these atmospheres. Dynamic strain aging can take place at deformation temperatures below ambient when the concentration of solute atoms is high, or at temperatures above ambient when the solute concentration is low.59

PLASTIC DEFORMATION

4.29

4.6.1 Static Strain Aging In practice, strain aging is of great importance in press-formed high-strength lowcarbon (or bake-hardening) steels during paint baking where strength anisotropy is strongly developed. The strain-aging rate also depends on the extent of deformation and increases when the deformation occurs at elevated temperatures or lower strain rates. Usually, about 15% reduction in thickness gives the maximum effect. Strain aging is common in sheet and plate steels and sheet-metal forming. It is undesirable in deep-drawing steels. Figure 4.14a schematically describes the effects of static strain aging on the flow stress of a low-carbon steel.8 In continuous wire-drawing operations, strain aging can develop to such a large extent that, in extreme cases, the stress generated by the reduction of the last die leads to splitting and cracking in the center of the wire. For small amounts of strain aging, a reduction in torsional properties of otherwise defect-free wire will occur. Strain aging can also develop erratic behavior in a forming operation because of the erratic variation of the yield strength of the wire at elevated temperatures.60 Static strain aging in drawn pearlitic steels greatly influences the drawability and properties of the finished products such as ropes and cables. In addition, higher susceptibility to central burst formation and to delamination is also observed. Static strain aging in a drawn pearlitic steel wire is controlled by the decomposition of cementite. This appears to occur in two distinct stages, each linked to different atomistic mechanisms. The first stage occurs at low aging temperature interval (between 60 and 100°C) and is characterized by a slight increase in yield strength and a decrease in electrical resistivity and background damping. The second stage of aging takes place at higher temperatures (between 120 and 200°C) or longer aging times and is exhibited by a considerably larger increase in yield strength and an increase in electrical resistivity, whereas background damping reaches very small values.60a Note that the second stage of aging is not reported in low-carbon steels.60b Certain coating treatments such as hot-dip galvanizing can yield a greater extent of strain-aging embrittlement in areas that were cold-worked to the critical amount eC; this can lead to brittle fractures. This problem can be eliminated by annealing the part before it is galvanized. Alternatively, strain aging can be controlled by using interstitial-free steels which are Al-killed and contain e *) and very low temperatures (T < T*), C and N atoms are unable. to .segregate to the moving dislocations because ¯d > Vs. At very low strain rates (e < e *) and very high temperatures (T > T*), any V solute atmosphere formed can easily move with the dislocations, and deformation occurs without DSA effects.30,64 DSA caused by interstitial elements is reported to be irregular and finds difficulties to be systematically analyzed. However, five types of serrated flow curves in the deformation of substitutional solid solution alloys have been observed which are attributed to DSA effects. These serration types are termed A, B, C, D, and E (Fig. 4.15a). Type A is characterized by periodic or repeated appearance of deformation bands and occurs after a low strain. This is observed in Al-Mg alloys deformed at room temperature or a mild steel deformed in blue brittle temperature range (⬃175°C). Type B refers to the stress oscillations around a constant level of the s-e curve. It is sometimes called Lüder’s or PLC bands and occurs at a specific angle (⬃55°) to the specimen tensile axis, which may be influenced by the rolling texture.65 Type C serration is recognized from load or yield drops below the prevalent level of the flow curve, and type D from peak or plateau in the s-e curve due to band propagation. Type E serrations usually appear at high strains or near the maximum load.64,65 The increase in flow stress and strain-hardening rate during DSA results partly from the increased migration and locking of solute atoms to dislocations (Fig. 4.15b); also a reduced dislocation annihilation occurs.66 Precipitates and atmosphere drag also enhance the hardening. When the low-carbon steel is strained at a normal tensile strain rate of about 1.75 ¥ 10-4 m/ms, DSA effects occur in the stress-strain curve at temperatures between 100 and 250°C, with very low total elongation and high tensile strength (Fig. 4.15c). If the interstitial solute content is substantial, DSA can be observed at room temperature. At very high strain rate, as in impact testing, these same limits take place between 350 and 650°C.63,64 In steels, clustering of N atoms may occur in solution with substitutional atoms such as Mn; this decreases their diffusion rate and, therefore, shifts these DSA effects to higher temperatures.66 DSA in steel wire is a function of steel composition, drawing strain, highest temperature attained during wire drawing, and the extent of time at that temperature.

E

A B Stress σ

εc (A) εc (B)

C εc (C) > εc (A,B) D A

B

D with B Strain ε (a)

200 °C

ρ cm–2 × 10–9

25

15

RT

5

0.04

0.08

0.12

0.16

γ (b)

b 400

σ N / mm2

a 300

200 FIGURE 4.15 (a) Types of serrations due to DSA.64 (b) Dislocation density r versus plastic shear strain g for a 0.035% C steel deformed at room temperature and 200°C. (c) DSA of a mild steel. Nominal stress versus elongation at (a) room temperature and (b) 150°C.66

100

0.1

e

0.2

(c)

4.32

PLASTIC DEFORMATION

4.33

Table 4.5 summarizes the effects of DSA on the room-temperature properties of five steels, such as increased strength, YS/UTS ratio, fatigue life and the DBTT, and decreased ductility and notch impact toughness.63 In polycrystals, the DSA-like serrations of flow stress and work hardening become more pronounced with decrease in grain size. It has been suggested that grain boundary regions are preferred sites for DSA and for consequent increased rate of dislocation accumulation. DSA has been found in a large number of dilute interstitial and substitutional solid solutions as well as in most of the dilute and substitutional concentrated commercial alloys. It is also observed in specimens subjected to uniaxial tension, compression, and torsional loading. Whether due to solute atmosphere, ordered regions, or precipitates, it is clear that DSA can be a very effective method of strengthening61 and can be used in combined forming and strengthening operations in the DSA temperature range.59 It is of importance to study the effect of DSA on mechanical behavior such as negative strain rate sensitivity (SRS) and unstable plastic flow (or strain localization). These phenomena are interrelated with mechanical anisotropy and may influence the formability at room temperature.64a DSA effects in drawn pearlitic steels (1) are characterized by lower intensities than those found in low-carbon steels, (2) start or attain their maximum values at higher temperatures, and (3) are controlled by the decomposition of cementite, as evidenced by different techniques such as Mössbauer spectroscopy and neutron diffraction.60b . The dependence of kinetics of DSA is expressed with the strain rate e as -Q e˙ = Kr m b exp RT

(4.49)

where K is a constant, rm is the density of mobile dislocations, b is the Burgers vector, Q is the activation energy involved in the process responsible for the PLC effect QPLC or for the maximum stress Qmax, R is the universal constant, and T is the absolute temperature. Blue Brittleness. One of the well-known manifestations of DSA is the blue brittleness phenomenon which occurs during plastic deformation at a strain rate of 10-3 m/ms in the blue-heat range of 230 to 370°C (or 450 to 700°F). This shows an extremely low tensile ductility and notch-impact resistance and high tensile strength. The onset of necking takes place when the rate of strain hardening equals the flow stress. Baird and Jamieson have observed that the loss of ductility does not occur with high N levels (>0.03%).66 Carbide- and nitride-forming elements are usually added to prevent blue brittleness.27,63

4.7 DEFORMATION TWINNING Mechanical or deformation twinning has long been accepted as a second important mode of plastic deformation in crystalline solids that have a deficient number of independent slip systems. Recently, it has been recognized that deformation twinning greatly influences the strength and ductility of some interesting materials.67 The actual extents to which deformation twinning can influence plastic flow in polycrystalline solids appear to depend on the amount of total strain associated with twinning. Deformation twinning has been found to be formed in many bcc, hcp, and lower-symmetry metals and alloys; in many fcc metals and alloys; in ordered alloys

TABLE 4.5 Changes in Charpy V-notch Impact Properties Due to Dynamic Strain Aging63

Shelf energy Steel

J

ft ◊ lb

Transition temperature at 10.9 J (8 ft ◊ lbf)

FATT, 50% ductile

°C

°F

°C

°F

Yield strength Grain size, mm

MPa

ksi

4.34

1008† As-rolled Prestrained 3%, 250°C (480°F)

26 21.5

19 16

-65 -40

-85 -40

-75 -45

-103 -49

14 14

262 372

38 54

1020† As-rolled Prestrained 3%, 250°C (480°F)

21.5 18

16 13.5

-55 -10

-67 14

-50 -25

-58 -13

14 14

352 507

51 73.5

1035† As-rolled Prestrained 3%, 250°C (480°F)

17.5 13.5

13 10

-35 +3

-31 37

-20 -5

-4 23

13 13

379 576

55 83.5

1522† As-rolled Prestrained 3%, 250°C (480°F)

26 24

19 17.5

-70 -50

-94 -58

-55 -39

-67 -38

10 10

386 503

56 73

1010 renitrogenized‡ As-rolled Prestrained 3%, 250°C (480°F)

24.5 20.5

18 15

-63 -52

-81 -62

... ...

... ...

10 10

331 427

48 62

FATT: fracture appearance transition temperature. † Specimens 5 ¥ 5 ¥ 55 mm (0.2 ¥ 0.2 ¥ 2.2 in). ‡ Specimens 2.5 ¥ 10 ¥ 55 mm (0.1 ¥ 0.4 ¥ 2.2 in).

PLASTIC DEFORMATION

4.35

and other intermetallic compounds; in elemental semiconductors and compounds; in nonmetallic compounds such as calcite and sodium nitrate; and even in complex minerals and crystalline polymers.14,68–74 Table 4.6 lists the twinning systems in some important metals and alloys.14,68 Twinning occurs when a portion of the crystal takes up an orientation that is related to the orientation of the untwinned lattice in a distinct symmetrical manner. The twinned portion of the crystal forms a mirror image across the twinning plane (also called composition plane) of the parent matrix. The following are the characteristics of the deformation twins:15,27,74–76 1. It is produced by a sudden localized shear process and involves a small, but well-defined, plastic deformation, in contrast to the apparently chaotic process of formation and growth of formation and growth of slip bands during glide deformation.68 2. In the twinned portion, realignment of atoms within a finite crystalline volume of the parent phase reproduces the original symmetry and crystal structure, but with a different orientation. 3. Deformation is a pure shear (Fig. 4.16)71 in which a cooperative atomic movement occurs; that is, individual atoms move by a fraction of the lattice interatomic spacing relative to each other. It produces a change in shape in the form of a small lens or plate; the parallel (or nearly parallel) sides represent the low-index planes, called a twin habit plane or twinning plane. This distorts the surrounding matrix. 4. Twins are visible after repolishing and etching the sample.77 5. The shear strain g associated with twinning is homogeneous (i.e., uniformly distributed), in which every adjacent atomic plane in the twinned portion is involved. 6. Its formation occurs very rapidly (in microseconds) (often with an audible click sound and with a sharp drop in yield) in materials with even moderate plastic slip resistance. Its formation consists of two parts: the nucleation of a twin nucleus and development or growth of the twin.69 Twin initiation stress is much larger than the stress required to propagate a preexistent twin. 7. Deformation twinning is always accompanied (or preceded) by the formation of some microslip, even though this slip may be difficult to recognize.68 Greater deformation (48 to 75%) in fcc alloys results in the formation of deformation twins which increase in density with cold working. 8. Twins usually form easily in large grains. Twin thickness is a function of grain size; for example, thicker twins form in coarse-grained specimens.68 Deformation twins have imperfect structures comprising many stacking faults. In many cases, the twinning shear can only transform a sublattice relative to the exterior, and reposition all other atoms properly within the twinned region, requiring internal atom switches which are called shuffles.68 9. There are a number of important geometrical differences between twinning and slip. Like slip, the occurrence of twinning depends on the least stress to be initiated in the crystal undergoing shear. The flow stress levels required for slip or twinning behavior of a particular material are not constant, but depend on the crystal structure of the metal or alloy, temperature of deformation, strain rate, mode of deformation, microstructure (including the grain size, texture, and presence of second-phase precipitates or dispersoids), chemical composition, and prestrain (or prior deformation).68 10. The nucleation of a twin is accompanied by a sudden load drop (responsible for serrated stress-strain curve), while the growth of a twin displays smoother behavior.77

TABLE 4.6 Deformation Twining Elements in Some Important Metals and Alloys5,14 Metal

K1

h1

gr

Notes

Face-centered cubic Cu

(111)

[112¯]

0.707

Only recrystallization twins (with these elements) occur in Ag, Al, Au, g -Fe, and Co

Body-centered cubic Cr, a-Fe, Mo, Na W

(112) (441) (332) (112)

[111¯]

0.707

[111¯]

0.707

Diamond cubic Ge

(111)

[112¯]

0.707

Hexagonal close-packed Be Cd Mg Ti Zn Be Mg Ti

Bi Hg Sb

a-Zr

(101¯2)

[1¯011] [101¯1¯] [1¯011] [1¯011] [101¯1¯]

0.199 0.171 0.129 0.189 0.139

(101¯1) (101¯3) (101¯1) (112¯1) (112¯2) (112¯3) (112¯4) (303¯4)

[1¯012] [1¯1¯26] [112¯3] [1¯1¯22] [2¯2¯43] ·202¯3Ò

1.066 0.638 0.957 1.194 0.468 ?

(101¯3)* (1¯012) (1¯012) (1¯012) (101¯2) (112¯1) (112¯2) (112¯3)

·303¯2Ò* [101¯1] [101¯1] [101¯1] ? ≤ ≤ ≤

? 0.118 0.447 0.146 ? ≤ ≤ ≤

c/a = 1.568 c/a = 1.886 c/a = 1.624 c/a = 1.587 c/a = 1.856 Additional forms Ditto ≤ ≤ ≤ ≤ ≤ ≤ ≤ * 150 and 268°C

Tetragonal

b-Sn In

(301) (101)

a-U

(17¯2) (112) (121) (111) (130)

[1¯03] [101¯]

0.119 0.150

c/a = 0.541 c/a = 1.078

Orthorhombic [312] X[37¯2] X[100] [123¯] [31¯0]

0.228 0.228 0.329 0.214 0.299

X = irrational twin

PLASTIC DEFORMATION

4.37

4.7.1 Crystallography of Twinning Figure 4.1671 represents an atomic rearrangement where twinning occurs along the twinning plane, called K1 in the twinning direction. The open circles are the original atomic positions, whereas the black circles represent the new atomic positions after the twin displacement. The formation of mechanical twinning is shown schematically in Fig. 4.17 when a shear stress is applied to the single-crystal specimen as shown in Fig. 4.17a. The crystal deformed into the final form is shown in Fig. 4.17b. The nature of the atomic movement occurring in twinning is shown in Fig. 4.17c. The atomic rearrangement in the twinned portion occurs in such a way that it preserves the original crystal structure symmetry; that is, the size and shape of the unit cell remain unchanged. According to crystallographic theory, it is possible to retain the size or shape of the unit cell only when three noncoplanar, rational lattice vectors in the original crystal having the particular length and mutual angles remain unchanged after shear

IIK2 h2

IIK1 h1

FIGURE 4.16 Atomic arrangement of twinning transformations.71 (Reprinted by permission of Butterworths, London.)

FIGURE 4.17 Twin formation by a shear stress. (a) Application of shear stress. (b) Twin formation. (c) Shear deformation.75 (Reprinted by permission of John Wiley & Sons, New York.)

4.38

CHAPTER FOUR

deformation. For the determination of the first criterion (i.e., undistorted planes upon transformation), let us consider a spherical single crystal that has been sheared to produce a twinned region in the upper hemisphere and an untwinned region in the lower hemisphere (Fig. 4.18).78 Shear stress has transformed the hemisphere to an ellipsoid in which all planes joining the twinning plane and a point on the original matrix have moved to a position after twinning and have suffered a change in shape and length except for a single plane (shaded area in Fig. 4.18) above the twinning plane. Its length and shape remain undistorted upon twinning because it makes the same angle q with the twinning plane. It is obvious, from the above, that only two crystallographic planes in a shear transformation have remained unchanged in size and shape. The first is called the twinning plane or the first undistorted plane (equitorial plane) and is designated by the symbol K1. The other plane (shaded) intersects K1 in a line that is perpendicular to the shear direction and makes equal angles with K1 prior to, as well as after, the occurrence of shear transformation. This is called the second undistorted plane and is crystallographically designated as K2. The plane of shear is the plane that contains both the perpendicular to the twinning plane K1 and shear direction h1. The intersection of the plane of shear with the second undistorted plane K2 is the second shear direction h2, which is represented by an arrow. Two shear directions h2 and h¢2 correspond to the positions of K2 and K¢2 before and after twinning, respectively. The second criterion for twinning transformation is that the mutual angles between the second undistorted plane K2 and first undistorted plane K1 must be retained upon transformation. If we assume h to be the width of the twin, s the magnitude of shear displacement as represented in Fig. 4.18, and q the angle between K1 and K2 (or K¢2), then the shear strain or shear g can be related to the angle subtended between K1 and K2 by

FIGURE 4.18 Schematic diagram showing the transformation of a sphere into an ellipsoid by twinning.

PLASTIC DEFORMATION

g =

4.39

s = 2 tan(90 - q ) = 2 cot q h

(4.50)

Thus the shear strain becomes fixed and can be easily determined after measuring the mutual angle between two undistorted planes.76 If we consider any vector e in the plane K1 (Fig. 4.19), there remains only one vector h2 in plane K2 which makes the same angle a with e before and after twinning. This arises because h2 is the only vector perpendicular to the intersection of two planes K1 and K2. This condition is applicable to e and all other vectors lying in K1. Finally, if it is assumed that K1 is a rational plane containing rational directions and that h2 is a rational direction, then the requirements of having these noncoplanar vectors unchanged in shape and size can be realized. In this manner, it can be shown that h1, being perpendicular to the intersection of K1 and K2, is the only vector in plane K1 that makes the same angles with all other vectors lying in plane K2 before and after twinning. It can, therefore, be concluded that another condition for preserving the crystal structure during twinning is that direction h1 and plane K2 be rational. There is also a third possibility that the two planes K1 and K2 and the two vectors h1 and h2 are all rational. The classical theories of the crystallography of deformation twinning define a particular twinning mode. Therefore, in summary, there are three modes of lattice shear for retaining its symmetry and crystal structure, thereby satisfying the conditions for twinning: 1. A twin of the first kind (when K1 and h2 are rational plane and direction) 2. A twin of the second kind (when K2 and h1 are rational plane and direction) 3. A rational or compound twin (when all four elements K1, K2 planes and h1, h2 directions are rational) It is apparent that the lattice rotation produced during twinning will depend upon whether twins of the first or second kind or compound twins are formed.26,76 The theory assumes that the twin interface is planar whereas, in reality, twin boundaries contain linear steps which have the characteristics of dislocations. Twins, therefore, are controlled by the properties of twinning dislocations.79 4.7.2 Twin Boundary Interface Twin boundary may be classified as coherent, semicoherent, or incoherent. In the former case, there is a perfect one-to-one matching (i.e., registry) of the lattice

FIGURE 4.19 first kind.

Twinning of the

4.40

CHAPTER FOUR

FIGURE 4.20 Coherent (WV) and semicoherent (VU) twin boundaries; the latter contains partial dislocations.80 (Reprinted by permission of Cambridge University Press, New York.)

planes at or across the twin boundary WV (Fig. 4.20).80 This is also a special lowenergy grain boundary. In the semicoherent case, the twin boundary plane VU rotates off the symmetry plane (as shown in Fig. 4.20), which results in the disregistry d of the boundary. This contains partial dislocations as shown in the figure. The disregistry or misfit is defined as d=

ap - am am

(4.51)

where ap and am are the lattice parameters of the twinned and matrix lattices, respectively. Semicoherent boundaries include low-angle grain boundaries and may be further divided into glissile and nonglissile. Incoherent boundaries will be discussed in Chap. 5.

4.7.3 Twinning in BCC Materials In bcc metals such as Fe, Nb, or W, the twinning plane K1 is usually [112] and the shear direction h1 is (or equivalent planes and directions).68 Mechanical twins (or Neumann bands) are generally observed in polycrystalline pure iron (or a-iron), Fe-Si alloys, and Fe-Cr alloys81 deformed at low temperatures. They are more predominant in coarse-grained than in fine-grained iron and other bcc metals. In a-iron (and other bcc metals) the twins form on {112} planes with a shear of 0.707 along the direction.43 The geometry of the process is shown in Fig. 4.21.78 Figure 4.21a shows a (11¯0) section through a bcc structure. The twinning planes {112} packed in the stacking sequence of ABCDEFAB . . . is normal to the (11¯0) plane whose trace is shown on the diagram. The twinning shear is 1/ 2 in a direction on a {112} plane. This can be produced by the shear displacement of 1/6 partial dislocation on every successive {112} plane. This indicates that the bcc twinning results from the propagation of partial dislocations 1/6 on every {112} plane.70 Figure 4.21b shows the movement of a set of twinning dislocations, lying along XY, to a portion of the crystal which produces a twin-oriented region. Figure 4.21c illustrates the completely twinned crystal where the twinning dislocations have crossed the whole crystal. This type of dislocation has usually been observed in airon, Fe-Si alloys, and Fe-Cr alloys deformed at low temperatures. Mechanical twins form in bcc metals as long and thin plates—usually not more than 5 mm thick—because of the high shear strain energy involved;26 bcc twinning may be induced by a high deformation rate (e.g., impact loading or explosive shock

FIGURE 4.21 Schematic representation of twinning in a body-centered lattice. (a) Section parallel to (11¯0) showing the stacking sequence of (112) planes, X-Y is a row of 1/6 [111] twinning dislocations, one dislocation on every (112) plane. (b) The twinning dislocations have moved halfway across the crystal to produce a twinoriented region, changing the stacking sequence. (c) A twinned crystal.78 (Reprinted by permission of Pergamon Press, Plc.)

4.41

4.42

CHAPTER FOUR

FIGURE 4.22 Stress-strain curve in a single crystal showing load serrations due to twinning.76 (Reprinted by permission of PWS-Kent Publishing Co., Boston, Massachusetts; after Schmidt and Boas.)

loading at room temperature). In the conventional (slow) deformation process, stress required to produce twinning is usually larger than that for slip at room temperature, and in this case slip precedes twinning. The stress-strain curve for iron may show the resultant serrated yielding due to twinning after the occurrence of plastic deformation by slip74 (Fig. 4.22). However, the tendency to twin increases with falling temperature of deformation at normal or slow strain rate because critical shear stress for slip deformation increases, and then the deformation twinning will occur with ease at stresses much below that for slip. This is experimentally supported by the fact that twin formation has been found to occur at evidently lower stress than the flow stress for slip in Fe-Be alloys and pure iron at 4.2 K and Fe-10% Cr alloy at 112 and 147 K.81 The addition of Be, Si, and Cr to iron causes deformation twinning to occur more prominently.

4.7.4 Twinning in FCC Materials In fcc metals such as Ag, Cu, Al, and Ni, the twinning mode is [111]. That is, the twin plane is {111} slip plane having the stacking sequence ABCABC . . . , while the twinning direction is . The twin formation changes the stacking sequence into Ø ABCABACBA ≠ The arrows denote the twin boundary beyond which the crystal has opposite stacking to that in the untwinned portion.24 In high-stacking fault energy (SFE) fcc pure metals such as pure Cu or Ni, deformation twinning can only occur at high stress levels (e.g., 150 MPa for Cu and 300 MPa for Ni), which are usually reached at low temperatures (or under heavy deformations). Many high-SFE metals such as Al-4.8% Mg, 6061-T6 Al alloy, and Cu-5 at% Ge crystals that do not display twinning during conventional low-strain deformation, easily twin under severe stress levels developed by shock deformation.69 In these cases, a subdivision of a

PLASTIC DEFORMATION

4.43

FIGURE 4.23 Bundles in brass with 15% Zn rolled to 37% reduction viewed in longitudinal section (with rolling direction indicated).83 (Courtesy of E. Aernoudt, P. V. Houtte, and T. Leffers.)

large fraction of grains by dense dislocation walls (DDWs)/first-generation microbands (MB1s) into cell blocks (CBs) has been observed. For Al, a specific macroscopic role of the DDWs/MB1s for the fraction of the mechanical anisotropy has been suggested.82 When fcc metals are highly alloyed (e.g., Fe-Ni-Co-Cr-Mo alloys), they develop low SFE, which gives rise to easy nucleation of fcc twins by the dissociation of perfect dislocations. Here the twins form in high density on a fine scale when heavy deformation is effected. These twins subdivide the matrix and are functionally equivalent to finer grain size in that they act as strong obstacles to dislocation motions.80,82 In some cases, the microstructure (e.g., of brass with 15% Zn) is dominated by twin lamellae and bundles of twin lamellae, separated by matrix without twins (Fig. 4.23).83 In Hadfield steel (12 wt% Mn, 1 wt% C steel), very high flow stress can be developed by the deformation at low temperatures. The occurrence of deformation twinning in compression, together with the continued subdivision of the original austenite matrix into finer domains by the twin lamellae and the high density of dislocations within the twin boundaries, gives rise to a high workhardening rate. For deformation (or annealing) twins, it has been shown and observed that coherent twin boundaries act as strong barriers to dislocations. Most of the extra hardening due to twinning is, therefore, attributed to the existence of coherent twin boundaries.57

4.7.5 Twinning in HCP Materials In hcp metals, the pyramidal [101¯2] planes are the most common twinning system. In hcp and other low-symmetry crystals, a simple shear will not satisfy the twin relationship, causing some atoms to be slightly out of atomic positions. In addition to the shear, it is necessary to shuffle local arrangements of some of the atomic positions or undergo secondary twinning, as in the case of hcp metals, to complete the twinning relationship.69 Research on several hcp metals has proved that the stress for twinning on {101¯2}, {112¯1}, and {112¯2} planes increases with increasing temperature whereas that for {101¯1} twinning decreases. The increase in stress for {112¯2} twinning with increasing temperature may suggest that slip is a prerequisite for this twinning

4.44

CHAPTER FOUR

behavior in a-Ti.69 In hcp metals and alloys such as Zr and Ti-6Al-4V, very high strain rates (by shock loading or explosive deformation) often lead to {112¯ 1} deformation twins.

4.8 DEFORMATION TEXTURE When a polycrystalline metal is severely deformed by rolling, deep drawing, swaging, or wire drawing, certain crystallographic directions or planes of the majority of individual crystals or grains tend to rotate themselves in a non-random or preferred orientation with respect to the direction of deformation. This preferred (grain or crystal) orientation is usually called the deformation texture or cold-worked texture. Thus, texture introduces anisotropy in mechanical properties of metals and alloy sheet materials.84 It produces peaking or earing (Fig. 4.24) and nonuniform wall thickness during deep drawing of aluminum alloy and steel sheets. Texture is of special importance in designing Ti and Zr alloys with reasonable drawability and makes a distinction between an acceptable and a completely unstable material.85 The development of such structures is most conveniently characterized by the measurement of the plastic strain ratio or plastic anisotropy factor r, which is defined as the ratio of the true width (or lateral) strain ew to the true thickness strain et (i.e., r = ew /et) in a tensile test on the sheet-metal specimen. For isotropic sheet material tested in tension, r = 1; and for plastic anisotropy, r > 1. The r is a measure of capacity of a sheet to resist thinning. The higher the r value, the greater the resistance to thinning during deep drawing.32 This behavior, due to plastic anisotropy in the rolling plane, is undesirable because it leads to frequent interruptions of production runs and causes material wastage. It is, therefore, necessary to control earing to tolerable minimum by introducing an appropriate production schedule.86 In contrast, plastic anisotropy producing a high deformation resistance in thickness direction and less in the rolling plane is said to have a larger limiting draw ratio (LDR) and can be used advantageously in various aspects of formability.87 Thus, LDR is defined as the largest ratio of blank-to-cup diameters that may be drawn successfully. The LDR has been reported to vary linearly with the rm value, as shown in Fig. 4.25 for mild steel.88,89 In sheet metal forming technology, the drawability of the deep-drawn containers (e.g., oil filters and compressor housings)32 can be predicted from the average normal anisotropy rm (or r¯ ) in the sheet, that is given by rm (or r ) =

r0 + 2r45 + r90 4

(4.52)

FIGURE 4.24

Typical earing.

PLASTIC DEFORMATION

4.45

FIGURE 4.25 The deep drawability of mild steel, represented by the limiting draw ratio (LDR) as a function of the rm (or ¯r ) value for the material. (After Atkinson and McLean, 1965.)

The planar anisotropy which is responsible for the magnitude and position of ears in a deep-drawn cylindrical cup and to undesirable metal flow in the blank holder, in the general case, is usually defined by Dr =

r0 + r90 - 2r45 2

(4.53)

where r0, r45, and r90 are the experimentally measured plastic strain ratios in the rolling 0°, 45°, and transverse 90° directions in the sheet, respectively. When Dr is positive, the earing occurs in the 0° and 90° direction; when Dr is negative, the earing (i.e., uneven edges on cylindrical cups drawn from circular blanks)32 occurs in the 45° direction. When Dr = 0°, the earing disappears. Thus, planar anisotropy Dr is bad for earing, particularly with beverage cans drawn from Al alloys because (1) it increases the machining or other processing to be done on the drawn product and (2) it means an azimuthal variation in the thickness distribution of the product, which necessitates a thicker gage than that would be necessary for isotropic strains. Orientation distribution functions (ODFs) can be used to predict earing behavior; however, they can also be determined by a deep-drawing test. An rm value of 1 denotes complete isotropy or equal flow strength, in the plane of the sheet and in the thickness direction. A large rm value improves LDR, which indicates the high average depth (or wall height) and a high through-thickness strength relative to the strength in the plane of the sheet. A high rm value also improves the dent resistance property of formed panels.90–92 Rimmed and semikilled steels have rm values close to unity (1.0 to 1.3) due to their weak texture and are, therefore, not suited for the most demanding deep-drawing applications. In reality, a low rm value implies limited drawability. On the other hand, aluminum-killed steels with strong {111} textures have rm values in the range of 1.5 to 1.8. The Tior Nb-stabilized steels with strong {111} and {554} orientations have rm values approaching or exceeding 2.0. There is a direct relationship between rm value and the grain size of steel, as shown in Fig. 4.26. It is important to point out that the press-forming processes used in the industry to form automobile bodies, domestic consumer durables, beverage cans, and so forth can be considered as specific forms of deep drawing; therefore, a close control of r values affects a large sector of manufacturing industry. A combination of a high rm value and low Dr value gives optimum drawability. Table 4.7 lists the main texture components observed in the cold rolling and annealing textures of low-carbon and extra-low-carbon (ELC) steels, along with the calculated average rm and Dr values, relating to each texture component.93,94

3.0

Average Strain Ratio (rm)

2.6

Ti > 12 (C + N)

2.2

Aluminum - Killed

1.8 Rimmed Ti > 6 (C + N) 1.4

1.0 4

6

8

10

12

Grain Size (ASTM No.) FIGURE 4.26 Variation of rm with the grain size of steel.7 The normal range of conventional rimmed and killed steel is shown. In addition to chemical composition, the value rm is determined by coil processing. Two Ti-treated Al-killed steels are also shown. To produce an interstitial-free steel with a high rm value, the amount of Ti present must exceed 6 times the amount of C + N.

TABLE 4.7 Major Components in ColdRolling and Annealing Textures of LowCarbon Steels Texture Component

rm

Dr

{001} ·110Ò {112} ·110Ò {111} ·110Ò {111} ·112Ò {554} ·225Ò {110} ·001Ò

0.4 2.1 2.6 2.6 2.6 5.1

-0.8 -2.7 0 0 1.1 8.9

Source: After D. Daniel and J. J. Jonas, Met. Trans., vol. 21A, 1990, p. 331.

4.46

PLASTIC DEFORMATION

4.47

Preferred orientations are determined by X-ray methods. The X-ray pattern of a fine-grained randomly oriented material will show uniformly intense rings for all angles, whereas that of the preferred oriented grains will show breaking up of rings into short arcs or spots. The textures are usually represented on a stereographic projection by means of pole figures. Actually, they are two-dimensional plots of threedimensional distributions of crystal orientations and do not, therefore, describe the true texture. In this two-dimensional plot, the angular relationships between planes in the crystal aggregate are maintained. There are two (e.g., normal and inverse) pole figure methods that provide a description of the texture of a sheet material.95 In the former, X-ray data are plotted directly on the stereographic projection where the center of projection corresponds to the normal to the sheet plane in the rolling or in the wire axis. The latter represents the density distribution of an important direction in the polycrystalline specimen (such as wire axis or the sheet rolling direction) drawn on a stereographic projection of the lattice in standard orientation.95 This usually illustrates the intensities of a certain set of planes such as {100}, {110}, or {111}. The plane of projection is the plane of sheet where the rolling direction (RD) and the transverse direction (TD) lie. Recently, mathematical methods have been developed for the calculation of the three-dimensional ODF from the numerical data obtained from several pole figures,93 but they do not contribute much to the understanding of the theory of texture.96 Precht et al. have developed a simple method for the prediction and control of texture and associated earing in rolled 3004 aluminum which is based on the measurement of the X-ray diffraction peak height ratios of the {110} planes.97 Deformation texture of a metal depends on the stacking fault energy, the extent, the type, the operative mechanisms, and the temperature of deformation processes. As the degree of deformation is increased, the texture improves toward perfection. In polycrystalline materials, deformation texture can be broadly grouped into fiber texture and rolling texture.

4.8.1 Classification of Texture For practical purposes, polycrystalline texture can be grouped into two types, according to the symmetry of the preferred orientations in the specimen. When the crystals are aligned along one crystallographic direction , parallel to the major axis of the specimen, and around this common axis the crystals are in random orientation, the texture is termed fiber texture. It is denoted by , the fiber axis. In a rolled sheet or strip, the texture exhibits symmetry with the three orthogonal directions of the product such as the rolling direction (RD), the transverse direction (TD), and the normal direction (ND). Thus the rolling texture of such samples needs two parameters, {hkl}, for its description.98 4.8.1.1 Fiber Texture. This type of texture occurs by uniaxial deformation methods (e.g., wire drawing, extrusion, etc.) in which the grains are elongated in the forming direction. This texture is characterized by a crystallographic direction of low indices parallel to the wire axis, for example, fiber texture of wiredrawn bcc metals in the direction parallel to the wire axis.99 However, colddrawn wires of fcc metals may have a double-fiber texture such as + directions, in varying degree, parallel to the wire axis. The becomes predominant in metals with high SFE (which causes easy cross-slip), while those with low SFE (by the alloying additions) have a predominantly texture containing

CHAPTER FOUR

4.48

some direction. In hcp metals (e.g., Mg), basal planes with direction rotate and coincide with the wire axis. 4.8.1.2 Rolling Texture. A rolling texture {hkl} describes that the crystallographic planes {hkl} are parallel to the rolling plane, and that the particular crystallographic direction in that rolling plane is parallel to the rolling direction.98 In fcc metals and alloys, two rolling textures predominate. Generally, {110} texture (a-brass-type texture) is developed in the early stage of deformation, whereas [112] texture (copper-type texture) becomes effective with RD

2 1

3

4

7 8 9 10 11 12 1 2

3 2 1

2

5 6

4

5

6 7 8 9 10 1

1 2

2 2

1

FIGURE 4.27 A {110} pole figure of rolling texture of iron (90% reduction in one direction only after quenching from 925°C). A pole density of unit represents randomness.21 (Reprinted with permission from Trans. AIME, vol. 221, 1961, p. 764, a publication of The Metallurgical Society, Warrendale, Pennsylvania.)

RD 12 8 4 2 1 0.5

0.5

0.5 1 2 4

0.5 1 2 4 8 12

0.5

FIGURE 4.28 A {100} pole figure of the rolling texture of carbonyl iron (99.7% deformed).96 (Reprinted by permission of Dr. Riederer Verlag, GmbH, Stuttgart; after Liesner and Wahl.)

PLASTIC DEFORMATION

4.49

heavy plastic deformation in which extensive cross-slip occurs. Type of texture is also dependent on the SFE, deformation temperature, and grain size; for example, high SFE and high temperature of deformation produce the copper-type texture {112} , and lower-SFE material gives rise to a-brass-type texture. In bcc metals the predominant texture is {001} (cube-on-edge) with cube planes in the rolling plane; but other texture components such as {112} and {111} are also found (Table 4.7). Figure 4.2720 shows the {110} pole figure of a 90% cold-rolled iron sheet, whereas Fig. 4.2896 shows the {100} pole figure of the 99.7% (deformed) rolling texture of a carbonyl iron. HCP metals tend to have the basal plane oriented parallel to both the close-packed and the rolling directions. Note that the texture of the tubes is a function of the relative reductions of the wall thicknesses and the tube diameter in their fabrication. When the wall thickness and the tube diameter are reduced proportionally, the texture resembles that of a wire or rod. If only the wall thickness is reduced, the texture is similar to that of a rolled sheet.98

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 13a. 14. 15. 16. 17. 18.

Metals Handbook, 9th ed., vol. 8, American Society for Metals, Metals Park, Ohio, 1985. Making, Shaping and Treating of Steels, 9th ed., ed. H. E. McGannon, United States Steel Corp., Pittsburgh, Pa., 1971. Carbon and Alloy Steels, ASM International, Materials Park, Ohio, 1996; B. Taylor, Metals Handbook, vol. 14, 9th ed., ASM, Metals Park, Ohio, 1988, pp. 877–899. D. J. Mack, Trans. AIME, vol. 166, 1946, pp. 68–85. C. S. Barrett, W. D. Nix, and A. S. Tetelman, The Principles of Engineering Materials, Prentice-Hall, Englewood Cliffs, N.J., 1973. J. Gurland and R. J. Asaro, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 1258–1260. Sheet Metal Formability, American Iron and Steel Institute, Washington, D.C., 1984. G. E. Dieter, Mechanical Metallurgy, 3d ed., McGraw-Hill, New York, 1986. S. Kalpakjian, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4378–4381. A. Hamano, Met. Trans., vol. 24A, 1993, pp. 127–139. A. Blake, Practical Fracture Mechanics in Design, Marcel Dekker, New York, 1996. F. A. A. Crane and J. A. Charles, Selection and Use of Engineering Materials, Butterworths, London, 1987. B. Derby, in The Encyclopedia of Advanced Materials, Pergamon Press, Oxford, 1994, pp. 295–299. A. W. Thompson and J. F. Knott, Met. Trans., vol. 24A, 1993, pp. 523–534. M. V. Klassen-Neklyudova, Mechanical Twinning of Crystals, Consultant Bureau, New York, 1964. A. S. Argon, in Physical Metallurgy, 4th ed., eds. R. W. Cahn and P. Haasen, Elsevier Science B.V., Amsterdam, 1996, pp. 1877–1955. N. Brown and R. A. Ekvall, Acta Metall., vol. 10, 1962, pp. 1101–1107. W. D. Brentnahl and W. Rostoker, Acta Metall., vol. 13, 1965, pp. 187–198. A. Abel and H. Munir, Acta Metall., vol. 21, 1973, pp. 99–105.

4.50 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50.

CHAPTER FOUR

L. P. Kubin, Hardening of Metals, ed. P. Feltham, Freund, Israel, 1980, pp. 67–112. H. Jones, private communication, 1998. W. C. Leslie, Physical Metallurgy of Steels, McGraw-Hill, New York, 1981. M. Bernstein, Acta Metall., vol. 17, 1969, pp. 249–259. H. Conrad, in Iron and Its Dilute Solid Solutions, eds. C. W. Spencer and F. E. Werner, Interscience, New York, 1963, p. 315. W. G. Johnston and J. J. Gilman, J. Appl. Phys., vol. 30, 1959, p. 129. W. C. Leslie, J. T. Michalak, and F. W. Aul, in Iron and Its Dilute Solid Solutions, eds. C. W. Spencer and F. E. Werner, Interscience, New York, 1963, p. 119–212. R. W. K. Honeycombe, The Plastic Deformation of Metals, 2d ed., Edward Arnold, London, 1984. G. E. Dieter, Metals Handbook, vol. 14, 9th ed., ASM International, Metals Park, Ohio, 1988, pp. 20–27. B. W. Christ and M. L. Picklesimer, Acta Metall., vol. 22, 1974, pp. 435–447. H. Conrad, Acta Metall., vol. 11, 1963, pp. 75–77. M. Abe, in Constitution and Properties of Steel, vol. 7, vol. ed. F. B. Pickering, VCH, Weinheim, 1992, pp. 285–333. Handbook of Metal Forming, ed. K. Lange, McGraw-Hill, New York, 1985. B. L. Bramfitt, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4897–4899. C-Y. Sa, in Sheet Stamping Technology: Applications and Impact, SP-779, Society of Automotive Engineers, Warrendale, Pa., 1989, pp. 37–45. A. K. Ghosh, S. S. Hecker, and S. P. Keeler, in Workability Testing Techniques, ed. G. E. Dieter, American Society for Metals, Metals Park, Ohio, 1984, pp. 135–195. G. I. Taylor, Proc. R. Soc., London, vol. 89, 1934, p. 660. E. Orowan, Dislocations in Metals, American Institute of Mining, Metallurgical, and Petroleum Engineers, New York, 1954, p. 69. S. J. Basinski and Z. S. Basinski, in Dislocations in Solids, vol. 4, North-Holland Physics Publishing, Amsterdam, 1979, p. 263. V. I. Startsev, in Dislocations in Solids, vol. 6, ed. F. R. N. Nabarro, North-Holland Physics Publishing, Amsterdam, 1979, p. 145. D. J. Dingley and D. McLean, Acta Metall., vol. 15, 1967, pp. 885–901. F. B. Pickering, Physical Metallurgy and Design of Steels, Applied Science Publishers, London, 1978. D. Kuhlmann-Wilsdorf, Met. Trans., vol. 16A, 1985, pp. 2091–2108. H. J. McQueen, Met. Trans., vol. 8A, 1977, pp. 807–824. J. R. Low, in Iron and Its Dilute Solid Solutions, eds. C. W. Spencer and F. E. Werner, Interscience, New York, 1963, pp. 217–263. A. Cracknell and N. J. Petch, Acta Metall., vol. 3, 1955, pp. 186–189. E. O. Hall, Yield Point Phenomena in Metals and Alloys, Macmillan, London, 1970. N. Hansen, Met. Trans., vol. 16A, 1985, pp. 2167–2189. A. W. Thompson, M. Baskes, and W. F. Flangan, Acta Metall., vol. 21, 1973, pp. 1017–1028. R. W. Armstrong, in Yield, Flow and Fracture of Polycrystals, ed. T. N. Baker, Applied Science Publishers, London, 1983, pp. 1–31. R. A. Masumura, P. M. Hazzledine, and C. S. Pande, Acta Mater., vol. 46, no. 13, 1998, pp. 4527–4534. Ref. 19 in ref. 49.

PLASTIC DEFORMATION

4.51

51. A. H. Choksi, A. Rosen, J. Rarch, and H. Gleiter, Scripta Metall., vol. 23, 1989, p. 1679. 52. J. Meakin and N. J. Petch, Role of Substructure in Mechanical Behavior of Metals, Report ASD-TDR-63-324, U.S. Air Force Wright-Patterson AFB, Ohio, 1963, pp. 243– 251. 53. M. F. Ashby, Phil. Mag., vol. 21, 1970, pp. 399–424. 54. M. F. Ashby, Strengthening Methods in Crystal, eds. A. Kelly and R. B. Nicholson, Wiley, New York, 1971, pp. 137–192. 55. E. T. Wessell, Trans. AIME, vol. 209, 1957, p. 930. 56. J. S. Blackmore and E. O. Hall, JISI, vol. 204, 1966, p. 817. 57. M. V. Bever, D. L. Holt, and A. L. Titchener, The Stored Energy of Cold Work, in Progress in Materials Science, vol. 17, Pergamon Press, Elmsford, N.Y., 1973. 58. D. V. Wilson, in Encyclopedia of Materials Science and Engineering, suppl. vol. 1, Pergamon Press, Oxford, 1990, pp. 508–512. 59. W. C. Leslie, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4007–4011. 60. Carbon Steel, Wire and Rod, Iron and Steel Institute, Warrendale, Pa., 1993. 60a. V. T. L. Buono, M. S. Andrade, and B. M. Gonzalez, Met. Trans., vol. 29A, 1998, pp. 1415–1423. 60b. R. H. U. Queiroz, E. J. Foneseca, V. T. L. Buono, M. S. Andrade, E. M. PeSilva, and B. M. Gonzalez, Wire J. Int., June 1999, pp. 76–81. 61. W. C. Leslie and E. Hornbogen, in Physical Metallurgy, 4th ed., eds. R. W. Cahn and P. Haasen, Elsevier Science B.V., Amsterdam, 1996, pp. 1555–1620. 62. V. T. L. Buono, G. M. Gonzalez, and M. S. Andrade, Scripta Metall., vol. 38, no. 2, 1998, pp. 185–190. 63. Carbon and Alloy Steels, ASM Specialty Handbook, ASM International, Materials Park, Ohio, 1996, pp. 308–328. 64. P. Rodriguez, in Encyclopedia of Materials Science and Engineering, suppl. vol. 1, ed. R. W. Cahn, Pergamon Press, Oxford, 1990, pp. 504–508. 64a. A. Fjeldly and H. J. Roven, Met. and Mats. Trans., vol. 31A, 2000, pp. 669–678. 65. J. M. Robinson and M. P. Shaw, Int. Mater. Rev., vol. 39, no. 3, 1994, pp. 113–122. 66. N. J. Petch, in Advances in Physical Metallurgy, eds. J. A. Charles and G. C. Smith, The Institute of Metals, London, 1990, pp. 11–25. 67. Q. H. Tang and T. C. Wang, Acta Mater., vol. 46, no. 15, 1998, pp. 5313–5321. 68. J. W. Christian and S. Mahajan, Progr. Mat. Sci., vol. 39, 1995, pp. 1–157. 69. G. T. Gray, III, in Encyclopedia of Materials Science and Engineering, suppl. vol. 2, ed. R. W. Cahn, Pergamon Press, Oxford, 1990, pp. 859–866. 70. L. Remy, Met. Trans., vol. 12A, 1981, p. 847. 71. E. O. Hall, Twinning, Butterworths, London, 1954. 72. P. B. Partridge, Int. Metall. Rev., vol. 12, 1967, p. 169. 73. F. J. Turner, in Deformation Twinning, eds. R. E. Reed-Hill, J. P. Hirth, and H. C. Rogers, Gordon and Breach, New York, 1964. 74. T. L. Altshuler and J. W. Christian, Acta Metall., vol. 14, 1968, pp. 903–908. 75. J. D. Verhoeven, Fundamentals of Physical Metallurgy, Wiley, New York, 1975. 76. R. E. Reed-Hill and R. Abbaschian, Physical Metallurgy Principles, 3d ed., PWS-Kent Publishing Co., Boston, 1992. 77. R. W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3d ed., Wiley, New York, 1989. 78. D. Hull, Introduction to Dislocations, Pergamon Press, Oxford, 1975.

4.52

CHAPTER FOUR

79. A. Crocker, in Twinning in Advanced Materials, eds. M. R. Yoo and M. Wuttig, The Metallurgical Society, Warrendale, Pa., 1994, pp. 3–16. 80. P. Haasen, Physical Metallurgy, 2d ed., Cambridge University Press, New York, 1986. 81. M. J. Kelley and N. S. Stoloff, Met. Trans., vol. 7A, 1976, pp. 331–333. 82. D. Juul Jensen and N. Hansen, Acta Metall. Mater., vol. 38, 1990, p. 1369. 83. E. Aernoudt, P. V. Houtte, and T. Leffers, in Plastic Deformation and Fracture of Materials, vol. 6, ed. M. Mughrabi, VCH, Weinheim, 1993, pp. 89–136. 84. H. J. Prask and C. S. Choi, Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4895– 4897. 85. C. Tome, A. Pochettino, and R. Penelle, in ICOTOM—Eighth International Conf.: Texture of Materials, The Metallurgical Society, Warrendale, Pa., 1988, pp. 985–990. 86. P. M. B. Rodrigues and P. S. Bate, in Textures in Non-ferrous Metals and Alloys, eds. H. D. Merchant and J. G. Morris, The Metallurgical Society-American Institute of Mining, Metallurgical, and Petroleum Engineers, Warrendale, Pa., 1985, pp. 173–187. 87. R. Sowerby, in Textures in Non-ferrous Metals and Alloys, eds. H. D. Merchant and J. G. Morris, The Metallurgical Society-American Institute of Mining, Metallurgical, and Petroleum Engineers, Warrendale, Pa., 1985, pp. 99–115. 88. S. P. Keeler, in Automotive Sheet Metal Formability, Final Report on AISI Project no. 1201-409, American Iron and Steel Institute, Washington, D.C., 1989. 89. R. W. Cahn, in Processing of Metals and Alloys, vol. 15, ed. R. W. Cahn, VCH, Weinheim, 1991, pp. 429–480. 90. W. B. Hutchinson, Int. Metall. Rev., vol. 29, no. 1, 1984, pp. 25–42. 91. K. Yoshida et al., Proceedings of the 8th Biennial IDDRG Congress, Gothenburg, 1974, International Deep Drawing Research Group, pp. 258–268. 92. A. J. Klein and E. W. Hitchler, Metall. Eng. Q., vol. 13, 1973, pp. 25–27. 93. R. K. Ray, J. J. Jonas, and R. E. Hook, Int. Mats. Rev., vol. 39, no. 4, 1994, pp. 129–172. 94. D. Daniel and J. J. Jonas, Met. Trans., vol. 21A, 1990, p. 331. 95. C. S. Barrett and T. B. Massalski, Structure of Metals, 3d ed., McGraw-Hill, New York, 1966. 96. J. Grewen and J. Huber, in Recrystallization of Metallic Materials, ed. F. Haessner, Dr. Riederer Verlag GmbH, Stuttgart, 1978, pp. 111–136. 97. W. Precht, L. Christodoulou, and F. Lockwood, in Textures in Non-ferrous Metals and Alloys, eds. H. D. Merchant and J. G. Morris, The Metallurgical Society-American Institute of Mining, Metallurgical, and Petroleum Engineers, Warrendale, Pa., 1985, pp. 223–227. 98. H. Hu, The Encyclopedia of Advanced Materials, vol. 4, Pergamon Press, Oxford, 1994, pp. 2798–2807. 99. D. V. Wilson, J. Inst. Metall., vol. 94, 1966, pp. 84–93.

CHAPTER 5

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.1 INTRODUCTION Annealing behavior of cold-worked materials has been of considerable interest for many years from both a theoretical and a practical standpoint. This is a widely accepted industrial practice which is broadly divided, based on the effects of temperature, into three main stages of the process: recovery, recrystallization, and grain growth. Usually the term recrystallization is synonymous with primary recrystallization. Grain growth can be subdivided into normal or continuous grain growth and abnormal or discontinuous grain growth, also called secondary recrystallization. These classifications are approximate because overlapping of some stages is a common phenomenon. This chapter deals with the property changes, mechanisms, kinetics, and commercial importance of these procedures.

5.2 RELEASE OF STORED ENERGY When the temperature of a cold-worked metal is raised sufficiently, the stored energy of cold work is released as heat1 and is usually measured in a highly sensitive differential calorimeter as the difference in specific heat (usually called power difference DP) between the cold-worked and the undeformed (or annealed) specimens. Such measurements are carried out by anisothermal, isothermal, or isochronal annealing methods. These methods determine the total amount of stored energy as well as the kinetics of release of energy. A knowledge of kinetics of the energy release is vital in understanding recovery and recrystallization processes and in interpreting the mechanism of stored energy. In the anisothermal annealing method, the temperatures of cold-worked and annealed specimens are raised at equal and constant rates, and the power difference DP between the two specimens is determined as function of temperature (Figs. 5.1 and 5.2). The area below the curve is proportional to the amount of stored energy released. Generally the power-difference curves exhibit single or multiple peaks (Fig. 5.1), each representing a surge in the stored energy released. The peak occurring at the higher and lower temperatures denotes the recrystallization and 5.1

5.2

CHAPTER FIVE

FIGURE 5.1 Schematic representation of the rate of energy release as a function of annealing time.

FIGURE 5.2 Schematic plot of the rate of energy release as a function of annealing temperature for three different purities.

recovery processes, respectively. However, there are some cases where recovery occurs without any surge in the energy release. Figure 5.2 shows the schematic illustration of three different patterns of energy release curves as a function of temperature as obtained by the anisothermal annealing method. Curve A, which is for the highest-purity metal, represents (1) a very small amount of energy released for recovery and (2) the lowest recrystallization temperature, as measured by a main (i.e., large) power peak. A slightly impure metal is represented by curve B, where a small but appreciable amount of energy is released during recovery. This is obviously clear from the existence of a shoulder prior to the main peak. Metal with increased impurity is represented by curve C, in which it appears that nearly one-half of the stored energy is released preceding the main peak. These patterns of energy release curves have been experimentally observed for copper and silver with varying purity. Thus we can conclude here that

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.3

FIGURE 5.3 The rate of energy release as a function of annealing time for 99.999% copper deformed by extension of two strains. In each case the isothermal annealing temperature was 189.7°C.1,2 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

increasing the impurity content in the metal raises not only the recrystallization temperature but also the fraction of the stored energy Erecovery/Etotal released during recovery. As in the anisothermal annealing method, in the isothermal annealing method the power difference DP between the two specimens is plotted against the (isothermal) annealing time (Fig. 5.3).2 The area under the curve is proportional to the amount of energy released. In the isochronal annealing method, the power difference is plotted against different annealing temperatures for equal times. This method provides the data for the amount of stored energy released as a function of temperature and does not render any information concerning the kinetics of the released energy during individual annealing treatment. Hence, the resolution obtained by this method is poor. Figure 5.4a shows the schematic representation of a change in mechanical properties (e.g., hardness, strength, ductility, etc.) and other physical properties (e.g., electrical resistivity, density, cell size, etc.) occurring during the annealing process of plastically deformed metal at room temperature.

5.3 RECOVERY Recovery embraces all structural changes that do not involve the sweeping of the deformed structure by migrating high-angle boundaries.3 Alternatively, recovery is defined as the restoration of physical properties of the cold-worked metal without any observable change in microstructure. The driving force for this process is a decrease in the thermodynamic (or stored) energy differences between the deformed state and the recovered structure. It involves the annihilation of dislocations and the rearrangement of dislocations into lower-energy configurations.4 Recovery thus produces changes in structures and properties prior to the onset of new strain-free recrystallized grains. More strain-free regions that form during

FIGURE 5.4 (a) Changes in several physical and mechanical properties of plastically deformed metal as a function of temperature. (b) Fractional recovery of defects introduced by electron or neutron bombardment of liquid helium temperature.9 [(b) Reprinted by permission of Pergamon Press, Plc.]

5.4

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.5

recovery can act as the origin or nucleus for subsequent recrystallization annealing. Recovery is often measured directly by calorimetry from the changes in stored energy which are achieved by glide, climb, and cross-slip of dislocations because stored energy is directly related to the number and configuration of dislocations in the material. The calorimetric measurements can be made with no phase transformation over the temperature range of experiment. Recovery is measured indirectly by following the changes in mechanical or physical property.4 The recovered substructure is usually weaker—but more ductile, more stable for high-temperature service, and more corrosion-resistant—than that formed by cold working.5 A special type of recovery commonly encountered in heat treatment is called stress relief annealing, which is often desirable and is used to retain much of the increased strength resulting from cold work and to remove the residual stresses produced by cold-working operations.

5.3.1 Mechanical Properties The extent of recovery depends on (1) the kinetics of recovery itself, (2) the temperature or time at which recrystallization occurs, and (3) the nature of metallic materials. In general, complete recovery can occur, particularly in lightly deformed materials. The larger the deformation, the smaller the fraction of the property (e.g., hardness) change on deformation which can be restored. This is presumably due to the greater ease of recrystallization at increased deformation6 (see also Fig. 5.10). However, there is an exception: Large deformation in single crystals of hexagonal metals that is produced by easy glide can be completely recovered to the original structure and properties on annealing.6 Based on the changes in mechanical properties upon recovery, metallic materials are subdivided into two groups.7 Metals with low stacking fault energy (SFE), denoted by gSFE (e.g., Cu, Ni, a-brass, and austenitic stainless steel), usually show very little recovery in mechanical properties (e.g., hardness and strength) prior to recrystallization. This is presumably due to the occurrence of little climb and consequently little recovery or rearrangement of dislocation structure. Since hardness and strength (or flow stress) are controlled by concentration and disposition of dislocations, they are not affected at the recovery temperature. However, metals with high SFE (e.g., a-iron, Al, etc.) usually show an appreciable amount of softening during a recovery anneal. This is attributed to the fact that the dislocation rearrangement can easily occur by the rapid climb process within the deformed metal during recovery anneal.7,8 Solutes may influence recovery by their effect on the SFE, by pinning dislocations, or by altering the concentration and mobility of vacancies.4 However, a large drop in hardness and strength occurs during recrystallization because of the presence of strain-free new recrystallized grains.

5.3.2 Physical Properties During plastic deformation, concentration of lattice vacancies and dislocation density increase. As a result, the electrical resistivity is increased and density is decreased. During annealing, the resistivity continues to decrease and finally approach toward the original (annealed) value, together with the change in the stored energy released, increase in density, and considerable reduction of lattice strains. In contrast to normal deformation, deformation caused by high-energy electron or fast-neutron irradiation in nuclear reactors involves a complex mechanism;

5.6

CHAPTER FIVE

when such a damaged material is annealed, the resistivity decreases in a distinct manner, different from that due to cold work (Fig. 5.4b).3,8,9 The study of damage recovery provides the general understanding of lattice imperfections in metals and the physics of defect reactions occurring at higher-temperature irradiations. This knowledge is of technological significance to both fusion and fission reactor systems.9 The drop in resistivity, increase in density, and reduction in lattice strain during recovery anneal give an indication of a significant decrease of concentration of point defects. Thus the properties that are most influenced by recovery are those that are sensitive to point defects.

5.3.3 Microstructure (Cell Size) During recovery anneal, the cell walls or boundaries sharpen, and then the cells gradually grow to a large size while the interior dislocations are further attracted into them. However, the sharpness of such cell walls occurring during annealing is influenced by the sharpness of the original cell walls formed during plastic deformation, which, in turn, depends on its SFE, that is, the extent of dissociation of dislocations and their ability to climb. Thus higher and lower SFE of the metal gives rise to well-defined and ill-defined cell structures, respectively, after moderate or heavy plastic deformation.8

5.3.4 Mechanism Recovery annealing is a thermally activated process, which is reflected by the fact that its operative mechanism at low-temperature, intermediate-temperature, and high-temperature range involves vacancy motion, dislocation motion without climb, and dislocation motion with climb, respectively.

5.3.5 Rearrangement of Dislocations into Stable Arrays 5.3.5.1 Polygonization, Subgrain Formation, and Subgrain Growth. An especially simple form of recovery process occurring at higher (recovery) temperature is called polygonization. It is associated with single crystals that have been slightly (plastically) bent (about an axis parallel to their active slip planes) so that only one glide system operates and is subsequently annealed. After annealing, the curved crystal breaks up into small (single or perfect) crystal segments (called subgrains), each retaining the local orientation of the original bent crystal and separated by dislocation walls known as subboundaries (or tilt boundaries) that are right angles to the glide vector of the active slip plane.8,10 These represent a simple type of lowangle grain boundary. When an X-ray beam is diffracted by such bent crystals, Laue photographs with elongated or asterated spots are observed. When the X-ray beam is allowed to fall on the annealed specimen, elongated spots of the bent crystal are replaced by a series of tiny, sharp spots, which is a characteristic of the polygonized structure (Fig. 5.5).11 The sub-boundaries (or polygon walls) present in the polygonized structures can be revealed by etching; these sub-boundaries appear as dense rows of pits in the etched photograph (Fig. 5.6).12 To understand this, let us consider the polygoniza-

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

(a)

5.7

(b)

FIGURE 5.5 Schematic Laue pattern showing (a) elongated spot obtained from a bent single crystal and (b) breaking up of elongated spots into a series of small sharp spots after polygonization.11 (Reprinted by permission of PWS-Kent Publishing Co., Boston, Massachusetts.)

FIGURE 5.6 Complete polygonized structure in bent and annealed iron (single crystal) at 700°C is revealed by dense rows of etch pits.12 750X. (Reprinted by permission of Pergamon Press, Plc.)

tion process in terms of dislocation distribution. When a crystal is subjected to plane gliding, edge dislocations of both signs will emerge at the surface as a result of easy glide. However, in bend gliding, which is a prerequisite for polygonization, a large number of excess edge dislocations of one sign are generated in the crystal to accommodate the plastic curvature, especially near the outer surface6 (Fig. 5.7a).13 The density of these excess dislocations is given by 1/rb, where r is the average radius of curvature and b is the Burger’s vector. Annealing then causes these dislocations to rearrange themselves by lining up over one another to form the dislocation walls or tilt boundaries, normal to the Burgers vector, by the process as shown in Fig. 5.7b.13 It is thus clear that alignment of edge dislocations into the low-angle tilt boundaries in the polygonized structure occurs by the dislocation glide and climb, which leads to a large relief of each other’s elastic strain field (i.e., lowering of strain

5.8

CHAPTER FIVE

FIGURE 5.7 Schematic representation of the polygonization process: (a) Random arrangement of excess (positive) edge dislocations which remain on active slip planes after bend-gliding. (b) Regrouping of edge dislocations to form dislocation walls.13 (Reprinted by permission of Butterworths, London.)

energy). The driving force for this process is, therefore, the decrease of strain energy achieved in forming a low-angle tilt boundary as a result of polygonization. In general, the relationship between the pure tilt (or grain) boundary energy Eb and angle of misorientation across the boundary (or relative tilt angle of the two subgrains) q [艐 b (Burgers vector)/h (spacing of dislocations)] (Fig. 5.8a) is expressed by the Read-Shockley equation Eb = E0q ( A - ln q )

(5.1)

where E0 = Gb/4p(1 - n), G is the shear modulus, n is the Poisson ratio, A = 1 + ln(b/2pr0), and r0 is the radius (or size) of dislocation core (often between b and 5b). According to Eq. (5.1), the energy of a tilt boundary and energy per dislocation increases and decreases, respectively, with increasing misorientation (decreasing h), as shown in Fig. 5.9.4,14 This theory holds good for small values of q (< ~15°). Equation (5.1) can be expressed in normalized forms of Eb and q, i.e., in terms of Em and qm, when the boundary becomes a high-angle boundary (that is, q ~ 15°).4 E=

Emq Ê q 1 - ln ˆ Ë qm qm ¯

(5.2)

Initially, segments of tilt boundaries are produced (from the scattered dislocations) which are on the order of 10 dislocations high. Such boundaries then exert a small force on isolated dislocations some distance away. This causes these dislocations to glide into the neighborhood of the boundary and then to climb sufficiently to find a niche in the boundary. In the latter stage, polygonization progresses by the merger or coalescence of pairs of adjacent subboundaries or short-range boundaries, thereby providing a somewhat large boundary which is still of the lowangle (tilt) type. This is shown in Fig. 5.8b.8 This part of the process is associated with the migration of vacancies, to or from the climbing edge dislocations. The occurrence of the removal of vacancies from the lattice can be explained by the fact that a significant decrease in electrical resistivity has been observed prior to recrystallization. Work on iron has shown that polygonization occurs with great ease after annealing of moderately deformed high-purity iron at temperatures as low as 300°C. However, this tendency decreases with increasing impurity, especially carbon content.15

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.9

FIGURE 5.8 Schematic illustration. (a) Structure of a tilt boundary. The two subgrains are mutually tilted by an angle q. (b) Edge dislocations at a tilt boundary triple point.8 (Reprinted by permission of North-Holland Physics Publishing, Amsterdam.)

5.3.5.2 Relationship between Subgrain Size and Flow Stress. The relationship between the subgrain structure formed by recovery after cold work and the flow stress of a material can be expressed in three ways by assuming negligible dislocation density within the subgrains. 1. If we assume that subgrains behave as subgrains, the flow stress s will be given by a Hall-Petch equation: s = s 0 + k1 D -1 2

(5.3)

where s0, the friction stress, and k1 are constants and D is the subgrain diameter. 2. If we assume the operation of dislocation sources whose length will be closely related to their subgrain diameter, the recovered flow stress will be given by the Kuhlman-Wilsdorf15a equation

5.10

CHAPTER FIVE

FIGURE 5.9 The energy of a tilt boundary and energy per dislocation as a function of the crystal misorientation.4,14 (Reprinted by permission of Elsevier Science Publishing, B.V., Amsterdam.)

s = s0 + k2D-1

(5.4)

Equations (5.3) and (5.4) may both be expressed in the form s = s0 + k3D-m

(5.5)

3. If we assume the cell boundary in terms of dislocation tilt boundary, the area of subgrain per unit volume A becomes ~3/D; for small orientations (angles of tilt boundary), q = b/h and the length of dislocation per unit area of boundary L = 1/h. Hence r = AL =

3 3qb = Dh D

(5.6)

Assuming the relationship between flow stress and dislocation density of Eq. (4.36), we can write D r = s 0 + k4 Ê ˆ Ëq ¯

-1 2

(5.7)

which is similar to Eq. (5.5) with m = –12 and k3 = k4q 1/2. All these equations have been employed to analyze substructure strengthening; however, there is little agreement in their application. According to Thompson’s findings, the well-developed subgrain structures tend to exhibit m ~ 0.5, and cell structures exhibit m ~ 1.15b

5.3.6 Kinetics The recovery anneal has no incubation period. It is a homogeneous process because it occurs uniformly throughout the entire volume of the specimen.4 During isother-

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.11

mal annealing after cold work, all properties (e.g., resistivity, flow stress) that change during recovery have a characteristic kinetic form: The rate of recovery (of the relevant property) is initially very rapid, and then it steadily diminishes with increasing time and eventually ceases after a long time. In the most extensive study,16,17 the recovery of the initial flow stress of polycrystalline zone-refined iron was evaluated as a function of the amount and temperature of deformation and time and temperature of the recovery anneal. The fraction of residual strain hardening, which is a measure of recovery, was plotted against time at various temperatures, as shown in Fig. 5.10.18 The fraction of residual strain hardening was defined as 1- R = where

sr -s0 sm -s0

R (fraction or extent of recovery) =

(5.8) sm - s r sm - s0

(5.9)

sr is the flow (or yield) stress after recovery, sm is the flow stress after deformation, and s0 is the flow stress after full annealing. If we assume that the change of flow stress occurring during the recovery anneal is proportional to the concentration of defects, we can write the recovery process by the equation d(s m - s r ) = K (s m - s r )e -Q RT dt

(5.10)

where n, the order of the reaction, is an integer; K is the rate constant for the reaction; Q is the activation energy; R is the universal gas constant; and T is the absolute temperature. Rearranging Eq. (5.10) gives

FIGURE 5.10 Recovery kinetics represented by the change in residual strain hardening as a function of time during isothermal annealing at various temperatures for iron strained 5% in tension at 0°C: 1 - R denotes a fraction of flow stress increment which remains after annealing.18 (Reprinted by permission of John Wiley & Sons, New York.)

CHAPTER FIVE

5.12

d(s m - s r ) -Q RT dt n = Ú Ke m -sr )

Ú (s

(5.11)

which reduces to A = kte -Q RT

(5.12)

where A is an unknown function of (sm - sr) and t is the annealing time. Equation (5.12) can also be represented in logarithmic form as ln t = ln A +

Q RT

(5.13)

Equations (5.10) and (5.12) can be used to measure the time-dependent decay of any other physical property during recovery. We can determine the value of Q from the slope of the plot of ln t versus 1/T for a given value of (1 - R). The activation energy for recovery Q is usually the same as that for self-diffusion, which may be represented by the equation D = D0 e -Q RT

(5.14)

Q (self - diffusion ) = DH f + DH m

(5.15)

where

f

m

where DH and DH represent the enthalpy of vacancy formation and vacancy migration, respectively. The diffusion coefficient D can be measured from x = 2Dt

(5.16)

where x is the diffusion distance in centimeters, D is the self-diffusion coefficient in square centimeters per second, and t is the time in seconds. It has been shown by Leslie et al.18 and Michalak and Paxton16 that Q does not remain the same during recovery of the flow stress of iron but becomes closer to DHm at small recovery time and closer to Q (self-diffusion) at the latter stage (i.e., large recovery time). This implies that vacancy migration is the controlling factor at short recovery times and that self-diffusion (i.e., dislocation climb) is the controlling factor at longer times.

5.4 RECRYSTALLIZATION Recrystallization is the thermally activated microstructural evolution process in which a new set of comparatively strain-free grains nucleates at the expense of a deformed matrix until it is consumed.19 Alternatively, recrystallization is defined as the reorientation of crystals in a solid material by the passage of a high-angle boundary. The process is akin to phase transformation in the sense that it can be described phenomenologically in terms of the constituent nucleation and growth rates. Nucleation may be either time-dependent or site-saturated. The driving force for this process is the reduction in free energy, which is accomplished by the reduction of the dislocation network remaining after the recovery stage.20 Unlike recovery, recrystallization produces a drastic change of mechanical and other physical properties of the deformed metal to the level corresponding to those of the annealed

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.13

condition. For example, during recrystallization over a small temperature range, hardness, yield strength, and tensile strength are generally reduced (Fig. 5.4a). Similarly, the ductility increases rapidly. Physical properties such as electrical resistivity decrease, and for iron and steel, the magnetic permeability increases markedly to its original level. The rate of heating of the specimen to the annealing temperature plays a significant role. During rapid heating, recovery is less and the driving force for recrystallization is greater. Hence recrystallization occurs faster.20

5.4.1 Origin of Recrystallized Nuclei A transmission electron microscopic (TEM) study of recrystallization in cold-rolled silicon-iron and single crystals by Hu has endorsed the hypothesis that the strainfree recrystallized nuclei are formed by subgrain coalescence via the “evaporation” of edge dislocations constituting the subboundaries between them.21–23 The enlarged subgrains so formed, much larger than their neighbors, could act as nucleus. Figure 5.11 is a schematic model due to Jones et al.24 which illustrates the nucleation of recrystallized grain based on subgrain coalescence at an original high-angle grain boundary (along AB) in a polygonized structure. They established that subboundary dislocations link continuously with dislocations in a neighboring high-angle grain boundary; the disappearance of subboundary involves a perceptible rearrangement of the dislocations (by climb and glide) in the grain boundary, and in reality, it was observed that grain boundaries serve more effectively as dislocation sinks than as dislocation sources. Recently Doherty and Szupunar25 modified Li’s theoretical model of subgrain coalescence26 by assuming the climb mobility of dislocation loops via pipe diffusion instead of that of discrete edge dislocations by lattice diffusion to predict a much faster rate of coalescence. They also employed a nonuniform array of dislocations

FIGURE 5.11 Schematic model illustrating the nucleation of recrystallized grain based on subgrain coalescence at an original high-angle grain boundary (along AB) in a polygonized structure.22,24 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

5.14

CHAPTER FIVE

FIGURE 5.12 An optical micrograph of aluminum compressed 40% and annealed for 1 hr at 328°C, showing strain-induced boundary migration into both grains. The direction of growth is not always parallel to the plane sectioned, as can be seen from the two white “bulges” that do not contact the lower parent grain in the section seen.29 (Courtesy of S. P. Bellier and R. D. Doherty.)

to model the disappearance of a low-angle boundary adjacent to a high-angle boundary. The second nucleation model, which was first reported by Beck and Sperry27 and later studied by Bailey and Hirsch,28 is strain-induced boundary migration (SIBM). In this model a subgrain within a deformed grain grows into its adjacent grain, forming a bulge which has the same orientation as the original grain, but is predominantly free of dislocations. Figure 5.12 is an example of SIBM in a compressed and annealed aluminum. Some of the tongues originating from above or below the plane of section were shown to have the same orientation with a nearby region of the deformed structure, below the central band limits.29 The salient point of the micrograph is that it exhibits “two-way” SIBM, which cannot occur by one grain with a larger subgrain size than the other grain. Thus it suggests that a process such as subgrain coalescence has occurred at the grain boundary. In subsequent work with Paul Faivre, they showed that subgrain coalescence occurred at grain boundaries where a high-angle transition band met the grain boundary.29a Note that the difference in nucleation models between coalescence and SIBM lies only in the initial stage, whereas the later stages of nucleus growth are indistinguishable in the two models. Based on the studies on Al, Cu, Fe, and Cu, it has been inferred that a moderate amount of deformation promotes the occurrence of subgrain coalescence in the early stages, whereas high strain and low SFE promote SIBM.8 Based on the TEM study of rolled brass by Huber and Hatherly30 and further observation by Jones31 in Cu alloys and stainless steels, another nucleation mechanism was proposed which involved nucleation from recovery twins (formed at a very

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.15

early stage of subgrain growth) and growth by a SIBM mechanism.8 Similarly, the high-vacuum electron microscopic (HVEM) studies of tensile deformed and annealed single crystals by Wilbrandt and Haasen,32 and later those of Cu-Mn and Cu-P alloys with a very low SFE by Haasen,33 have prompted them to propose another multiple twinning mechanism (i.e., successive twin generation mechanism), as shown in Fig. 5.13. They emphasize that most of the recrystallized orientations do not exist in the deformed microstructure, which become modified by multiple twinning on specific twinning variants after nucleation. That is, orientations of the recrystallized grains are related to multiple twinning.32,33 This multiple twinning nucleation may be valid in materials that do form extensive annealing twins, e.g., brass and even copper, but it seems very unlikely in aluminum.33a Humphrey and Ferry (1996) have reported that there is much more twinning at a free surface than away from the surface. These results on the effect of a single surface in samples annealed after polishing for observation in HVEM clearly indicate the probability that the multiple twinning, generating new orientations, as reported by Haasen for the case of thin-foil TEM annealing with a free surface, is valued for a process occurring in thin-section annealing and not for materials annealed in bulk.33a The concept of preexisting or preformed nuclei which turn into a viable nucleus, as proposed by Hutchinson in 1992, is also gaining favor. It is thus clear that the orientation of new grains is clearly imprinted in the previous deformed microstructure. The nucleation sites for recrystallization in the deformed microstructure may be either those present prior to deformation, such as second-phase particles or grain boundaries, or those induced by the deformation, such as transition bands and shear bands.8 5.4.2 Kinetics of Recrystallization Like phase transformation and unlike recovery, the process of primary recrystallization proceeds by the nucleation and growth process in a sigmoidal manner with respect to time. Recrystallization nucleus may be defined as a crystallite of low internal energy growing into deformed material from which it is separated by a high-angle boundary. The basic characteristics of the kinetics of the primary recrystallization process are as follows: 1. An initial incubation period is necessary. 2. It is followed by slow rate of change at the initial stage, accelerated rate of change at the intermediate stage, and finally slow rate of change at the final stage. This is schematically shown in Fig. 5.14a. 5.4.2.1 The Johnson-Mehl-Avrami-Kolmogorov (JMAK) Model. The main features of a formal theory of recrystallization kinetics due to Johnson, Mehl, Avrami, and Kolmogorov, called the JMAK model, are discussed below.34,35 Let us consider that N˙ is the nucleation rate (i.e., number of new grains formed per unit time per unit volume of unrecrystallized metal) and G is the linear growth rate of any new grain. During linear growth rate G, the radius of a growing nucleus r is given by r = G(t - t0 )

(5.17)

where t is the time of recrystallization and t0 is the incubation period. This relationship is represented in Fig. 5.14b. If we consider the nucleus to be a sphere, then

Matrix R2 R10 R2 R1

Band

R2

R3

R10 –R9

Matrix

R5

Band

R7

R4

R1 R6

R7

R8 R2

4 µm (a)

(b)

FIGURE 5.13 (a) Schematic of recrystallized multiple twinned grain in Cu-0.03% P showing microstructure after 80% strain and 20 min annealed at 500°C; (b) twin chain corresponding to the microstructure.33 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.; after P. Haasen.)

5.16

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.17

FIGURE 5.14 Schematic diagram illustrating (a) the basic characteristics of the kinetics of the primary recrystallization process and (b) the variation of the radius of the new recrystallized grain with isothermal annealing time during the growth stage of primary recrystallization.

the recrystallized volume per nuclei is (4p/3)[G(t - t0)]3. Since the number of nuclei increases at the same time as the growth occurs, we can express dn = N◊ [1 - f (t )]dt

(5.18)

where dn is the number of recrystallized nuclei formed in time interval dt and for the recrystallized volume fraction of the material at time t. Equation (5.18) is valid for the actual recrystallization process, and it does not take into consideration the boundary impingement taking place, in some portion of the material, even at the early stage. The boundary impingement causes a reduction of the actual number of nuclei formed at any instant by an amount represented by the ghost nuclei which would have formed if the recrystallization had not advanced to this stage. The extended number of nuclei dnex is the sum of the actual nuclei dn and the ghost nuclei dn¢ at any stage of recrystallization. Thus we can rewrite Eq. (5.17) in the following form:

◊ (t )dt = N◊ dt dnex = dn + dn¢ = N◊ [1 - f (t )]dt + Nf

(5.19)

The extended volume fraction transformed fex(t) can thus be written as fex (t ) = Ú

t

0

4p ◊ [G(t - t0 )]3 Ndt 3

(5.20)

If we further assume N˙ and G to be constant and t0 = 0, we get fex (t ) =

p ◊ 34 NG t 3

(5.21)

Since the actual and ghost nucleus formed in any time interval dt will occupy the same volume per nucleus, we can write, according to Johnson and Mehl,36 the following relationship between extended volume fraction fex(t) and actual recrystallized volume fraction, f or f(t):

5.18

CHAPTER FIVE

untransformed volume df (t ) dn = 1 - f (t ) = = total volume dnex dfex (t )

(5.22)

This simple differential equation reduces to f or f (t ) = 1 - exp[ - fex (t )]

(5.23)

The above analysis assumes that 1. The grains grow isotropically in three dimensions until impingement occurs. The JMAK exponent n decreases if the grains are constrained to grow in one or two dimensions. 2. Nucleation sites are assumed to be randomly distributed. Combining Eqs. (5.21) and (5.23), we get -p ˆ ◊ 3 4 f (t ) = 1 - expÊ NG t Ë 3 ¯

(5.24)

For the three most common forms of specimens (wire, sheet, and lump), the limiting equations for isothermal recrystallization, as derived by Johnson and Mehl,36 are as follows: For one-dimensional recrystallization,

◊ Ê -hNGd w2 t 2 ˆ f (t ) = 1 - expÁ ˜ Ë ¯ 2

(5.25)

where dw represents the wire diameter. For two-dimensional recrystallization,

◊ Ê -hNd sG 2 t 3 ˆ f (t ) = 1 - expÁ ˜ Ë ¯ 3

(5.26)

where ds represents the sheet thickness. For three-dimensional recrystallization,

◊ Ê -hNG3 t 4 ˆ f (t ) = 1 - expÁ ˜ Ë ¯ 4

(5.27)

where h is the shape factor. Most investigators agree on the following fraction softening equation due to Johnson, Mehl, Avrami, and Kolmogorov (JMAK) who assumed N˙ to decrease exponentially with increasing time and G to be the same for all grains: f (t ) = 1 - exp( - kt n )

(5.28)

where f(t) and t have the usual meanings; k is a kinetic parameter related to constant growth rate, nucleation rate, and a shape factor; and n is the annealing or JMAK exponent. The value of n lies between 1 and 2 for one-dimensional recrystallization (i.e., nucleation in wire); between 2 and 3 for two-dimensional recrystallization (i.e., nucleation in thin sheet), and between 3 and 4 for three-dimensional (i.e., nucleation in lump form) recrystallization. This equation can be used to obtain the isothermal annealing curves in Fig. 5.15. Both the Johnson-Mehl and the Avrami equations will be discussed again in Chap. 7. Thus, the n value can provide useful fundamental information regarding the type and morphology of grain growth. Rearranging Eq. (5.28) and taking logarithms of both sides, we get

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.19

FIGURE 5.15 Percentage recrystallization as a function of annealing time at a fixed temperature. (Courtesy of H. Pops.)

log or

log log

1 = kt n 1 - f (t )

1 = log k + n log t 1 - f (t )

(5.29) (5.30)

which is a straight-line equation whenever the parameters log{1/[(1 - f(t)]} and annealing time t are plotted on double-log paper. In this case the annealing exponent is defined as the slope of the line, as shown in Fig. 5.16 for a large range of oxygen concentration and annealing temperatures. Slopes for all compositions and annealing temperatures are nearly equal to 1, and calculated n values are listed in Table 5.1. The fact that n is less than 2 and independent of oxygen content gives significant insight into the nature of the nucleation rate.37 To measure the JMAK exponent, it is essential to correct the time axis of the JMAK plot to t - t0, where t0 is the incubation time. In some recent studies by Samajdar and Doherty37a on the recrystallization of warm deformed aluminum, there was an extensive incubation period for samples deformed 83%; but when deformed 96%, the failure to correct for incubation time yielded absurdly high values of JMAK exponent (n = 25) for the material deformed 83%. However, a more reasonable value of n = 2.4 was found for the material that showed no incubation time.37a 5.4.2.2 Microstructural Path Methodology. A significant attempt to improve the JMAK model has recently been made by Vandermeer and Rath38–40 using the microstructural path methodology (MPM). In this model, more realistic and more complex geometric approaches are employed by using additional microstructural properties in the analysis and, if required, relaxing the uniform grain impingement constraint. The essential features of MPM are (1) the extraction of more detailed information about nucleation and growth rates from experimental measurements than those obtained by the JMAK approach, (2) greater flexibility than in the

CHAPTER FIVE

5.20

FIGURE 5.16 Typical isothermal transformation plots showing the effects of annealing time upon log{1/[1 - f(t)]}. Recrystallization data obtained for different oxygen contents and annealing temperatures.37 (Courtesy of H. Pops.)

TABLE 5.1 Values of Exponent n in Annealing Time-Transformation Equation.† All wires drawn 65% reduction area prior to annealing37 Annealing temperature, °C Oxygen, ppm 10 175 358 642 1000 †

149

204

260 1.2

1.50

0.88 1.23 1.27 1.41 1.13

n

f(t) = 1 - exp(- kt ) = volume fraction transformed.

JMAK model, (3) spatial or random distribution of recrystallized grains, and (4) consideration of global microstructural properties such as extended volume fraction of recrystallized grains and extended interfacial area per unit volume.4 As in the JMAK model, it is simple to use the concept of extended volume fraction recrystallized fex(t) or fex(= Ú V dn¢) [Eq. (5.23)] and the extended interfacial area per unit volume between recrystallized and unrecrystallized (deformed) grains by means of Eq. (5.31), due to Gokhale and DeHoff,41 Sex = Sex (t ) =

S 1- f

(5.31)

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.21

where V is the volume of the recrystallized grains and S is the interfacial area per unit volume between recrystallized and unrecrystallized grains. The progress of isothermal recrystallization, which can be determined metallographically, is given by t fex (t ) = N˙ (t )V(t -t )dt (5.32)

Ú

0

0

and t Sex (t ) = Ú N˙ (t )S(t -t 0 )dt 0

(5.33)

where V(t-t0) and S(t-t0) are the volume and interfacial area, respectively, at time t, of a grain which has nucleated at time t0¢; N˙ (t) dt is the number of new nodules nucleated per unit volume in the time interval between t0 and t + t0. If it is further assumed that the grains have the same spheroidal shape which is preserved during growth, the volume V(t-t0) and interfacial area S(t-t0) of individual nucleated grains are given by V(t -t 0 ) = Kv a(3t -t 0 )

(5.34)

S(t -t 0 ) = K sa(2t -t 0 )

(5.35)

where Kv and Ks are shape factor constants and a(t-t0) is a major semiaxis of the spheroid, which is related to the interface migration rate G(t) by t

a(t -t 0 ) = Ú G(t )dt t0

(5.36)

where a(t-t0), being a local property, may be predicted by measuring the diameter of the largest unimpinged grain intercept DL on a plate polished surface. By assuming DL to be the diameter of the earliest nucleated grain, we get DL (5.37) 2 For isothermal recrystallization, fex, Sex, and DL can be expressed by the timedependent power law functions of the form 1 fex = ln = kt n 1- f (5.38) S m Sex = = Kt (5.39) 1- f DL = Sts (5.40) If the derived functions are also assumed to be expressed by power law functions, we have a(t -t 0 ) =

and

N˙ = N1td-1

(5.41)

a(t-t0) = Ga(t - t0)r

(5.42)

where N1, d, Ga, and r are constants. Since shapes of spheroidal grain remain preserved during growth, a(t) = DL/2, and thereby Ga = S/2, and r = s. Vandermeer and Rath show that, for spheroidal grains, which preserve their shape during growth, d and r become d = 3m - 2n

(5.43)

CHAPTER FIVE

10

10000

1

1000

Sv/(1–Xv)

In [1/(1–Xv)]

5.22

0.1

0.01 0.1

1

10

100

10 0.1

Normalized time

1

10

Normalized time

(a)

(b)

Largest grain (µm)

80

60 40 20 0

0

1

2 Normalized time (c)

3

4

FIGURE 5.17 Annealing kinetics of an iron crystal deformed 70% and annealed at various temperatures showing the effect of normalized annealing time on variation of: (a) fex (= Xvex), (b) Sex (= Svex), and (c) largest recrystallized grain diameter.19 (Courtesy of R. A. Vandermeer and B. B. Rath.)

r=s=n-m

(5.44)

For Eqs. (5.38) to (5.42), it is evident that d = 1 represents the case of constant nucleation rate, d = 0 denotes the site saturation nucleation, and r = s = 1 represents a constant growth (or interface migration) rate. Hence, by measuring f, S, and DL experimentally as a function of annealing time, we can determine the values of n, m, and s experimentally; calculate d and r; and identify the form of the nucleation kinetics. In contrast, if the JMAK model is applied for constant nucleation and growth rates, we obtain n = 4, m = 3, s = 1, d = 1, and r = 1. Experimental Application. Figure 5.17 represents the recrystallization kinetics of deformed (111)[112] iron crystals at various temperatures which illustrates the variations of f(= Xv), S(= Sv), and the largest unimpinged recrystallized grain diameter as a function of normalized annealing time. The slopes of these lines in Fig. 5.17a, b, and c, respectively, yield n, m, and s values. The values obtained are n = 1.90, m = 1.28, and s = 0.60 which, in turn, give d = 0.04 and r = 0.62. It is inferred from these data that nucleation is site-saturated (due to negligibly small d value), the grains grow as spheroids (due to r ~ s), and growth rate decreases with time (due to low r value). However, for s π r, the grain shape varies during recrystallization.19 The model has also been extended to incorporate the influence of recovery during the recrystallization anneal.39

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.23

FIGURE 5.18 Radius of largest grain versus isothermal annealing time at 350°C for 2.8 and 5.1% elongation of aluminum.42 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

.

5.4.3 Experimental Determination of G and N For the measurement of G, a series of identical specimens are first given a specific amount of strain, then isothermally recrystallized for different lengths of time, and finally quenched to room temperature. The radius of the largest grain (which has not suffered impingement with another growing grain) observed individually in metallographically prepared specimens is plotted as a function of time of isothermal recrystallization, as shown in Fig. 5.18.42 The slope of this curve determines the rate of growth G, while the intercept on the time axis determines the incubation period t0. For the determination of N˙ , first the number of new recrystallized grains per unit area Ns is plotted as a function of time, as shown in Fig. 5.19a. The slope of this curve yields the (two-dimensional) surface nucleation rate N˙ s, as illustrated in Fig. 5.19b. The two-dimensional nucleation rate can be approximately related to the three-dimensional nucleation rate N˙ , which can be found in any book on quantitative metallography.

.

5.4.4 Effect of Process Variables on G and N The concepts of N˙ and G are useful in explaining the influence of different variables on the recrystallization process. Since G is invariant with time, its values can be easily compared. On the other hand, N˙ rapidly increases with increasing temperature whereas N˙ varies with time: Hence it is not directly compared. Figure 5.20 shows the effect of the extent of deformation and grain size, respectively, on G. The results represented in Figs. 5.20a and 5.18 illustrate that the growth rate increases up to about 10% strain; beyond this level, G still increases but at a slower rate up to about 15% strain. Figure 5.20a also demonstrates the plot of incubation time versus extent of deformation. The incubation time falls rapidly with strain to a small level at about 15% strain, beyond which it is constant. This implies ease of nucleation (i.e., accelerated recrystallization) at high strain.

CHAPTER FIVE

5.24

0.07 Ns, number / cm2 – sec

Ns, grains / cm2

120 100 5%, 350°C

80 60 40 20 0

2,000

4,000 6,000 Time, sec

0.06 5%, 350°C

0.05 0.04 0.03 0.02 0.01

8,000 10,000

0

0

2000

(a)

4000 6000 Time, sec

8000

(b)

FIGURE 5.19 (a) Aluminum after 5.1% elongation: surface density of the recrystallized grains versus time.42 (b) Aluminum after 5.1% elongation: surface nucleation rate N˙ s at the surface versus time.42 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

% elongation 5

10

15

30

20

7.5

AI 329°C

10

20

5

G

20

5.0 N G

N G

G

N or G × 10–6

30

1000 sec

G × 10–6 cm/sec

350°C 15

10

2.5

10

N

i.t. 0

0

0.05

0.10

0.15

0.20

0

× 105

0

0

0 0

5

10

15

Effective strain

Percent elongation

(a)

(b)

20

FIGURE 5.20 (a) Effect of prior strain in aluminum on growth rate G and induction period t0 during recrystallization at 329°C.42 (b) Effect of a prior strain in aluminum on growth rate G and nucleation rate N˙ during recrystallization at 350°C.42 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

5.4.5 Effect of Recrystallization Time and Grain Size The important characteristic of a transformation (e.g., recrystallization) is the time needed to transform a given volume fraction. We do not experimentally determine the time for f(t) = 1.00 because it is very difficult. However, we often calculate the time to achieve a given volume fraction of transformation, say, f(t) = 0.5 or 0.95. We get the following relationship by using Eq. (5.24): 0.662 t0.5 = Ê ◊ 3 ˆ Ë NG ¯

14

(5.45)

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.25

FIGURE 5.21 Effect of time and temperature on rate of linear growth in zone-melted iron (cold-rolled 60%) showing rapid decrease of linear growth rate with time at constant annealing temperature.18 (Reprinted by permission of John Wiley & Sons, New York.)

where t0.5 denotes the time required to obtain 50% volume transformed. We know that both N˙ and G increase when the temperature is raised. With this in mind, Eq. (5.45) indicates that we can get a certain volume fraction of recrystallization accomplished in a shorter time by increasing the values of N˙ and G, which, in turn, lowers the 1-hr recrystallization temperature. That is, there is an inverse relationship between Trecry and N˙ and G. It can also be generalized from Eq. (5.45) that a change in G has a far-reaching influence on t0.5 or any other annealing time corresponding to a given volume fraction of transformation. In other words, there is a sharp decrease of G with increasing annealing time compared to that of N˙ which has been experimentally found in cold-worked and annealed zone-refined iron specimens (Fig. 5.21).18 The number of nuclei formed per unit volume is given by N◊ N◊ t0.5 = 0.90Ê ˆ ËG¯

3 4

(5.46)

If d is the average diameter of the recrystallized nuclei, the number of nuclei per unit volume will be roughly 1/d3, so that G d f (t )=0.5 = 1.035Ê ◊ ˆ ËN¯

14

(5.47)

We can generalize Eq. (5.47) for the recrystallized grain size as G d = (constant )Ê ◊ ˆ ËN¯

14

(5.48)

It is apparent from Eq. (5.48) that the ratio of N˙ /G is useful to determine the recrystallized grain size. Fine recrystallized grain size can be obtained by maintaining a high N˙ /G ratio, that is, high nucleation rate and a slow growth rate or large

CHAPTER FIVE

5.26

number of nuclei formation with very little growth. In contrast, a low ratio represents a slow nucleation rate with respect to the growth rate and thus represents coarse recrystallized grains. Figure 5.20b shows a plot of N˙ and G along with their ratio N˙ /G versus the amount of strain for a typical metal (aluminum). These curves illustrate that as the degree of cold deformation (prior to annealing) decreases, N˙ falls more rapidly than G; that is, the N˙ /G ratio drops more sharply. When the amount of cold deformation decreases below a particular (about 3%) strain, called critical strain (which is different for different metals), the N˙ /G ratio becomes practically zero and, consequently, it is not possible to form recrystallized grains in a reasonable length of time. Note that the amount of critical strain is very different for different metals. Many factors restrict the number of recrystallized nuclei which successfully grow into fully formed grains. Such factors greatly reduce the final grain density. Other factors influencing the recrystallized grain size include annealing temperature and heating rates used in the annealing treatment. A high annealing temperature yields a high density of potent nuclei and a fine grain size if the grain growth and secondary recrystallization are absent. A high heating rate leads to a higher temperature for the start of recrystallization. Hence, the density of the active nucleation sites is increased, and the likelihood of a strongly preferred nucleus starting to grow at relatively low temperature is diminished. A slow heating rate produces excessive growth of a few nuclei, resulting in an extremely coarse and inhomogeneous structure.43

5.4.6 Strain-Anneal Technique Well-developed single crystals are successfully prepared by the strain-anneal technique, in which the specimen is first strained just above the critical strain and then annealed at a slow heating rate and the lowest possible temperature so that the N˙ /G ratio is drastically reduced.

5.4.7 Activation Energies Qn and Qg The activation energy of recrystallization Q is a measure of the driving force, i.e., of the energy difference between the cold-worked and annealed states. Since the Arrhenius equation applies for this process, we can write37 1 = Ae -Q RT (5.49) t Note that activation energy for pure metals is typically one-half that for selfdiffusion.44 Since activation energy for recrystallization increases with increasing impurity contents, oxygen concentration in the high-conductivity Cu wire leads to the formation of metal oxides upon annealing, thereby removing impurities from solid solution.45 Moreover, N˙ and G vary exponentially with temperature (that is, N˙ µ e-Qn/RT and G µ e-Qg/RT), which indicates that nucleation and growth in recrystallization are processes that are controlled by their respective activation energies (for example, Qn and Qg decrease with increasing strain and decreasing initial grain size). Surprisingly, the values of both Qn and Qg have been found to be of the same order of magnitude, provided the strain is higher than about 5%.42 Recrystallization rate =

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.27

5.5 RECRYSTALLIZATION OF TWO-PHASE ALLOYS As most commercial alloys contain more than one phase, an understanding of the recrystallization behavior of two-phase materials is of great importance and scientific interest. The second phase may be in the form of dispersed particles, which are present during the deformation; or, if the matrix is supersaturated solid solution, the particles may precipitate during subsequent annealing. In aluminum-treated low-carbon steel, it has been observed that a low annealing temperature results in the precipitation of AlN on sub-boundaries prior to recrystallization, thereby causing a considerable retardation of the process. However, high annealing temperature accelerates the recrystallization to such a degree that the steel may completely recrystallize before the onset of precipitation of AlN.46 In steels, the presence of pearlite or spheroidized cementite has been found to accelerate recrystallization.47

5.5.1 Particle-Stimulated Nucleation Coarse, widely dispersed second-phase particles can produce a local concentration of lattice distortion caused by the applied deformation which, in turn, accelerates recrystallization by altering N˙ rather than G. This partial stimulated nucleation (PSN) has been observed in many Al, Cu, Fe, and Ni alloys and is usually only found adjacent to large (≥1 mm) particles; however, a lower limit (0.8 mm) was deduced from the indirect measurements of Gawne and Higgins on Fe-C alloys.48 In this manner, rapid recrystallization leads to a final grain size of the order of interparticle spacing, in a relatively pure Al matrix. In most cases, a maximum of one grain is found to nucleate at any particle, although it seems that multiple nucleation occurs at very large particles, as seen in Fe-O alloy (0.33% O2) (Fig. 5.22). Another example of PSN is the precipitation of ~1-mm FeAl3 particles in Fe-Al alloy which enhances N˙ .

FIGURE 5.22 Particle-stimulated nucleation of recrystallization at oxide inclusions in iron; 60% rolling reduction, 2 min at 540°C.18 (Reprinted by permission of John Wiley & Sons, New York.)

5.28

CHAPTER FIVE

A dense distribution of finely dispersed second-phase particles has been found to hinder recrystallization by delaying both the nucleation and growth processes.40 For example, dense distributions of fine (0.1- to 1-mm) particles have a retarding effect on the growth rate, which results in a fine grain structure. However, microstructural inhomogeneities often lead to the preferential growth of just a few grains. The effect of a fine and dense particle dispersion is thus an increase in Trecry, and also a coarse final structure.43 Very small ( D0, Eq. (5.68a) reduces to D = Kt

(5.69)

Various models and theories of grain growth predict that at large times (5.70) D = K ¢t 1 n where t is the annealing time, K¢ is a proportionality constant that depends on material composition and temperature, and n is the grain growth exponent. A grain growth exponent n = 2 (i.e., parabolic growth law) was predicted by the simple theory of Burke and Turnbull103 and is mostly considered as an ideal value for n. Mullins and Venals104 have predicted n = 2 for a single-phase polycrystal. However, experimental values of n are usually larger than 2 and may change with time. These are called normal grain growth equations. 5.9.1.5 Grain-Coarsening Behavior. Coarse grains are usually avoided during cold-working operations because of the reduction in strength, toughness, and hardness with increasing grain size. Moreover, the plastic flow under stress becomes uneven, which causes the smoothness of the metallic surface to be impaired; this produces surface roughness or orange-peel appearance. Hence, the annealing conditions are selected in such a way as to limit the secondary recrystallization.105 The coarsening or Ostwald ripening of the second-phase particles used to inhibit austenite grain growth in steel is usually found to follow the Wagner equation for diffusion-controlled growth 8gD[M ]Vmt (5.71) 9RT where r is the particle radius after a time t; r0 is the initial particle radius; g is the interfacial energy; D is the diffusion coefficient of the relevant (or rate-limiting) species; [M] is the dissolved content of the solid; Vm is the molar volume; R is the gas constant; and T is the absolute temperature. Plain carbon steels without grain-refining additions exhibit grain coarsening by normal grain growth (Fig. 5.37). When grain-refining particles such as AlN, VC, VN, NbCN, TiC, and TiO are added, fine grains persist up to a coarsening temperature, and the grain-coarsening temperature increases. In Al-killed steels, growth inhibition fails by abnormal grain growth process at temperatures between 1050 and ~1200°C (Fig. 5.37a). Above this temperature, normal grain growth occurs. However, if a coarser dispersion of more stable particles such as oxides or TiCN is present (Fig. 5.37b), all grain growth is hindered up to a very high temperature. r 3 - r03 =

5.9.2 Formation of Annealing Twins Annealing twins have been commonly observed in recrystallized structures, particularly in fcc metals (such as Cu-group metals and alloys, lead, nickel-base superalloys, and austenitic steels) and disordered alloys as well as in many intermetallic

CHAPTER FIVE

5.48

Abnormal

–2

0 Plain carbon steels

2

ASTM Grain Size No.

ASTM Grain Size No.

–2

4 Mixed 6 8 900

Fine 1000 1100 Temperature, C (a)

1200

0

Carbon Plain Steel

2 4 6 8 900

Ti-0 steel 1000 1100 1200 Temperature, C

1300

(b)

FIGURE 5.37 (a) Austenite grain sizes in plain carbon and aluminum-killed low-carbon steel at various tyemperatures showing normal and abnormal grain growth. (b) Austenite grain sizes at various temperatures in Ti-O grain-refined steels showing normal grain growth.97,98

FIGURE 5.38 Annealing twins in annealed 70-30 brass. Some coherent twin (CT) and incoherent twin (IT) boundaries are marked.4 (Courtesy of M. Ferry.)

compounds, ceramics, and minerals. These twins take the form of parallel-sided lamellae, bounded by {111} planes or coherent twin (CT) boundaries and at their ends, steps or terminations, by incoherent twin (IT) boundaries, as shown schematically in Fig. 5.38. Twins may form during recovery, recrystallization, and grain growth (see also Sec. 4.7). Grindvaux and Form106 have established, by direct observation, that most annealing twins occur during primary recrystallization and few additional twins form during subsequent grain growth. When coherent twin interfacial energy gT is greater than the ordinary grain boundary energy gB, as in aluminum, where gT/gB = 0.2, then annealing twins are rarely visible.

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.49

FIGURE 5.39 Mechanism for twin formation during grain growth following recrystallization.4

The important models for the formation of annealing twins can be classified into three groups: (1) growth accident model, (2) grain encounter model, and (3) model consisting of nucleation twins by stacking faults or fault packets. The first model assumes that a coherent twin boundary forms at a migrating grain boundary due to a stacking error during growth under energetically favorable conditions. The second model assumes that different grains initially separated “encounter” each other during grain growth. If these grains tend to be in twin orientation to each other, the boundary between them becomes a coherent twin boundary after reorienting itself. In the third model, a grain boundary during its migration nucleates a twin so that its incoherent segment lies at the grain boundary. The twin then grows presumably by the migration of the other noncoherent boundary.107 Figure 5.39 shows the postulated mechanism of twin formation during grain growth following recrystallization. As grain growth occurs, it is assumed that the triple point between grains A, B, and C moves vertically. As grain growth advances, a growth fault may be generated, and a twin will appear at the preceding B-A-C triple point, and such a fault will be stable in order to grow the twin (T), as shown in Fig. 5.39c. If the relative orientations of the grains are such that the energy of the boundary AT is lower than that of AC, then, because the energy of the coherent twin boundary AT is very low, there may be a reduction in total boundary energy albeit the extra boundary area is created, and thus the twin configuration will be stable and grow. This condition, in two-dimensions, is given by gATL13 + g TCL23 + g TBL12 < gACL23 + gABL12

(5.72)

where gij is the energy of the boundary between grains i and j and Lxy is the distance between points x and y.4 The growth will stop if the triple point ABC reacts with another triple point, say, BCE, leading to a grain configuration with a less favorable energy balance (Fig. 5.39c). On this model, the number of twin lamellae should vary linearly with the number of triple-point interactions; this evidence was observed by Hu and Smith.108 The actual atomistic mechanisms of twin formation during grain growth and during recrystallization are likely to be similar. Annealing twins have also been observed in bcc metals and alloys in which the crystallography is more complex in that one twin can possess three distinct coher-

5.50

CHAPTER FIVE

FIGURE 5.40 Annealing twins in an Fe-Al alloy after prolonged grain growth.8 (After R. W. Cahn and J. A. Coll, Acta Metall., vol. 4, 1961, p. 683.)

ent interfaces, as shown in Fig. 5.40. The low energy of the coherent boundary can be inferred from the fact that the normal grain boundary is only slightly diverted where a coherent boundary adjoins on it (at P). Recently, the methods of mesotextural measurement have been used to study the twin formation during recrystallization and grain growth.109 Twin formation has also been cited in one model of nucleation in primary recrystallization, namely in Cu; this model was fully established by Haasen.33

5.9.3 Secondary Recrystallization When the annealing of an initially deformed material is continued for a long time (even) after the complete formation of a (primary) recrystallized structure, a few grains start growing preferentially and very rapidly until they impinge upon one another. Thus, they consume all small neighboring (recrystallized or normal) grains and produce a very coarse-grained structure, on the order of several centimeters. In extreme cases, single-crystal metals can be produced, for example, by annealing commercially pure Mo wire, drawn to a proper degree, at 2000°C.110 This is called secondary recrystallization or abnormal growth. The requirement for secondary recrystallization is the strong impediment of normal grain growth, with the exception of a few grains that act as “nuclei” for secondary recrystallization; here the large grains are not freshly nucleated but are merely large-grown grains of the primary structure. The discontinuous growth of selected grains has similar kinetics to primary recrystallization and has some microstructural similarities, as shown in Fig. 5.41. Secondary recrystallization is promoted if one or more of the following inhibiting conditions for normal grain growth are realized:105

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.51

FIGURE 5.41 Secondary recrystallization in Fe-3% Si during an anneal at 1373K.105 (Detert, 1978.)

1. Introduction of an array of less stable particles capable of totally inhibiting grain growth even for a grain with an infinite size advantage favors secondary recrystallization.97 Examples are the presence of finely dispersed second-phase particles such as MnS and AlN in Fe-3% Si alloys and at least 1% retained austenite in certain grades of 12% Cr steels used for the manufacture of turbine disks.111 As the grain growth inhibitors coarsen and dissolve, abnormal grain growth occurs (Fig. 5.37). Abnormal grain growth is viable in alloys where normal grain growth has stagnated due to particle pinning. 2. The average grain diameter D after primary recrystallization has reached a limiting value which is twice the thickness of the sheet t or is equal to the diameter of the wire d. That is, grain coarsening stops when D = 2t or D = d. Hence, abnormal growth may not be possible. 3. Presence of a strong single-orientation texture component in a fine-grained recrystallized material favors the formation of abnormal grain growth on further annealing at high temperatures. The grain boundaries within a highly textured volume have a lower misorientation and, therefore, a much lower energy and mobility than those within a normal grain structure. 4. Abnormal grain growth may result either from an abnormally high boundary mobility or from a higher driving force. Abnormal grain growth is possible if abnormal grain growth is faster than the growth in the average grain assembly. 5. It is obvious from condition 2 that sufficient thermal energy or Ostwald ripening is necessary to facilitate significant grain boundary displacement. That is, abnormal grain growth is likely to occur if the temperature is raised above the graincoarsening temperature and as the particle dispersion becomes unstable. The coarsening temperature is defined as the temperature at which particles can grow to, or exceed, the critical radius (i.e., limiting size).111 Like primary recrystallization, secondary recrystallization is a nucleation and growth process. An incubation period is usually observed. A plot of the volume fraction of the secondary (recrystallized) grains versus isothermal annealing time produces a characteristic sigmoidal curve. Figure 5.42 shows such a curve which represents the kinetics of secondary recrystallization in Fe-3% Si alloy, upon isothermally annealing, producing the cube texture.112 Like primary recrystallization, the progress of secondary recrystallization can be described by the Avrami relationship

CHAPTER FIVE

% Secondary recrystallization

5.52 100 80 60

Anneal temperature 1050°

40 20 0

20 40 60 80 Anneal time, min

100

FIGURE 5.42 Sigmoidal curve for the secondary recrystallization for the formation of the cube texture in Fe-3%Si upon isothermal annealing.22,112 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

f (t ) = 1 - exp( - kt n )

(5.73)

where f(t) is the fraction of secondary recrystallized grains, t is the annealing time, and k and n are constants. However, the driving force for secondary recrystallization is the large decrease in total interfacial energy of the grain boundary. Dunn and Walter113 have given the following equation expressing the relationship between relative contributions of grain boundary, surface energy, sheet thickness, and driving force of a secondary grain in a weakly textured matrix: Grain boundary energy term g b r - g b r = 2 Dg s t Surface energy term

(5.74)

where gb is the grain boundary energy per unit area, Dgs is the difference in surface energy, ¯r is the average grain radius in a stable matrix, r is the radius of the secondary (or potential secondary) grain, and t is the sheet thickness. This ratio is maximum when r >> ¯r, and it is zero when r = ¯r.

5.9.4 Commercial Application of Secondary Recrystallization Principles A better solution for avoiding the intergranular fracture problems associated with the application of fine filamentary Mo and W wires for electron tubes, emitters of thermoionic electrons, incandescent lamps, and so forth is to use them as long, pure, single-crystal wires. This is achieved by exploiting the secondary recrystallization phenomenon.110 A coarse-grained or secondary recrystallized structure is highly desired in the production of soft ferrous magnetic sheets (e.g., Fe-Si and Fe-Mn alloy sheets) to be used as transformer cores. Low coercive force (or magnetizing force H) is a prerequisite in the design of soft magnets used as transformer cores. The presence of (1) dispersed second-phase particles, (2) minimal nonmagnetic inclusion, and (3) the negligible orientation differences between the neighboring grains are characteristic features for the production of such soft magnetic materials, because all these reduce coercive force. Coercive force decreases with increasing grain size because the domain pattern across the grain boundary is simple as a result of the smaller orientation difference between the neighboring grains. A simple domain pattern can also be produced at the free surface when this is parallel to the direction of easy magnetization. Thus, coercive force can be reduced to a very low level by using rolling and anneal schedules to produce either a cube-textured or a cube-on-edge textured material.114

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.53

When a strong cube-texture is obtained, two easy directions lie in the plane of the sheet parallel to the rolling and transverse directions. On the other hand, when a cube-on-edge texture is produced, it makes the easy direction in all grains almost parallel to one another and to the rolling direction of the sheet. The former texture has superior magnetic properties relative to the latter texture.13 This technique is very important because transformer material is applied in the form of thin sheets to achieve minimum hysteresis loss. When strips made of Fe-3% Si alloy containing a fine dispersion of MnS particles are given a final treatment in the temperature range of 900 to 950°C, it produces a (110) [001] cube-on-edge (or Goss) texture of the secondary grains with a directionally preferred (“easy”) magnetization along the strip-rolling direction, thereby requiring a smaller mass of transformer laminations, usually with a 0.30- to 0.35-mm thickness.105 Dunn and Walter115 have produced the secondary recrystallization texture in high-purity irons and 0.6% Si-Fe. They have found that the surface energies control the preferred orientation, which, in turn, depends on the annealing atmosphere. If the annealing atmosphere is contaminated with high O2 content, g100 (surface energy of the {100} planes) becomes the lowest {hkl} surface energy that leads to the preferential grain growth in {100} orientation. On the other hand, if O2 is low or absent in the atmosphere, g110 (surface energy of {110} planes) becomes the lowest {hkl} surface energy that permits the secondary grains to grow preferentially in the (110) [001] orientation. Thus secondary recrystallized texture can be either {100} or {110} , or a mixture of both, depending upon the O2 content in the annealing atmosphere.100 Another variation to improve and obtain a sharp preferred (110) [001] texture is accomplished by the controlled dispersions of small AlN particles which precipitate in the deformed material. Minor additions of boron, sulfur, and nitrogen which segregate preferentially at grain boundaries have also been used by Grenoble and Fiedler to improve the preferred (110) [001] texture.116 In contrast to the findings of Dunn and Walter,115 Taguchi and Sakakura117 have observed that when single crystals of 3% Si-Fe alloy containing needlelike AlN particles and having (001) [100] orientation were rolled with 70 to 80% reduction and annealed between 800 and 1000°C, they noticed the secondary recrystallization texture with the initial (001) [100] orientation, irrespective of the purity and thickness of crystals, the purity of gas atmosphere, and the formation of surface scales. The driving energy for this texture is surprisingly not the surface energy. According to them, the effectiveness of AlN to cause (001) [100] secondary recrystallization texture should have the following characteristics: 1. Very fine needles (~1 mm in length) of AlN, which precipitate at 800 to 1000°C, are easily dissolved into solution within a short period by raising the temperature to, say, 1200°C, for 5 min. However, they reprecipitate on slow cooling. 2. These fine needles have an hcp structure with a = 3.104 Å, c = 4.965 Å, and c/a = 1.60. 3. The orientation relationship between needles and parent crystal, as obtained by analysis of the electron diffraction pattern, is given as follows:

{10.1}AIN // {120}a

(5.75a)

{12.2}AIN // {122}a

(5.75b)

Mostly, the needles tend to precipitate on {100}a or {120}a.

5.54

CHAPTER FIVE

Secondary recrystallization with either (100) [001] or (110) [001] texture has also been reported in Fe-3.25% Si rolled strips from sintered compacts by controlling the purity, particle size of the metal powder, temperature, and O2 content of the annealing atmosphere.118 5.9.4.1 Production of Grain-Oriented Silicon Steel Sheets. Two processes are used to manufacture grain-oriented silicon steel (GOSS) sheet as a transformer core material for electrical applications. The first is called the Armco or two-stage coldrolling process, as summarized in Table 5.4. The essential requirements in the Armco process include (1) nucleation of {110} grains, (2) ability of these grains to grow, and (3) grain growth inhibition of other orientations. The process consists of initial hot rolling, two light cold-rolling stages, and three annealing stages. The necessary conditions for their abnormal growth are achieved by microstructural control. A fine dispersion of MnS precipitates produced by rapid cooling of the slab prior to hot rolling is resistant to rapid grain coarsening, which maintains the small matrix grain size during the early stages of the high-temperature annealing. As Ostwald ripening and dissolution progress, abnormal grain growth with a strong cube-on-edge or GOSS {110} texture is produced. Abnormal grain growth is also favored by the presence of sharp texture. Preferential grain growth in desirable orientations at the surface is also achieved by the addition of S to the MgO surface coating. This prevents the growth of surface grains, and the sulfide formed is eventually eliminated by reaction with the H2 atmosphere for steels containing ~0.25% Al and ~0.01% N. The second method, called the Nippon Steel process, was developed by Taguchi et al.119 and Sakakura.120 This process consists of a single-stage cold-rolling reduction process and requires both MnS and AlN precipitates as grain growth inhibitors. This process involves rapid cooling after hot rolling step, large cold rolling, decarburization at 850°C in wet H2 atmosphere, and addition of metal nitrides and sulfur in the final annealing treatment to control the decomposition of AlN particles. Although both processes are being continually refined, it is believed that the Nippon Steel process leads to a stronger GOSS texture but to a large grain size. A cube-oriented sheet has been described by Arai and Yamashiro121 for a 3.26% Si-Fe alloy directly cast to 0.37-mm thickness prior to final cold rolling. Here surface-energy-controlled grain growth occurred with a strong {100} texture in the final sheet thickness of 0.15 mm without using any inhibitor. TABLE 5.4 Processing of Grain-Oriented Silicon Steel Sheets4 Armco process a) Soak at ~1340°C; hot-roll to 2 mm. b) Cold-roll to 0.5 to 1 mm. c) Anneal at 950°C in dry H2/N2 (80 : 20). d) Cold-roll to finished size. e) Decarburize at 800°C in wet H2; dew point 50°C. f) Coat with MgO. g) Texture anneal at 1150°C in pure H2.

Nippon Steel process a) Soak at ~1350°C; hot-roll to ~2 mm. b) Hot-band anneal 1125°C, air-cool to 900°C, water-quench to 100°C c) Cold-roll d) Decarburize at 850°C in wet H2; dew point 66°C. e) Coat with MgO + 5% TiO2. f) Texture anneal at 1200°C in H2/N2 (75 : 25).

RECOVERY, RECRYSTALLIZATION, AND GRAIN GROWTH

5.55

5.9.4.2 Surface-Controlled Secondary Recrystallization. A process called surface-controlled secondary recrystallization or alternatively tertiary recrystallization, was discovered by Detert122 and Walter and Dunn123 in thin ( r*, a critical embryo (called a nucleus) will grow in size. It is clear that further increase or decrease in size corresponding to the critical value reduces the DG value. The activation energy for nucleation or the free-energy change of formation of the critical nucleus, in the presence of elastic strain energy, is obtained after combining Eqs. (6.5) and (6.6) as DG*Hom =

3 16pg ab 2 3(DGv + DGe )

(6.7)

Alternatively, Eq. (6.5) can be expressed in the following form because it is convenient to focus attention on the number of atoms in the embryo instead of its radius: DG = n DGv + hn2/3gab + n DGe

(6.8)

where n is the number of atoms in the embryo and h is the shape factor. Similarly, the critical number of atoms and the critical free-energy change accompanying the formation of a potent coherent nucleus are given by 2hg ab È ˘ n* = Í Î 3(DGv + DGe ) ˙˚ and

DG*Hom =

3

3 16h3g ab 4 27 (DGv + DGe )2

(6.9)

(6.10)

Note: Equations (6.6), (6.7), (6.9), and (6.10) are applied for coherent nucleation with sharp interfaces.

6.2.1 Strain Energy Strain energy plays an important role in the nucleation of solid-state phase changes. Strain energy developed by the formation of a new phase in a matrix phase is grouped into two types. The first one is caused by lattice mismatch or misfit strain due to different lattice parameters between two different crystalline phases. The second one is associated with the volume strain or dilatational strain due to the volume difference between two structures.10 Thus the elastic energy is a strong function of morphologies of the two phases such as shapes, orientations, and mutual arrangements of the precipitates.11 The elastic strain energy associated with a coherent second-phase precipitate has been studied by Laszlo,12 Robinson,13 and Eshelby.14 Eshelby’s approach, based on the transformation strain, will be presented here. Let us consider an infinite isotropic linear elastic matrix of shear modulus m, modulus of elasticity E, and Poisson’s ratio n which contains a fully coherent ellipsoid precipitate with the corresponding elastic constants m*, E*, and n* formed by transformation. If the transformation of the matrix into the coherent precipitate occurs without producing constraints of the surrounding matrix, the transformation

NUCLEATION IN SOLIDS

6.5

is said to be a stress-free transformation and can be described by its uniform stressfree transformation strain eijT*.8,14 In a pure dilatation transformation, a change in size without a change in shape of the precipitate occurs, which is represented by eijT* = edij, where dij is a Kronecker delta [a second-rank unit isotropic tensor (dij is 1 if i = j and is 0 if i π j)] and e is the constant linear misfit strain. When the elastic constants are the same in the matrix and precipitate particle (that is, m = m*, E = E*, and n = n*), the strain energy per unit volume of the precipitate is given by8,14,15 T E(e 11 ) = 2 m(1 + v)(e11T ) 1-v 1-v 2

DGe = DGe0 =

2

(6.11)

which is independent of precipitate shape. However, when the matrix and precipitate have different elastic constants, DGe becomes shape-dependent (Fig. 6.3).8,15 Figure 6.3 is the variation of DGe/DGe0 with particle eccentricity (or ellipsoid aspect ratio) b = c/r, where c denotes semithickness and r is major radius of the particle for n = 0.291 and µ = 8.6 ¥ 1011 dyne/cm2 (the elastic constants for a-iron), with n* = n and m */m = 1/3, 1, and 3.15 For c/r = •, the shape of the particle becomes a needle; for c/r = 1, the shape becomes a sphere with the maximum strain energy existing; and for c/r tcr, it becomes incoherent. This represents a qualitative picture of the effect of precipitate size on the nature of the precipitate/matrix interface. In fact, the precipitate forms initially with a coherent interface, and as it grows in size, it becomes partially coherent (or semicoherent); finally, coherency is completely lost and the precipitate becomes incoherent. This has been experimentally observed in many solid Æ solid transformations. The coherent precipitates are invariably associated with the orientation–habit-plane relationship (e.g., formation of disk-shaped Guinier-Preston (GP) zones in Al-Cu alloys with orientation— habit-plane relationship {100}zones // {100}matrix). Similarly, Widmanstätten precipitates have been initially found to be coherent with sharp interfaces and an orientation— habit-plane relationship.

NUCLEATION IN SOLIDS

6.7

FIGURE 6.4 Schematic representation of the barrier to nucleation for coherent and incoherent precipitates versus its thickness.16 (Reprinted by permission of John Wiley & Sons, New York.)

6.2.2 Vacancy Effect In coherent nucleation, excess vacancies do not contribute to the driving force. For incoherent nucleation, thermal vacancy concentration relieves any transformation strain that results in a stress-free critical nucleus.8 Hence the change in free energy of forming a spherical incoherent embryo is given by DG =

4 3 pr (DGv + DGvac ) + 4pr 2g ab 3

(6.15)

where DGvac is the driving force due to excess vacancies; other terms have their usual meanings. Similarly, the critical free-energy change DG*Hom,incoh is given by DG*Hom,incoh =

16pg 3 2 3(DGv + DGvac )

(6.16)

Note: For homogeneous nucleation in liquid Æ solid transformation, the strain energy barrier is nonexistent. Hence in this case, Eqs. (6.6), (6.7), (6.9), and (6.10) can be applied after deleting the DGe term. (See Chap. 3 for more details.)

6.3 CLASSICAL HOMOGENEOUS NUCLEATION RATE IN SOLIDS Since the probability of observing any embryo of a given size with n atoms in the equilibrium state is shown to be proportional to exp(-DG/kT), the number of embryos Nn containing n atoms can be expressed by the relation8 DG ˆ N n = N expÊ Ë kT ¯

(6.17)

6.8

CHAPTER SIX

where N is the total number of (active) atomic sites per unit volume, k is the Boltzmann constant, and DG is the total free-energy change represented in Eq. (6.5). Similarly, it can be shown, for an equilibrium number of critical-sized embryos per unit volume, that DG* ˆ N*n = N expÊ Ë kT ¯

(6.18)

The addition of one more atom to each of these critical embryos (or microclusters) will make them stable nuclei. These never decompose; instead, they continue to grow. If this occurs with a frequency b*, the overall nucleation . rate N of a new phase in the solid state is given by the general nucleation equation15,16 DG* ˆ t0 ◊ N = Zb*N expÊ - ˆ expÊ Ë t¯ Ë kT ¯

(6.19)

where Z is the Zeldovich nonequilibrium factor (typically Z ⬵ 0.1); b* is the frequency factor, which is the rate at which a single atom joins the critical nucleus; N is the density of atomic nucleation sites of the matrix phase to establish a steadystate nucleation condition; t0 is the incubation time; t is the isothermal reaction time; and DG* (= DG*Hom) is the critical free energy of activation for nucleation. This equation is not exact. For one thing, it does not take into account the effect of strain energy on nucleus shape. In fact, Lee et al.19 have recently shown that strain energy does not influence the critical nucleus shape unless it constitutes a large fraction of the absolute value of the volume free-energy change. In general, the values of Z, b*, DG*, and t0 can be calculated from a specific model of critical nucleus shape and are all specific to the system and to the type of nucleation .process considered..8,20,21 Usually, the classical nucleation rate consists of transient Nt and steady-state Nss periods, which can be in the form of equations such as DG* ˆ N◊ SS = Zb*N expÊ Ë kT ¯

(6.20)

t0 ◊ ◊ Nt = N SS expÊ - ˆ Ë t¯

(6.21)

and

In the transient period, the nucleation rate rises continuously until the incubation time is reached. In the steady-state period, the nucleation rate is constant.20 Note that the steady-state period may not hold for long before a depleted matrix concentration causes a decrease.

6.4 HETEROGENEOUS NUCLEATION In this section we will deal with the modified theory of nucleation in the presence of impurity particles in the assembly on the container wall or at grain boundaries, grain edges, grain corners, or on dislocations.

NUCLEATION IN SOLIDS

6.9

6.4.1 Nucleation at Container Wall Let us consider the formation of the b embryo in the form of a spherical cap on a flat container wall, as shown in Fig. 6.5a. Assuming the surface energy between the a/b interface gab to be isotropic (i.e., constant), the contact angle between the embryo and surface is q; the condition for static equilibrium can be given by3

gas = gbs + gab cos q

(0 £ q £ p)

(6.22)

where gas, gbs, and gab are the surface energies of the a/wall interface, b/wall interface, and a/b interface, respectively. When q, the so-called wetting or dihedral angle, lies beyond the stated limits, Eq. (6.22) cannot remain valid; in that situation, either a (no wetting) or b phase will spread over the surface (complete wetting). The free energy associated with heterogeneous formation of a b embryo may be written as DGHet = Vs (DGv + DGe ) + Aabg ab + Absg bs - Absg as

(6.23)

where Vs is the volume of the cap-shaped embryo; Aab and Abs are the areas of the a/b and b/wall interfaces; and gab, gbs, and gas are the surface energies per unit area

FIGURE 6.5 The formation of a b embryo on (a) a flat impurity surface s, (b) an a grain boundary surface, (c) a grain edge, (d) a grain corner, and (e) a dislocation. [(b–d) After Clemm and Fisher; (e) after Cahn.]

CHAPTER SIX

6.10

of the various interfaces, respectively. From the geometry shown in Fig. 6.5a, Eq. (6.23) can be expressed in terms of the contact angle q and cap radius r: 4 3 (DGv + DGe ) + 4prab2 g ab ˘˙S(q ) DGHet = ÈÍ prab Î3 ˚ where

S(q ) =

(2 + cosq )(1 - cosq )2 4

=

2 - 3 cosq + cos3 q 4

(6.24)

(6.25)

In Eq. (6.24) the term in the first set of brackets is the same as the free energy associated with the formation of a spherical embryo by homogeneous nucleation if the DGe term is not different from homogeneous nucleation. However, the second term after the brackets is a function only of the angle of contact between the container wall and embryo. Therefore, we can obtain r*ab in the same manner as we did for homogeneous nucleation: r*ab = -

2g ab DGv

(6.26)

The relationship for the radius of the critical nucleus or the radius of curvature of the spherical cap, as expressed in this equation, is the same in all cases of nucleation (including homogeneous) within a single matrix phase, differing only in the values of interfacial energy used. Consequently, the DG* for heterogeneous nucleation DG*Het, if DGe is the same, can be written as DG*Het = DGHom * S(q )

(6.27)

Comparing Eqs. (6.7) or (6.10) and (6.27), we see that the activation energy barrier for heterogeneous nucleation DG* Het is lower than DG* Hom by a shape factor S(q). For example, when q > 0, this term is positive; when q = 10°, S(q) ⬵ 10-4; when q = 30°, S(q) ⬵ 0.02; when q = p/2, S(q) = 1/2; and when q = p, S(q) = 1. When q Æ 0, DG* Het decreases to zero; that is, the b phase wets the substrate S in the presence of the a phase.

6.4.2 Nucleation on Grain Boundaries, Grain Edges, and Grain Corners The barrier to nucleation on grain boundary DG*B, is reduced by two mechanisms: (1) destruction or elimination of the portion of the planar a-a grain boundary, which is the surface separating the two matrix grains, as shown by a dashed line in Fig. 6.5b, and (2) a decrease in the critical nucleus size when compared to homogeneous nucleation.3 When we consider that the embryo formed on the grain boundary is a symmetrical doubly spherical cap, the condition for static equilibrium can be given by g aa = 2(g ab cosq )

(6.28)

where gaa is the grain boundary energy and gab is the particle-matrix boundary energy. If we ignore the strain energy term, the free energy associated with grain boundary nucleation is given by DG = VB DGv + Aabgab - Aaagaa

(6.29)

NUCLEATION IN SOLIDS

6.11

where VB is the volume of embryo, Aab is the area of the a/b interface of the energy gab created, and Aaa is the area of a-a grain boundary of energy gaa destroyed during nucleation. Alternatively, Eq. (6.29) can be written, using the geometry of Fig. 6.5b, in the form3,22 DG =

2 3 pr DG v (2 - 3 cosq + cos 3 q ) + g ab 4pr 2 (1 - cosq ) - g aa pr 2 sin 2 q 3

(6.30)

Again it can be shown that the critical radius of curvature of the nucleus (i.e., of the spherical cap) for grain boundary nucleation is given by rB* = -

4g ab DGv

(6.31)

and the critical free energy for nucleation at the grain boundary DG*B is related to that of homogeneous nucleation DG*Hom by the equation DGB* = DG*Hom (1 2)(2 - 3 cosq + cos3 q )

(6.32)

= DG*Hom 2[S(q )]

(6.33)

It should be pointed out that the ratio of DG*B/DG* Hom decreases with the increasing ratio of gaa/gab. In a similar manner, for nucleation on grain edges, where three planar boundaries meet in a line at an angle of 120° and the shape of an embryo is bounded by three spherical surfaces (Fig. 6.5c), greater reduction in the barrier energy for nucleation (than that required for grain boundary nucleation) is found. For nucleation at grain corners, where four different grains and the four different grain edges meet (Fig. 6.5d), a much greater reduction in the barrier energy to nucleation occurs and, according to Clemm and Fisher, may even cause nucleation without any nucleation barrier.23 The reason is as follows: At a grain edge or corner, the surface area of the grain boundary destroyed in comparison with the critically sized nucleus is larger than that for a planar boundary. Cahn24 has plotted the ratios of DG*B/DG*Hom, DG*E/DG*Hom, and DG* C/DG* Hom as a function of cos q (Fig. 6.6), from which it is obvious that DG*C DG*E DG*B < < DG*Hom DG*Hom DG*Hom That is, for all values of cos q, we have DG* C < DG* E < DG* B Such reduction in the barrier to nucleation does not cause a large increase in the nucleation rate, because the density of atomic sites for nucleation decreases sharply as the nucleation barrier changes in the order: homogeneous, grain boundary, grain edge, and grain corner. Cahn has approximately expressed the following equations for the number of atoms per unit volume on the various nucleation sites as NB for grain boundary, NE for grain edges, NC for grain corners: d NB = N AÊ ˆ Ë D¯

d NE = N AÊ ˆ Ë D¯

2

d NC = N A Ê ˆ Ë D¯

3

(6.34)

where NA is the number of atomic sites per unit volume, d is the grain boundary thickness, and D is the mean grain diameter. Cahn has calculated the nucleation

CHAPTER SIX

6.12 1.0

0.8

DG*Hom

DG*B,C,E

0.6

Grain boundaries

0.4

Grain edges Grain corners

0.2

0

0.25

0.5

0.75

1.0

Cos q FIGURE 6.6 The ratio of DG*B/DG* Hom, DG* E/DG* Hom, and DG* C/DG* Hom as a function of cos q (= gaa/2gab).24 (Reprinted by permission of Pergamon Press, Plc.)

rates in all these conditions and has summarized his findings in Fig. 6.7. This is the plot of [kT ln (D/d)]/DG*Hom (along the vertical axis) versus gaa/gab (along the horizontal axis), which illustrates the highest nucleation rates for homogeneous nucleation; as we proceed from (a) grain boundary to (b) grain edge to (c) grain corner nucleation, the rates decrease.

6.4.3 Nucleation at Dislocations Cottrell25 and Koehler and Seitz26 first proposed that dislocations can act as catalysts for nucleation of a new phase in solid. They suggested independently that the accommodation of the misfit strain by the strain field of dislocations would decrease the activation energy for the formation of a nucleus.27 Many electron-microscopic observations have also shown that dislocations act as excellent catalysts for precipitation processes and numerous alloy systems in the solid state,28–30 especially at small values of DGv (small undercooling) and if there is a modest elastic misfit strain energy due to the modest volume difference between the precipitate and the matrix. For small undercooling, this effectiveness of the nucleation site depends upon the magnitude of the Burgers vector, the extent of supersaturation, and the type of dislocations.31 Note that edge dislocations are more effective nucleation sites than screw dislocations. Nucleation on dislocations may occur with the incoherent, semicoherent, or coherent embryo with the matrix. In this section, we will discuss only incoherent and coherent interfaces.

NUCLEATION IN SOLIDS

6.13

[KT ln)D/d]/DG*Hom

0.25

0.20

Homogeneous

0.30

Boundary

0.5

0.10

Edge

0.05 Corner 0

0.5

1.0

1.5

gaa/gab

2.0

FIGURE 6.7 The plot of [kT ln (D/d )]/DG* Hom versus gaa/gab showing the highest nucleation rates for homogeneous nucleation and the lowest rates for grain corner nucleation.24 (Reprinted by permission of Pergamon Press, Plc.)

6.4.3.1 Incoherent Nucleation on Dislocations. When the incoherent precipitate forms on a dislocation, it causes a decrease in the length of the dislocation core, a decrease in a portion of the dislocation strain field, and a decrease in strain energy. In other words, all the strain energy of the dislocation in the volume of the dislocation core, now occupied by the new phase (nucleus), is relieved. This strain reduction of dislocation core enhances the nucleation.31–34 Cahn32 has proposed the incoherent nucleation model at dislocations by considering the nucleus as shown in Fig. 6.5e. The free-energy change produced by the formation of a small length of cylindrical nucleus of radius r around the dislocations is given by the relation DG = pr2 DGv - A ln r + 2prg + constant 2

(6.35) 2

where A is a constant equal to mb /4p (1 - n) for edge dislocation and to mb /4p for screw dislocation, m is the shear modulus, b is the Burgers vector, n is Poisson’s ratio, and g and DGv have the usual meanings. The first term is negative; the second term, being the elastic strain energy of dislocation within the radius r, is also negative; and the third term is positive.32 When both the first and second terms are not appreciable, the DG versus r curve 2 has the form A (Fig. 6.8) with parameter a[ = mb2DGv/2p2gab or mb2 DGv/2p2gab(1 n)] < 1. This parameter represents the catalytic power of dislocations. On the other hand, when the above terms are large compared to the surface-energy term, the free-energy curve has the form B (with a > 1), demonstrating the nonexistence of barrier energy for the nucleation of embryos on dislocations. In this situation, growth-controlled transformation occurs.3 Figure 6.9 illustrates the ratio of inco-

CHAPTER SIX

6.14

FIGURE 6.8 Schematic representation of the free energy of formation of a dislocation nucleus as a function of its radius.32 Curve A, a < 1; curve B, a > 1. (Reprinted by permission of Pergamon Press, Plc.)

1.0 0.9 0.8 DG*D, incoh / DG*Hom

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 a

FIGURE 6.9 The ratio of the critical free energy for nucleation on a dislocation to that for homogeneous nucleation as a function of the parameter a.32 (Reprinted by permission of Pergamon Press, Plc.)

herent nucleation energy on dislocations (DG* D,incoh) and in the matrix as a function of the parameter a. If a < 0.4, the dislocation does not act as a potent nucleation site. When a > 1, precipitation occurs simultaneously without any nucleation barrier, resulting in a bulge of radius r* over the stable cylindrical precipitate (Fig. 6.5e).28,32 Nucleation on dislocations may also be aided, to a small extent, by solute atmosphere. It has been shown that small precipitates are more stable with a coherent interface, whereas large precipitates are more stable with an incoherent interface.27,35 6.4.3.2 Coherent Nucleation on Dislocations. A coherent nucleus with the matrix does not form at the dislocation core. The strain energy and core energy persist in the presence of the formation of coherent particles as a result of continuity of the lattice planes. However, in some regions of the dislocation strain field,

NUCLEATION IN SOLIDS

6.15

nucleation might occur in order to minimize DGe.27 The critical free energy for coherent nucleation at nearby dislocation DG*D,coh can be expressed after replacing T gab in Eq. (6.7) by [gab - mb (1 + n) |e11 |/9p(1 - n)], which takes into account the effect of elastic interactions (for a spherical nucleus) with a nearby edge dislocation8,15,27,36

DG D*,coh

T 16p [g ab - mb(1 + v) e 11 ] 3 9p (1 - v) = 2 3(DGv + DGe )

(6.36)

where m is the shear modulus of the matrix, n is Poisson’s ratio, eT11 is the misfit strain of the particle, b is the Burgers vector of dislocation, and gab, DGv, and DGe have the usual meanings.

6.5 MECHANISM OF LOSS OF COHERENCY It is thus apparent once again from the above discussions that the precipitate initially nucleates coherently and becomes incoherent during growth. The mechanism of loss of coherency (in order to become an incoherent precipitate) that is summarized below also holds in precipitation sequences occurring in a wide range of precipitation-hardening alloys (Sec. 6.7): 1. Creation or punching of prismatic dislocation loops at, or close to, the particle/matrix interface due to high shear stress arising from coherency strain37,38 2. Formation and growth of small dislocation loops inside the precipitate by the collapse of clusters of vacancies or interstitials37,39 3. The accumulation of point defects from the matrix 4. The climb or glide or absorption of dislocations from an exterior source (in the matrix), or from the “grown-in” dislocation network, to the particle/matrix interface40

6.6 SPINODAL DECOMPOSITION Spinodal decomposition (i.e., continuous phase separation) has received wide theoretical attention since its current ideas were put forth by Hillert and Cahn in 1961 and 1962.41–44 Spinodal decomposition represents a departure from (or a limiting case of) the conventional theories of diffusion-controlled phase transformation. Spinodal decomposition occurs in systems that exhibit a simple miscibility gap in an otherwise single-phase region and is usually associated with the ordering reaction. Figure 6.10b is a simple hypothetical binary phase diagram with a miscibility gap. A spinodal line observed on the phase diagram represents the loci of g≤ = (∂2G/∂c2)T,P = 0 (Fig. 6.10a) at different temperatures; this line has also been described thermodynamically as a limit of stability and represents the boundary between metastable and unstable regions of the phase diagram.45,46 The region of negative curvature in the free energy versus composition diagram defines the region of spinodal decomposition, while the region lying between the points corresponding to g≤ = 0 and g¢

6.16

CHAPTER SIX

FIGURE 6.10 (a) Schematic free energy versus composition curve, at temperature T2, showing the regions of stability; (b) phase diagram illustrating the equilibrium miscibility gap (the locus of the common tangent points ¥) and the spinodal (the locus of the inflection points of the free energy versus composition curve, •).45 (Reprinted by permission of Academic Press, Orlando, Florida.)

= 0 in the free-energy versus composition curve defines normal nucleation and growth processes. In general, the heat treatment steps involved in the spinodal decomposition are (1) solution treatment of the alloy at a temperature T1 above the miscibility (or solubility) gap to produce a single-phase material; and (2) quenching to an intermediate temperature within the spinodal region (for example, T2) and holding (i.e., aging) at that temperature for a short time or continuously cooling the specimen from T1

NUCLEATION IN SOLIDS

6.17

FIGURE 6.11 Enlarged schematic free energy versus composition curves. (a) The free-energy change of an unstable phase C0; (b) freeenergy changes during decomposition of a metastable phase of composition C0.45 (Reprinted by the permission of The Metallurgical Society, Warrendale, Pennsylvania.)

to room temperature within the spinodal region.47 During this period the supercooled alloy becomes unstable because a small departure from composition C0 will lower the total free energy (Fig. 6.11a). Hence the decomposition of unstable solid solution into two-phase mixtures with essentially the same crystal structure, but with composition different from that of the parent phase, proceeds with a lowering of successive lines of Fig. 6.11a until the line corresponding to the lowest free-energy state is reached. This state is defined by the common tangent of the mixture C¢a and 48 C≤ Spinodal structures are, therefore, fine-scale, homogeneous, two-phase mixa. tures produced by phase separation under certain conditions of temperature and composition. In contrast, if an alloy of composition C0 is quenched at a small undercooling or supersaturation, that is, at a temperature T≤1 outside the spinodal region but within the solvus of the miscibility gap (where g≤ > 0), then the infinitesimal fluctuation from C0 increases the free energy and the solution becomes a metastable phase. However, if nuclei are formed with a large localized composition variation toward C≤ a , a lowering of the free energy can be effected. The reason is that for a small composition change, a positive interfacial term is more pronounced. Thus the metastable region is represented by the portion of the phase diagram where g≤ > 0, and the phase transformation proceeds by classical nucleation and growth process.

6.6.1 Nucleation and Growth Process versus Spinodal Decomposition Unlike classical nucleation and growth process that occurs in metastable solutions, the spinodal reaction is a spontaneous unmixing or diffusional clustering process in unstable solutions. The number of features that differentiate nucleation and growth process from spinodal decomposition are listed in Table 6.1. Figure 6.12a shows the composition profile associated with the classical nucleation and growth process. A composition discontinuity occurs at the distinct precipitate/matrix interface. The

6.18

CHAPTER SIX

TABLE 6.1 Factors Distinguishing Nucleation and Growth Process from Spinodal Decomposition Nucleation and Growth Processes

Spinodal Decomposition

1. Nucleation and growth occur within a metastable supersaturated solid solution during the early period.

1. Spinodal decomposition occurs within the supersaturated solid solution, which is inherently unstable to small fluctuations in compositions; therefore, the solution decomposes, spontaneously producing A-rich and B-rich regions.

2. The nucleus formed at a sufficient undercooling is a distinctly separated particle of new phase which may have the crystal structure, as well as orientation, different from the matrix.

2. The nucleus formed at a sufficient undercooling is not really a distinctly separated particle of the new phase but has the same crystal structure and orientation as the parent phase.

3. It is a large concentration fluctuation over a small volume, i.e., involving short-range composition change.

3. In the vicinity of spinodal, it is a small concentration fluctuation over a large volume, i.e., involving long-range composition change. The transformation thus occurs simultaneously throughout the matrix.

4. The classical nucleation process occurs in the region of the free energy versus composition plot where the curvature is positive, i.e., g≤ > 0 (Fig. 6.10).

4. The spinodal decomposition occurs within the region where the curvature of the free energy versus composition plot is negative, i.e., g≤ < 0 (Fig. 6.10).

5. This is associated with the establishment of a distinct precipitate-matrix interface with a positive surface-energy barrier. Consequently, a nucleation barrier exists which needs to be overcome prior to the start of transformation. This is a diffusion-controlled process.

5. The precipitate-matrix interface initially is not sharp but is diffuse in nature, without possessing a distinct structural discontinuity. Associated with this is a gradient energy which is overcome by g¢¢ if the interface is largely diffused. In this extreme situation, there is no thermodynamic barrier to spinodal decomposition. However, this reaction is governed by the activation energy of diffusion.

6. Normally there is an incubation time.

6. There is no incubation time.

7. Within the nucleation and growth region, the interdiffusion coefficient, ˜ > 0 (i.e., downhill diffusion) occurs, D and the composition fluctuations increase nonexponentially with time.

˜ < 0 (i.e., 7. Within the spinodal region, D uphill diffusion) occurs, and composition fluctuations increase exponentially with time.

composition in the precipitate nucleus rises immediately to C≤a with the reduced matrix concentration in the vicinity of the nucleated particles. With the precipitate growth, normal downhill diffusion occurs (from C0 to C¢a) in the depleted zone of the matrix toward the precipitate. Figure 6.12b shows the composition profile associated with the spinodal decomposition. The composition increase, which extends

NUCLEATION IN SOLIDS

6.19

Concentration

C"a Co C'a

Early Distance

Later

Final

(a)

C"a Co C'a

Early

Later

Final

(b)

FIGURE 6.12 Schematic evolution of composition profiles to show the difference between (a) classical nucleation and growth and (b) spinodal decomposition.48 (Reprinted by the permission of The Metallurgical Society, Warrendale, Pennsylvania.)

over a large area, is initially much less. During the precipitate growth, the solute concentration increases toward the precipitate region by uphill diffusion (i.e., diffusion up the concentration gradient).48

6.6.2 Coherent Spinodal Decomposition There are two types of spinodal decomposition, namely, chemical spinodal decomposition (as discussed in the previous section) and coherent spinodal decomposition. In either case the decomposition occurs by a continuous process; that is, there must be an existence of continuity of free energy versus composition curve from one phase to another. Chemical spinodal decomposition also occurs in fluids and noncrystalline solids. In coherent spinodal decomposition, the coherent fluctuations in solids can lead to effective elastic strains across the diffuse interface; in chemical spinodal decomposition, however, the gradient energy contribution due to diffuse interface is more pronounced. The coherent phase diagram is always a metastable phase diagram and lies within the unstressed equilibrium phase diagram because it involves a reversible metastable equilibrium, which tends to be constrained because the lattice remains continuous.45 Figure 6.13a shows the coherent and incoherent free-energy composition curves for a binary solid solution at temperature T2. It is clear from this diagram that coherent spinodal decomposition lies between the inflection points at aIV and bIV of the free energy versus composition curve. The coherent free-energy curve is higher than the incoherent equilibrium free-energy curve. Figure 6.13b shows both coherent and incoherent miscibility gaps and the respective coherent and chemical spinodals.45 The coherent critical temperature occurs at a lower temperature that is depressed by DT from the incoherent critical temperature.43

6.6.3 Nonclassical Homogeneous Nucleation Theory Cahn and Hilliard49 have defined the classical nucleation regime as one which involves the constancy of composition near the center of the nucleus so that the volume free energy and the interfacial energy can be taken separately into account. In contrast, the nonclassical regime is one in which the presence of composition variation with position throughout the critical nucleus necessitates the use of both the energies together. Cahn and Hilliard49,50 have developed a continuum nonclas-

6.20

CHAPTER SIX

FIGURE 6.13 (a) Corresponding coherent and incoherent freeenergy curves for a binary solid solution at temperature T2. The coherent spinodal is represented by the inflection points on the freeenergy curve at a IV and b IV.The coherent free-energy curve is higher than the incoherent equilibrium free-energy curve (after Hilliard). (b) Coherent and incoherent miscibility gaps and the respective coherent and chemical spinodal.45 (Reprinted by permission of the Academic Press, Orlando, Florida.)

sical theory of homogeneous nucleation which depends upon the free energy of an inhomogeneous binary solution and does not involve a change in crystal structure. The assumptions included in this theory were that both phases were incompressible fluids of constant molar volume and that the range of fluctuations within the matrix fluid is large with respect to the interatomic or intermolecular spacing. The fluid flow assumption holds with reasonable accuracy in solid Æ solid reactions provided both phases have the same crystal structure and orientation, the same lattice para-

NUCLEATION IN SOLIDS

6.21

meters, and the same isotropic elastic properties. The Cahn-Hilliard theory is applied to high temperature (i.e., transformation temperature not much below the critical temperature Tc of the equilibrium or metastable equilibrium miscibility gap) and close to the spinodal decomposition, where the interface is diffuse. Recently, LeGoues et al.51 have found that in many cases an alternative discrete latticepoint nonclassical model (i.e., sharp interface model) is more applicable at high temperatures. For the continuum Cahn-Hilliard model, we first assume an incompressible binary solution of constant molar volume. The free energy G0 of a homogeneous solid solution of composition C0 is then given by G0 = Ú g(C0 ) dV

(6.37)

v

where g(C0) is the local free energy per unit volume and V is the volume. The free energy of a solution of composition C can be given by the Cahn-Hilliard relationship as È Ê h2E ˆ 2 2˘ G = Ú Íg(C ) + K (—C ) + Á ˜ (C - C0 ) ˙ dV v Ë 1- v¯ Î ˚

(6.38)

where g(C) is the free energy per unit volume of a homogeneous solid solution of composition C; K is the gradient energy coefficient, defined by K=-

∂ 2G 1 ∂ 2G (∂— 2C ) + ∂C 2 (∂ —C )2

(6.39)

which is a positive proportionality constant (for clustering system); K(—C)2 is the positive gradient energy; and, therefore, the barrier energy to nucleation (like interfacial energy), h is the linear change in the lattice parameter per unit composition change (1/a0)(∂a/∂c) (where a0 is the lattice parameter at the average composition C of the solid solution), E is Young’s modulus, n is Poisson’s ratio, and 0 2 2 Úv[h E/(1 - n)](C - C0) dV is the total elastic strain energy of an infinite isotropic solid solution across the diffuse interface arising from coherent composition fluctuations. The total free-energy change resulting from a composition variation in an initially homogeneous solution is given by

[

2

2

]

DG = G - G0 = Ú g(C ) - g(C0 ) + K (—C ) + h 2Y (C - C0 ) dV v

(6.40)

where Y = E/(1 - n). On expanding g(C) about the average composition C0 as g(C ) @ g(C0 ) + (C - C0 )g ¢ +

1 (C - C0 )2 g ¢¢ 2

and inserting in Eq. (6.40) for a conservation system, we get 0 1 2 2 2 DG = Ú ÈÍ (C - C0 ) g ¢¢ + K (—C ) + h 2Y (C - C0 ) ˘˙ dV v Î2 ˚

(6.41)

If we further assume that the difference in composition between the initial homogeneous solution C0 and one with a composition C varies sinusoidally with distance, then we may write

CHAPTER SIX

6.22

C - C0 = A cos

2pX = A cos(bX ) l

(6.42)

where b is the wave number and l is the wavelength. Combining Eqs. (6.41) and (6.42) yields 0 1 2 DG = Ú ÈÍ A2 cos 2 (bX )g ¢¢ + K (—C ) + h 2YA2 cos 2 (bX )˘˙ dV v Î2 ˚

(6.43)

DG 1 2 = A ( g ¢¢ + 2 Kb 2 + 2h 2Y ) V 4

(6.44)

that is,

This equation is the free-energy change per unit volume between the homogeneous solid solution and the solid solution with a composition fluctuation, as shown by Eq. (6.42). It can be seen that the condition for a homogeneous solid solution to become unstable with respect to sinusoidal fluctuations of wavelength and 2p/b and thereby decompose spinodally is that DG should be negative,41,43 that is, g≤ + 2Kb2 + 2h2Y < 0

(6.45)

For coherent spinodal composition, b is negligibly small, that is, l Æ • and g≤ + 2h2Y < 0

(6.46)

2

The loci of g≤ + 2h Y = 0 at different temperatures represent the limit of stability; these loci form the line that is usually called the coherent spinodal line in the phase diagram. This lies within the chemical spinodal line (loci of g≤ = 0), as shown in Fig. 6.13b. It is clear from Eq. (6.43) or (6.44) that DG will have either a negative or a positive value depending on the magnitudes of various quantities. For example, for a particular value of g≤, when b increases from a very low value (long wavelength) to a very high value (short wavelength), DG varies from negative (an unstable solid solution) through zero to a positive value (a stable solid solution). Thus a critical value of b (called bc) exists, above which the solution is stable to composition fluctuations and below which the solution is unstable to infinitesimal composition fluctuations. Since DG = 0 at the critical value bc, we can write g≤ + 2Kbc2 + 2h2Y = 0

(6.47)

which yields Ê g ¢¢ + 2h 2Y ˆ bc = Á ˜ Ë ¯ 2K

12

(6.48)

Alternatively, the critical wavelength, lc = 2p/bc, becomes41,44 8p 2 K ˆ Ê lc = Á ˜ Ë g ¢¢ + 2h 2Y ¯

12

(6.49)

Equation (6.47) asserts that the gradient or surface-energy term varies as Kb2 and limits the decomposition on a fine scale.41

NUCLEATION IN SOLIDS

6.23

6.6.4 Diffusion Equation Let us consider the kinetics of spinodal decomposition in terms of a diffusion equation involving gradient in chemical potential. Cahn formulated an equation relating the spontaneous diffusional flux to the gradient chemical potential and demonstrated that for the flux to be spontaneous, it must lead to the lowering of free energy.41,48 Thus Cahn’s formulation of the diffusion equation for the initialstage kinetics of spinodal decomposition in modified form is48,52 ∂C M = [( g ¢¢ + 2h2Y )— 2C - 2 K— 4C + nonlinear terms] ∂t Nv

(6.50)

where M is the positive atomic mobility and Nv is the number of atoms (or molecules) per unit volume of homogeneous solution of composition C0. Let us compare Eq. (6.50) with the usual equation for Fick’s second law, namely, ∂C = D˜ (— 2C ) ∂t

(6.51)

The term (M/Nv)(g≤ + 2h2Y) in Eq. (6.50) can be identified with the interdiffusion ˜ after including the strain energy term but neglecting the higher-order coefficient D ˜ is negaand nonlinear terms. It is thus clear that, in an unstable solid solution, D tive; that is, uphill diffusion takes place. 6.6.4.1 Solution of Diffusion Equation. Assuming linear approximation and that g≤, K, h, and M are independent of composition, the general solution of Eq. (6.50) can be expressed in the form C - C0 = A(b , t ) cos (b , r )

(6.52)

where A is the amplitude of the Fourier component of the composition wave of wave number b (vector) at time t and r is the position vector. The general solution of Eq. (6.52) over all values of b and r gives 1 C(r, t ) - C0 = Ê ˆ Ë 2p ¯

3

Ú A(b , t ) exp (ibr ) db

(6.53)

b

The initial or as-quenched amplitude of the Fourier component of wave number b at t = 0 is A(b , t ) = A(b , 0) exp [R(b )t ]

(6.54)

where R(b) is the time-dependent amplitude factor and is defined by R(b ) = -

M ( g ¢¢ + 2h2Y + 2 Kb 2 )b 2 Nv

(6.55)

and R(b) is positive if g≤ + 2h2Y + 2Kb 2 £ 0

(6.56)

Plots of R(b) versus b and l are shown by solid curves in Figs. 6.14 and 6.15, respectively. The dashed line is the solution [Eqs. (6.54) and (6.55)] to the classical diffusion equation [Eq. (6.50)] after neglecting the gradient energy term, the elastic

6.24

CHAPTER SIX

FIGURE 6.14 Amplification factor R(b) versus wave number b. The dashed curve denotes the solution [Eqs. (6.54) and (6.55)] to classical diffusion equation [Eq. (6.50)] after neglecting the gradient energy term, elastic strain energy, and the nonlinear term. The solid curve represents the solution [Eqs. (6.54) and (6.55)] to Eq. (6.50) without the higher-order nonlinear b terms.52 The bm is the wave number at maximum amplification factor, and bc is the critical wave number. (Reprinted by permission of ASM International, Metals Park, Ohio.)

FIGURE 6.15 The amplification factor R(b) plotted against wavelength l (= 2p /b) instead of the wave number b for the same solutions as given in Fig. 6.14.52 (Reprinted by permission of ASM International, Metals Park, Ohio.)

strain energy term, and the nonlinear term. The solid curve represents the solution [Eqs. (6.54) and (6.55)] to Eq. (6.50) without the higher-order nonlinear b terms.

6.6.5 Early Stages of Spinodal Decomposition Within the spinodal region g≤ + 2h2Y < 0 and R(b) > 0 for all values of b, decomposition occurs spontaneously by the evolution of fluctuations with wave numbers around a specific growth rate R(bm). For low values of b (high l), the gradient energy is negligibly small and the partitioning of the atomic components is slow; however, for high values of b (short l), which is greater than bm, the gradient energy term begins to dominate. For b > bc, the decomposition cannot take place because R(b)

NUCLEATION IN SOLIDS

6.25

will decay as a result of shorter compositional fluctuations (i.e., shorter distances between clusters). For a one-phase field or a metastable portion of a two-phase field of the phase diagram (Fig. 6.13b), we have g≤ + 2h2Y > 0. In this case, R(b) < 0 and Eq. (6.54) predicts that any existing composition fluctuations will decay out. For a particular temperature and composition within the spinodal region, there exists a critical wave number bc (= 2p/lc) that satisfies g≤ + 2h2Y + 2Kbc2 = 0

(6.47)

for which R(b) becomes zero. It follows from Eqs. (6.46) and (6.47) that coherent spinodal represents the locus of bc = 0. We can thus rewrite Eq. (6.55) in terms of bc by using Eq. (6.47): R(b ) =

2 KM 2 (bc - b 2 )b 2 Nv

(6.57)

This function is shown by the solid curve in Fig. 6.14. The maximum growth rate is found at bm =

bc 2

(6.58)

After equating the derivative of Eq. (6.57) to zero, bm is called the spinodal wave number. Combining Eqs. (6.57) and (6.58), we get the maximum value of R(b)52 2 KM 4 bm Nv KM 4 bc = 2Nv

R(b m ) =

(6.59) (6.60)

Thus R(bm) is strongly dependent on bm.

6.6.6 Experimental Study of Spinodal Decomposition The study of the kinetics of spinodal reactions is made by using small-angle Xray and neutron scattering by monitoring the changes in the intensity distribution of the composition fluctuation around the direct beam. The lattice imaging technique by transmission electron microscopy has also been used to characterize the modulated and tweed structures occurring during the initial period of decomposition.4 The isothermal kinetics of an early stage of spinodal decomposition can be determined from a quenched and aged specimen of a spinodal alloy (e.g., Al-22 at% Zn0.1 at% Mg alloy specimen aged at 125°C) by a plot of small-angle X-ray spectra intensity versus the scattering angle 2q, where b = 2p/l ⬵ deg 2q/14 (Fig. 6.16).53 This angle is a characteristic of the wavelength of composition modulations corresponding to the peak position lm. Note that there is a steady change in spectral shape from a broad peak at low intensity for quenched specimen to a very sharp peak at much greater intensity for specimen aged for 80 min. The small-angle region representing small b (high l) is growing in intensity, while the large-angle region is shrinking. This is predicted by assuming the occurrence of a critical wave number bc.

6.26

CHAPTER SIX

FIGURE 6.16 SAS spectra from an Al-22 at% Zn-0.1 at% Mg alloy specimen aged at 125°C using the correction for the effect of beam height. The radiation employed was CuKa with a characteristic wavelength of 1.541 Å.53 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

FIGURE 6.17 The [001] electron diffraction pattern from spinodally decomposed Cu-4 wt% Ti alloy aged at 400°C (750°F) for 100 min, illustrating satellites flanking the matrix reflection.4 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

Figure 6.17 is a [001] electron diffraction pattern from a spinodally decomposed Cu-4 wt% Ti alloy showing the satellite configuration flanking the matrix reflection.4 The appearance of “side bands” or satellite spots in the electron diffraction pattern is a characteristic feature of spinodal decomposition.54

NUCLEATION IN SOLIDS

6.27

If the strain energy term in the free-energy expression is very small (small misfit) or if the material (e.g., Fe-Cr-Co permanent magnetic alloy) is elastically isotropic, the microstructure developed will be isotropic (Fig. 6.18).55 The two-phase mixture is interconnected in three dimensions and does not exhibit any directionality.55 However, aligned modulated structure is developed; that is, preferential growth or a dominant concentration wave occurs in anisotropic materials (cubic crystals), for example, a Cu-Ni-Fe alloy, along the elastically soft matrix directions (Fig. 6.19),55 where minimum strain energy exists.52 In several Cu-Ni-Fe spinodal alloys, a coherency loss (of modulated structures) has been found when l > 800 Å (80 nm).56 The tweed structure also appears to form in alloys where coherent tetragonal precipitates become embedded in a cubic matrix, being aligned along the direction.11

6.6.7 Spinodal Alloys Spinodal decomposition has been observed in a wide variety of metallic alloy systems, for example, Al-base alloys such as Al-Li, Al-Cu, Al-Zn, and Al-22 at% Zn-0.1 at% Mg; Cu-base alloys such as Cu-4 to 5 at% Ti, Cu-Zn, Cu-Co, Cu-NiFe, Cu-Ni-Si, Cu-Ni-Sn, and Cu-Ni-Cr; Fe-base alloys such as Fe-Al, Fe-Be, Fe-Mo, Fe-Cr, Fe-Ni-Al, Fe-Ni-C, Fe-Cr-Co, and Fe-Al-Cr-Ni; Mn-base alloys such as

FIGURE 6.18 Transmission electron micrograph of isotropic spinodal structure developed in permanent magnetic Fe-28.5Cr10.6Co(wt%) alloy aged at 600°C (1100°F) for 4 hr.55 (Reprinted by permission of ASM International, Metals Park, Ohio; after A. Zettser.)

6.28

CHAPTER SIX

FIGURE 6.19 Transmission electron micrograph showing spinodal microstructure in a 51.5Cu-33.5Ni-15Fe (at%) alloy solution treated at 1050°C (1922°F) for 2 hr and aged at 775°C (1427°F) for 15 min. The dark regions are Ni-rich, and the light regions are Cu-rich. Foil normal is ⬃[001], and the alignment along elastically soft (100) matrix direction is apparent. The wavelength of the modulated structure is ⬃25 nm (250 Å). (Courtesy of G. Thomas.)

Mn-Cu; Ni-base alloys such as Ni-Al, Ni-12 wt% Ti, Ni-Cr, Ni-V, Ni-Mo, Ni-Be, and Ni-Si; Nb-base alloy such as Nb-Zr; Ti-base alloys such as b Ti-Cr; and Au-base alloys such as Au-77 at% Ni and Au-Pt alloys.4,57,58 Many of the splat-cooled alloys are also expected to exhibit spinodal decomposition, and in many systems, for example, Al-4 wt% Cu alloy below 130°C, Cu-15Ni-8Sn (wt%), and Cu-10Ni-6Sn (wt%) alloys, modulated structures precede the formation of GP zones or transition phases.

6.6.8 Applications This mode of transformation is useful in improving the physical and mechanical properties of commercial alloys. For example, spinodal structure favors high coercivity in the production of permanent magnetic materials. This structure can be optimized by thermomechanical treatment, step aging, and magnetic aging. This structure is important in high-cobalt Alnico, Cu-Ni-Fe, and Fe-Cr-Co alloys.55 Spinodal structure in Cu-Ni-Fe appears to possess good mechanical stability under fatigue conditions because sheared precipitates would withstand very severe distortion prior to the development of an effective wavelength.59

6.7 PRECIPITATION HARDENING In Secs. 6.1 through 6.4, we have described homogeneous nucleation and heterogeneous nucleation on various nucleation sites. This knowledge is the foundation for understanding diffusional transformations. For example, let us begin our discussion with the precipitation-hardenable alloys (also called age-hardenable alloys). If an alloy system exhibits (1) moderate or extended solubility that decreases with decreasing temperature, for example, Al-rich region of Al-Cu system (Fig. 6.20),60 and (2) the formation of one or several metastable transition phases prior to, or in addition to, the equilibrium phase, it fulfills the conditions necessary for precipitation hardening. These alloys are called precipitation-hardening alloys because their hardness or yield strength increases with time at a constant temperature

Atomic percent copper 1100

0

10

20

30

40

50

60

70 80 90 100 1064.87 b0

1000 g0

L 900

b

6.29

Temperature °C

e1

800 700

g1

660.452°C e2

600

h1

548.2°C

500

d

567°C

z1 h2

400 300

(AI)

q

(Cu) a2

z2

0 Al

10

20

30

40 50 60 Weight percent copper

70

80

90

100 Cu

Phase

Composition, wt% Cu

Pearson symbol

Space group

(Al) q h1 h2 z1 z2 e1 e2 d g0 g1 b0 b a2 (Cu)

0 to 5.65 52.5 to 53.7 70.0 to 72.2 70.0 to 72.1 74.4 to 77.8 74.4 to 75.2 77.5 to 79.4 72.2 to 78.7 77.4 to 78.3 77.8 to 84 79.7 to 84 83.1 to 84.7 85.0 to 91.5 88.5 to 89 90.6 to 100

cF4 tI12 oP16 or oC16 mC20 hP42 (a) (b) hP4 (c) (d) cP52 (d) cI2 (e) cF4

Fm3¯ m I4/mcm Pban or Cmmm C2/m P6/mmm ... ... P63/mmc R3¯ m ... P4¯ 3m ... Im3¯ m ... Fm3¯ m

... ... 61 to 70

tP6 cF16 hP5

... Fm3¯ m P3¯ m1

Metastable phases q¢ b¢ Al3Cu2

(a) Monoclinic? (b) Cubic? (c) Rhombohedral. (d) Unknown. (e) D022-type long-period superlattice

FIGURE 6.20 Al-Cu phase diagram with description of phases.60,60a (Reprinted by permission of ASM International, Materials Park, Ohio.)

CHAPTER SIX

6.30

after quenching from a solution treatment temperature. These sequences of precipitate nucleation are of considerable interest both in the understanding of the mechanisms of phase transformations and in the controlled evolution of the mechanical properties of precipitation-hardening alloys.61 The use of precipitation hardening as a strengthening technique is very important in certain alloy systems, notably Al-based alloys, Cu-Be, steels (including maraging steels), oxide systems (such as Ti-rare earth oxides), and amorphous alloys.62 The precipitation nucleation sequences occurring in Al-, Cu-, Fe-, and Ni-based precipitation-hardening alloys have been given in Table 6.2.63–65 Table 6.3 lists the mechanical properties and applications of some Al-, Cu-, and Ni-based precipitation-hardening alloys,66,67 while Table 6.4 enlists only the mechanical properties of important precipitationhardening stainless steels.67–69 Precipitation strengthening also includes dispersion strengthening, which involves the most commonly known mechanism of overaging, namely, the Orowan mechanism. For dispersion strengthening, the dispersoid particles are introduced by natural or artificial aging as well as by other methods such as powder metallurgy and mechanical alloying. Precipitation strengthening used widely in ferrous alloys is recognized in the quench aging of low-carbon steels, in microalloyed steels, and in tempering of martensite.70 Readers will find some interesting discussions on these topics elsewhere in this book.

6.7.1 General Characteristics of the Precipitation-Hardening Process 1. The hardness (and tensile strength) measured at room temperature increases up to a maximum (called peak hardness) in the hardness versus aging-time curve (Fig. 6.21).71,72 This peak hardness occurs at an optimum size distribution and degree of coherency. . 2. As the aging temperature is increased (that is, N and G increase rapidly), the peak hardness value decreases but is readily attained. 3. A single-stage (peak) hardening takes place at higher aging temperatures for high supersaturation or low aging temperatures for lower solute content (or supersaturation). In contrast, a two-stage hardening occurs at lower aging temperatures for higher supersaturation exhibiting high peak-hardness level69 (Fig. 6.21). TABLE 6.2 Precipitation Sequence Observed in Common Precipitation-Hardening Systems Base metal Al

Alloy Al-Cu (containing up to 1.7–4.5 wt% Cu)

Al-Ag (containing up to~23 at% Ag) Al-Mg Al-Zn

Precipitation sequence Spinodal decomposition Æ GP1 zones rich in Cu (plates, fcc) on {100}a Æ q ≤ phase (plates, tetragonal) on {100}a Æ q ¢ phase (plates, tetragonal) GP zones rich in Ag (spheres, fcc) on {111}m Æ g ¢ (Ag2Al, plates, hcp) Ordered GP zones (plates or rods, fcc) Æ b ¢ (plates, fcc) Spinodal decomposition Æ GP zones (spheres, fcc) Æ a ¢ (plates, rhombohedral) Æ a ¢ (cubic)

Equilibrium precipitate

q (CuAl2, bct)

g (Ag2Al, hcp) b (Mg3Al2) Zn

TABLE 6.2 Precipitation Sequence Observed in Common Precipitation-Hardening Systems (Continued) Base metal

Alloy

Precipitation sequence

Al-Cu-Mg (2xxx)

GPB zones rich in Mg and Cu (rods) on {110}m Æ S¢ (plates) on (021) planes Al-Mg-Si (6xxx) GP zones rich in Mg and Si (needles, fcc) along {100}a Æ b ¢¢ ordered zones rich in Mg and Si Æ b ¢ phase (rodlike, hcp), Mg2Si Al-Zn-Mg(-Cu) (7xxx) GP zones rich in Zn and Mg (spheres) Æ h¢ (MgZn2, plates, hcp) on {111}a Æ h (MgZn2, rods or plates, hcp) along {111}a Al-Li d ¢ (Al3Li) Al-Li-Zr d ¢ + a ¢ (Al3Zr) Al-Li-Mg d ¢ (Al3Li)

T [(Al,Zn)48Mg32, cubic]

d ¢ + S¢(Al2CuMg) + T¢1 (Al2CuLi) + a ¢

T1 (Al2CuLi) + d + a ¢ (Al3Zr)

Cu-2Be

GP zones Æ g ≤ Æ g ¢ (bct) first forms with a {112}a habit plane and later with a {113)a

g (CuBe, ordered, bcc)

Cu-1–3 at% Co Cu-2.5% max Ti

GP zones (spheres) b ¢ phase (Cu4Ti, D1a structure)

b (Co, plates) —

Fe-C

e carbide (Fe2.4C disks, hcp)

Fe-N PH stainless steels 17-4 PH

a ¢¢ (disks, Fe8N, bct) on (100)m

Fe3C (plates, orthorhombic) g ¢ (Fe4N, fcc)

Al-Li-Cu-Mg (high Mg/Cu ratio) Al-Li-Cu-Mg-Zr

Ni

b (Mg2Si, plates)

GP zones Æ S¢ (Al2CuMg)

Al-Li-Cu

Fe

S (Al2CuMg, orthorhombic)

d (AlLi) d + a¢ Al2MgLi, bcc, rod like q(CuAl2) Æ T1 (Al2CuLi) T1 (Al2CuLi, thin plates, hcp) S (Al2CuMg)

Al-Cu-Li

Cu

Equilibrium precipitate

GP1 zones Æ GP2 zones Æ d ¢, q ¢, and/or T¢1 T¢1 (Al2CuLi, thin plates)

17-7 PH

e phase, Cu-rich clusters (spheres, bcc) NiAl (sphere, fcc)

A286 17-10 P alloy

g ¢ (Ni3TiAl) M23(CP)6 or (CrFeP)23C6

Incoloy 901, 902

Ordered g ¢ [Ni3(Al,Ti), fcc, L12 structure, cuboidal] Ordered g ¢¢ (Ni3Nb, bct, DO22 structure, disks) + g ¢ [Ni3(Al,Ti), fcc, in small amounts] Ordered g ¢ Ni3(Al,Ti)

Haynes 242

Ordered Ni2(Mo,Cr)

Inconel X-750 Inconel 718

6.31

e (fcc) b (NiAl, laths and plates) h (Ni3Ti) h [Ni3(AlTiNb) plate,hexagonal] d Ni3Nb (orthorhombic) Ni3Ti, hcp, DO24 structure

TABLE 6.3 Typical Mechanical Properties and Applications of Some Precipitation Hardening Al-, Cu-, and Ni-Based Alloys66,67

Base metal

Alloy number (UNS #)

Chemical composition, wt%

Tensile strength Temper

ksi

MPa

Yield strength ksi

MPa

Elongation, (%)

Typical application

Wrought alloys Al

6.32 Cu

Ni

2024 (92024) 3004 (93004)

Al-4.4Cu-1.5Mg -0.6Mn Al-1.2Mn-1.0Mg

6061 (96061) 7075 (97075) 7178 (97178)

Al-1.0Mg-0.6Si -1.3Cu-0.2Cr Al-5.6Zn-2.5Mg1.6Cu-0.23Cr Al-6.8Zn-0.3Mn -2.7Mg-2.0Cr

C17200

T86

64

440

57

395

6

H38

41

—

36

250

4–6

T6

45

310

40

276

12

T6

83

572

73

503

11

Aircraft structures, rivets, truckwheel hardware, screw machine products Rigid containers (cans), chemical handling and storage, sheet-metal works, builder’s hardware, incandescent and fluorescent lamp bases Trucks, towers, marine structures, canoes, railroad cars, pipelines, furniture Aircraft and other structural parts

T6, T65

88

605

78

540

10

Aircraft and aerospace structures

98Cu-2Be -0.6Mn

175

1207

140

965

7

C17000

98Cu-1.7Be -0.3Co

180

1240

155

1070

4

Monel K-500 (N05500)

66Ni-30Cu -2.7Al-0.6Ti

151

1041

111

765

30

Bourdon tubing, flexible metal hose, bellows, clips, washers, retaining rings, firing pins, springs, flexible metal hose, bushings, valves, pumps, shafts, diaphragms, contact bridges, bolts, screws, navigational instruments, nonsparking safety tools Bellows, bourdon tubing, fuse clips, fasteners, lock washers, retaining rings, switch and relay parts, electric and electronic parts, roll pins, springs, spine shafts, rolling mill parts, welding equipment, valves, pumps, diaphragms, nonsparking safety tools Corrosion-resistant and high-strength parts such as pump shafts, oil well drill collars, tools and instruments, doctor blades and scrapers, impellers,

valves, gears, springs, valve trims, fasteners, marine propeller shaft, gyroscope applications Nimonic 80 (N07080) Nimonic 90 (N07090) Udimet 500 (N07500) Inconel 718 (N07718)

6.33

Inconel 718 SP (N07719) Inconel X-725 (N07725) Inconel X-750 (N07750) Haynes 240

Incoloy 925 (N09925)

Nimonic 105

Ni-1.4Al-0.008B*0.10C*-2.0Co*19.5Cr-0.2Cu*-3.0Fe* -1.0Si*-2.25Ti Ni-1.4Al-0.13C*18.0Co-19.5Cr-3.0Fe*1.0Mn*-1.5Si*-2.4Ti 53Ni-18.5Co-18Cr-4Mo3Ti-3Al 52.5Ni-19Cr-3Mo5.1Nb-0.9Ti-0.5Al18.5Fe-0.15Cu*-0.07C* Same as above with 0.05%C* 57.0Ni-20.75Cr8.25Mo-3.4Nb1.4Ti-0.35Al0.35Mn*-0.03C* 73.0Ni-15.5Cr-3Mo -1.0Nb-2.5Ti-0.7Al7Fe-0.25Cu*-0.04C* 65Ni-25Mo-8Cr-2.5Co*2Fe*-0.8Mn*-0.8Si*0.5Al*-0.5Cu*-0.03C*0.006B* 42Ni-0.3Al-0.03C*0.5Nb*-21.5Cr2.25Cu-22.0Fe min1.0Mn*-3.0Mo0.50Si*-2.15Ti Ni-0.12C*-1.0Si* -1.0Mn*-15Cr-20Co5Mo-1Ti-4.7Al-0.2Cd* -1.0Fe*-0.0065B -0.15Zr*

180

1240

150

1036

22

See Sec. 6.7.4

162

1120

92

635

24

See Sec. 6.7.4 Aerospace and chemical processing industries, seal and container rings, duct segments, casings, fasteners, rocket nozzles, pumps, etc.

TABLE 6.3 Typical Mechanical Properties and Applications of Some Precipitation Hardening Al-, Cu-, and Ni-Based Alloys66,67 (Continued)

Base metal

Alloy number (UNS #)

Chemical composition, wt%

Tensile strength Temper

ksi

MPa

Yield strength ksi

MPa

Elongation, (%)

Typical application

Cast alloys Al

6.34 Cu

Ni

296.0 (A02960)

Al-4.5Cu-2.5Si

T6

40

275

26

179

5

355.0 (A13550)

Al-5Si-1.3Cu-0.5Mg

T61

35

241

25

172

3

356.0 (A13560)

Al-7.0Si-0.3Mg

T61

38

260

28

195

5

712.0 (A17120) C82400

Al-5.8Zn-0.6Mg0.5Cr-0.2Ti Cu-1.7Be-0.3Co

T5

35

241

25

172

5

150

1034

140

965

1

C95400

Cu-4Fe-11Al

105

725

54

373

8

Monel 505 (N05505) Inconel 705 (N09705)

63Ni-29Cu-4Si

127

876

97

669

3

Safety tools, cams, bushings, molds for forming plastics, pump parts, valves, bearings, gears, parts for submarine telephone cable repeater system and hydrophone, and plunger tips for die casting Bearings, bushings, gears, worms, valve seats and guides, pickling hooks, pump impellers, rolling mill slippers, slides, and nonsparking hardware Valve seats

68Ni-15Cr-9Fe-6Si

110

758

95

665

3

Exhaust manifolds

* Indicates maximum value.

Aircraft fittings, fuel pump bodies, wheels and gun control parts, railroad car seat frames, connecting rods, compressors, fuel pump bodies Aircraft supercharger covers, pump housings, aircraft fittings and engine crankcases, air compressor pistons, fuel pump bodies, liquid-cooled cylinder heads, water jackets, and blower housings Aircraft pump and control parts, automotive transmission cases, aircraft fittings, water-cooled cylinder blocks, and wheels Machine parts

TABLE 6.4 Minimum Mechanical Properties of Some Precipitation-Hardening Stainless Steels67–69 AISI (UNS) designation (ASTM specification)

Composition

17-4PH

630 (S17400) (A564, A693, A705)

0.09C-1.0Mn-1.0Si17.0Cr-4.0Ni-3.6Cu0.25(Nb + Ta)

15-5PH

(S15500) (A564, A705)

0.07C-15Cr-1Mn4.5Ni-3.5Cu0.35(Nb + Ta)

(S13800) (A564, A693, A705)

0.05C-0.02Mn-0.1Si12.75Cr-8Ni-2.25Mo0.01Nb

Custom 450

(S45000) (A564, A705)

Custom 455

(S45500) (A564, A705)

Trade name

Product forma

Condition

Hardnessc Rc

UTS, MPa (ksi)

YS, MPa (ksi)

Elong. (%)

1310 (190) 1070 (155) 1000 (145) 930 (105)

1170 (170) 1000 (145) 860 (125) 725 (105)

10d 12d 13d 16d

40–35e 45e 45e 50e

40 35d 32d 28d

48 42d 38d 36d

6–10c,f 6–10c

45–35f —

45 43

— —

6–10f 8–12 15

20–40f 45 50

40 43 28

— — —

Reduction in areab (%)

Min Max

Martensitic

PH13-8Mo

6.35

13-8 SuperTough

B,F,P,Sh,St B,F,P,Sh,St B,F,P,Sh,St B,F,P,Sh,St

(Similar to 17-4PH) 1520 (220) 1380 (200)

1410 (205) 1310 (190)

B,F,P,Sh,St P,Sh,St

H950 H1000 H1000

1406 (204)

0.05C-1Si-15.5Cr6Ni-0.75Mo-1.5Cu(8 ¥ C)Nb

B,F,P,Sh,St B,F,P,Sh,St B,F,P,Sh,St

H900 H1000 H1150

1240 (180)g 1170 (170) 1100 (160) 1030 (150) 860 (125) 515 (75)

0.05C-0.5Mn-0.5Si12Cr-8.5Ni-1.1Ti0.3(Nb + Ta)-2Cu-0.5Mo

B,F, shapes H900h B,P,Sh,St H950 B,P,Sh,St H1000

1620 (235) 1586 (230) 1483 (215)

1520 (220) 1517 (220) 1345 (195)

8 12 14

30 50 55

47 48 44

— — —

H900 H1000 H1025 H1050

1758 (256) 1593 (231) 1565 (227) 1517 (220)

1648 (239) 1503 (218) 1475 (214) 1413 (205)

14 16 15 17

62 66 67 66

49 47 47 46

— — — —

Fe-13Cr-8Ni-2Mo-1Al

Custom 465

H900 H1025 H1075 H1150

0.02C-0.25Mn-0.25Si12.0Cr-11.0Ni-1.65Ti1.0Mo

Semiaustenitic 17-7PH

631 (S17700) (A693)

0.07C-1.0Mn-1.0Si-17Cr7Ni-1.15Al

B,Sh,St B,Sh,St B,Sh,St

CH900 RH950 TH1050

1650 (240) 1450 (210) 1240 (180)

1590 (230) 1310 (190) 1030 (150)

1 1–6d 3–7d

— — —

46 41d 38

— 44d —

PH15-7Mo

632 (S15700) (A693)

0.07C-1Mn-1Si-15Cr7Ni-1.15Al

B,Sh,St B,Sh,St B,Sh,St

RH950 TH1050 Cold-rolled and aged

1550 (225) 1310 (190) 1650 (240)

1380 (200) 1170 (170) 1590 (230)

1–5d 2–5d 1

— — —

43d 38d 46

46d 46d —

TABLE 6.4 Minimum Mechanical Properties of Some Precipitation-Hardening Stainless Steels67–69 (Continued)

Trade name

AISI (UNS) designation (ASTM specification)

Stainless W

635 (S17600)

AM 350

S35000 (A693)

AM 355

S35500 (A693, A705)

Composition 0.08C-1Mn-1Si16-17.5Cr-6.0-7.5Ni0.4Al-0.4-1.2Ti 0.07C-0.5Mn-0.5Si-1617Cr-4-5Ni-2.5-3.25Mo -0.07-0.13N 0.01C-0.5Mn-0.5Si15-16Cr-4-5Ni-2.53.25Mo-0.07-0.13N

Product forma

Condition

UTS, MPa (ksi)

YS, MPa (ksi)

Elong. (%)

Reduction in areab (%)

Hardnessc Rc Min Max

B,P,Sh,St B,P,Sh,St B,P,Sh,St P,Sh,St P,Sh,St

H950 H1000 H1050 H850 H1000

1310 (190) 1240 (180) 1170 (170) 1275 (185) 1140 (165)

1170 (170) 1110 (160) 1070 (150) 1030 (150) 1000 (145)

8 8 10 2–8d 2–8d

25 30 40 — —

39 37 35 42 36

— — — — —

F P,Sh,St

H1000 H850

1170 (170) 1310 (190)

1070 (155) 1140 (165)

12 10

25 —

37 —

— —

615i (89)i 945i (137)i 593 (86)

255i (37)i 605i (88)i 290 (42)

70i 25i 45

76i 39i 63

82i,j 30f,i

— —

641 (93)

248 (36)

48

70

75j

—

1000 (145)

690 (100)

24

37

29

—

Austenitic 17-10P

6.36

17-14Cu-Mo

A286

JBK-75

a

600 (S66286)

0.5C-1Mn-1Si-17Cr10.5Ni-0.28P 0.12C-1Mn-16Cr-14Ni2.5Mo-0.4Mo-0.3Ti-3Cu

B B

0.08C-1.4Mn-0.4Si15Cr-26Ni-1.3Mo-0.3V2Ti-0.35Al-0.003B

P,Sh,St P,Sh,St

Annealed Aged Solution-annealed 1230°C, 0.5 hr, WQ, and aged 730°C, 5 hr Annealed 980°C (1800°F) Aged 720°C (1330°F)

0.015C-0.005Mn-0.02Si14.5Cr-29.5Ni-1.25Mo2.15Ti-0.25Al-0.21V0.0015B

B, bar; F, forging; P, plate: Sh, sheet; St, strip. Values are for bar products. Where minimum value is also given, maximum value applies only to flat-rolled products. Both max and min values may vary with thickness for flat-rolled products. d Value varies with thickness or diameter. e Value is generally lower for flat-rolled products and varies with thickness. f Higher value is longitudinal; lower value is transverse. g Tensile strength only applicable up to sizes of 13 mm (1/2 in.). h Up to and including 150 mm (6 in.). i Values are typical. j Rockwell B hardness. Reprinted by permission of ASM International, Materials Park, Ohio. b c

NUCLEATION IN SOLIDS

6.37

FIGURE 6.21 The hardness versus aging-time curves for Al-Cu alloy at (a) 130°C and (b) 190°C.71,72

4. The precipitation occurs by a nucleation and growth process. The precipitate particles that are nucleated during the early aging treatment are coherent with the matrix and are metastable phases that can form more rapidly than the equilibrium phase. 5. When aging is allowed to proceed beyond the peak hardness, the hardness decreases in a monotonic manner into the overaged condition. This is attributed to the loss of coherency of the precipitate particles. Double (or two-step) aging treatment is sometimes used to produce initially a fine distribution of GP1 zones at a temperature below the zone solvus, followed by zone coarsening at an elevated temperature. Such treatments are sometimes necessary to reduce precipitate-free zones (PFZs). However, the choice between double-aging treatment and alloying in order to optimize the structures and properties should be exercised based on the economic factors.70

6.7.2 Al-Base Alloys 6.7.2.1 Al-Cu Alloys. Let us focus our attention now on a typical Al-4.0 wt% Cu alloy. The basic steps involved in precipitation-hardening heat treatment are as follows:

6.38

CHAPTER SIX

1. Solution heat treatment (solutionizing) at a temperature of about 550°C (i.e., near the eutectic temperature and between the solvus and solidus temperatures) to produce a uniform and homogeneous single-phase solid solution. 2. Quenching in water to room temperature to retain an unstable, supersaturated solid solution (with copper) by suppression of the precipitation of equilibrium q phase. This is also effective in retaining most of the concentration of vacancies present in thermal equilibrium at the solutionizing temperature.73 Note that enhanced vacancy concentration formed by quenching from the solutionizing temperature gives rise to an increased rate of solute diffusion, thereby an enhanced rate of clustering. The vacancies may finally condense to form dislocation loops, which tend to act as nucleation sites for precipitation or may exit to various sinks such as grain boundaries. The loss of vacancies at these sinks results in precipitate-free zones. 3. Aging by holding for a sufficient length of time at room temperature (called natural aging) or at an intermediate temperature below about 190°C (called artificial aging) to precipitate a finely dispersed second phase within the matrix. In general, the (artificial) aging temperature for any precipitation-hardening alloy lies between ⬃15 and 25% of the temperature difference between the solutionizing temperature and room temperature.74 Metastable and Equilibrium Phases. The age-hardening sequence in Al-4% Cu alloy below 130°C is described as a s Æ spinodal decomposition Æ GP1 zones Æ GP2 zones (q ¢¢) Æ q ¢ Æ q The prefix GP is used to indicate the initials of the discoverers, Guinier75 and Preston,76 of these zones. Age-hardening of this alloy, however, at 130°C has been found to have the following precipitation sequence: a s Æ GP1 zones Æ GP2 zones (q ¢¢) Æ q ¢ phase Æ q phase (CuAl 2 ) On the other hand, there is a direct precipitation of equilibrium phase at high temperatures, which is represented by

as Æ q spinodal decomposition. There is strong evidence that the formation of GP1 zones below 130°C initially proceeds by a spinodal decomposition. Rioja and Laughlin77 made an electron diffraction study of an Al-4 wt% Cu alloy aged at room temperature for 5 hr and showed that the initial decomposition product was the modulated microstructure comprising the periodic and aligned two-phase mixture (i.e., solute-rich and solute-poor regions) which was attributed to spinodal decomposition. This is followed by GP1 zones arranged in a semiperiodic array. Hence they proposed that zone formation occurred continuously from the spinodal decomposition. Other workers have suggested that there is no nucleation barrier for a spinodal mechanism for GP1 zone formation or for direct GP1 zone formation.28 A kinetic study of precipitation also supports a spinodal mechanism. In other important age-hardening alloys (e.g., Inconel 80 and the Nimonics), spinodal decomposition occurs initially. gp1 zones. The existence of GP1 zones is detected as streaks in the X-ray or electron diffraction patterns arising from the fact that they are very thin; they are characterized by compositional segregation together with the strain field induced by coherent zones (Fig. 6.22).8 GP1 zones are formed directly from the solid solution as the first precipitate on aging at 130°C and consist of circular disk- or platelike clusters of Cu atoms on the

NUCLEATION IN SOLIDS

6.39

FIGURE 6.22 Electron micrograph of an aged Al-4% Cu alloy. The fine equiaxed particles are either GP1 zones or q≤; the elongated particles are q ¢, which have formed on dislocations.8 (Reprinted by permission of North-Holland Physics Publishing, Amsterdam; after G. W. Lorimer.)

{100} planes of the matrix. These disks have fcc crystal structure (Fig. 6.23)28,78,79 and are about 2 atoms (0.4 to 0.6 nm) thick and 30 atoms (⬃10 nm) in diameter and are usually coherent with the matrix because the atoms just replace the Al atoms in the lattice. Figure 6.24 is Gerold’s model of a GP1 zone in Al-Cu.79 It shows the extension of the associated displacement field of the surrounding Al-rich matrix to approximately 15 layers on each side of the single Cu-rich plane.28 This model is in good agreement with the investigation of (1) Al-17 at% Cu alloys aged at room temperature for 12 years and examined by diffuse X-ray scattering80 and (2) Al-1.5 wt% and Al-4.0 wt% Cu alloys aged at 80, 100, and 130°C, studied by weak beam electron microscopy.81 However, later results have confirmed that GP1 zones are really a mixture of single-layer and multilayer pure copper regions on {100}a planes.82 The compositional range of GP1 zones varies between ⬃100 at% Cu (according to Gerold’s model) and 30 to 40 at% Cu. The formation of GP1 zones is a homogeneous nucleation and growth process because this establishes the three conditions: random formation without any perceptible incubation time and in very high number densities (up to 1018/cm3), which is several orders of magnitude larger than the density of available heterogeneous nucleation sites. However, excess vacancies appear to play a vital role in their formation. On the other hand, q ≤, q ¢, and q phases usually form by a heterogeneous nucleation process. Lorimer and Nicholson83 have indicated that GP1 zones act as nucleating sites for other metastable precipitates. gp2 zones. (q ≤ phase). On aging for a longer time or at higher temperatures (130 to 170°C), the diffraction streaks are resolved into a well-defined intensity maxima. The precipitate, formed homogeneously82 in this stage, is a disk-shaped q≤ phase; the precipitate size is larger than that of GP1 zones, being 1 to 4 nm (10 to 40 Å) thick and 10 to 100 nm (100 to 1000 Å) in diameter. The crystal structure is

6.40 FIGURE 6.23 The crystal structures of the precipitates formed during aging in Al-4% Cu alloy. The matrix phase is a; GP1 zones, q ≤, and q ¢ are transition phases of increasing stability; and the stable precipitate is q.1,28 (Reprinted by permission of North-Holland Physics Publishing, Amsterdam; after Hornbogen78 and Gerold.79)

NUCLEATION IN SOLIDS

6.41

FIGURE 6.24 Gerold’s model of a GP1 zone in AlCu.79 (Reprinted by permission of Van Nostrand Reinhold International.)

tetragonal (Fig. 6.23) with a = 4.04 Å and c = 7.68 Å, not due to ordering but due to coherency constraints with the matrix. They are fully coherent platelike precipitates with a {001}a habit plane. The orientation relationship with the matrix is (001)q ≤ // (001)a; [100]q≤ // [100]a. GP2 zones are the main contributors to the strength of AlCu alloys, which is nearly double when compared to the solid solution itself (Fig. 6.21). Like the GP1 zones, q≤ phases are visible as a result of coherency strain fields caused by misfit normal to the plates (Fig. 6.22). The composition of q≤ was deduced to be ⬃17 at% Cu by comparing the q ≤ lattice parameter with that of splatquenched metastable fcc Al alloy samples containing 17.3 at% Cu. q ¢ phase. After prolonged aging times, the q ≤ phase yields an elongated diskshaped q ¢ phase with a thickness of 10 to 150 nm (100 to 1500 Å). The q ¢ phase has a more ordered tetragonal structure than the q ≤ (Fig. 6.23), with a = 4.04 Å and c = 5.8 Å. The approximate composition of q ¢ is CuAl2, and it obeys the same orientation relationship with the matrix as q ≤, while edges of the plate may be either semicoherent or incoherent.84 Hence, it is appropriate to say that q ¢ is semicoherent with the matrix. The density of the q ¢ phase in the matrix is lower than that of q≤, which nucleates on matrix dislocations (Fig. 6.25)8 and grain boundaries. q phase. After a very long aging time, the equilibrium q, CuAl2, forms, which is incoherent with the matrix phase and is bct with a = 6.066 Å and c = 4.874 Å. The q forms at planar interfaces, dislocations, grain boundaries, or q ¢/matrix boundaries. Figure 6.25 shows the nearly equiaxed particles which have nucleated on the q ¢/matrix interface. Free-Energy Relationship. Let us consider the free-energy relationship in the nucleation of GP1 zones, the transition phases q ≤ and q ¢, and the equilibrium phase q. Figure 6.26 shows the schematic bulk free energy versus composition diagram at a given temperature T for the Al-Cu system with composition C0. Since GP1 zones and the matrix have the same crystal structure, they lie on the same free-energy curve. The solubilities of q, q ¢, q ≤, and GP1 zones, respectively, in equilibrium with the matrix at a given aging temperature T are higher than those of the parent phase. From the common tangent rule, it is clear that GP1 zones have the highest solubility (lowest stability) and that the solubility decreases in the order q ≤, q ¢, q. That is, the solute concentration a1 of the matrix in equilibrium with GP1 zones is higher than a2 for q ≤. Similarly, the solute concentration a2 of the matrix in equilibrium with the q ≤ phase is higher than a3 forq ¢, and the solute concentration a3 of the matrix in equilibrium with q ¢ is higher than a4 for q. Thus, only alloys of solute content greater than a1 can form GP1 zones; alloys of solute content between a2 and a1 can form the q ≤ phase; alloys in the composition range between a3 and a2

6.42

CHAPTER SIX

FIGURE 6.25 Precipitation of q ¢ and q phases in Al-Cu alloy. The equiaxed q particles have nucleated at the q ¢ plate/matrix interfaces and have grown to consume the metastable precipitate.8 (Reprinted by permission of North-Holland Physics Publishing, Amsterdam; after R. Carpenter.)

can form the q ¢ phase; and alloys in the composition range from a4 and a3 must decompose directly to the stable q phase. We can thus conclude that the solubilities a1, a2, a3, and a4 represent the maximum stability limits of GP1 zones, q ≤, q ¢, and q, respectively, at a given temperature T; and that the number of intermediate reaction stages decreases with the decrease of supersaturation. These matrix compositions represent the points on the solvus lines for GP1 zones, q ≤, q ¢, and q, at a particular aging temperature. By taking into account similar free-energy relationships for all temperatures and the common tangent rule, the appropriate phase diagram (Fig. 6.27)8,78 may be constructed which will include the extra solubility lines, that is, metastable solvus curves. It illustrates that as the metastable equilibrium solvus curve drops to a lower temperature range, the stability of the transition phase it represents is lowered. The complete precipi-

NUCLEATION IN SOLIDS

6.43

FIGURE 6.26 Schematic bulk free energy versus composition diagram at a given temperature T for the Al-Cu system with composition C0. It illustrates the relative free energies of GP1 zones and of q ≤, q ¢, and q phases.

L 600

a+L q Solvus

Temperature, °C

a

400

q¢ Solvus q¢¢ Solvus

200 GP Solvus

0 0 Al

1

2

3 4 Wt % Cu

5

6

FIGURE 6.27 Al-rich portion of the Al-Cu phase diagram including the metastable solvus curves.8 (Reprinted by permission of NorthHolland Physics Publishing, Amsterdam; after Hornbogen.)

tation sequence comprising GP1 zones and transition phases can only be achieved if the alloy is aged below the GP1 zones solvus. When aging is carried out at a temperature between the q ≤ solvus and q ¢ solvus, the precipitation sequence will comprise only q ¢ as the first metastable phase. Thus, with the increase of aging

6.44

CHAPTER SIX

FIGURE 6.28 Schematic TTT diagram of stable and metastable phases in the Al-Cu system.

temperature, the number of transition phases formed diminishes; and above q ¢ solvus, no transition phase appears. However, in all situations the precipitation sequence is completed with the formation of a stable q phase. The effect of aging temperature on the time required to nucleate various phases is shown by a schematic TTT diagram in Fig. 6.28. Some important points may be elucidated here. Since the hardening sequence at low temperature comprises as Æ GP1 zone Æ q ≤ Æ q ¢ Æ q and given the fact that DG*GP1 < DG* q≤ < DG* q¢ < DG* q (where DG* denotes the activation barrier energy for nucleation of the respective phase in the subscript), the time t0 required to nucleate each phase is given by t0(GP1) < t0(q≤) < t0(q¢) < t0(q). That is, GP1 zones nucleate more easily than q ≤, q ¢, and q; q≤ nucleates more easily than q¢ and q; and so on. Figure 6.26 illustrates the relative free energies of GP1 zones and those of q≤, q¢, and q phases. This is exemplified in Fig. 6.28, where the upper temperature corresponding to the asymptotic curve is the solvus temperature for a particular phase in an alloy of composition C0. For example, GP1 zones will not form above T1 (as shown in Fig. 6.28) as a result of its instability arising from the increased free energy of this alloy of composition C0. Effect of Trace Additions. Polmear and his coworkers85,86 have noticed that small additions of Mg and Ag into Al-Cu alloys promote a precipitation of disk- or platelike W phase with a {111}a habit plane. The chemistry of the W phase is close to the q (Al2Cu) phase but with trace amounts of Mg and Ag.87,88 6.7.2.2 Al-Cu-Mg (or 2xxx) Alloys. Some of the Al-Cu-Mg alloys, notably 2014, 2024, 2025, 2124, 2219, and 2618 alloys, have been developed for aircraft construction. For high Cu/Mg ratio (= 8), q phase is favored and the aging sequence is summarized as89–91 a s Æ GP1 zones Æ GP2 zones (q ¢¢) Æ q ¢ Æ q where GP1 zones are Cu-rich thin plates, q≤ is fully coherent intermediate precipitate probably nucleated at GP1 zones, and q¢ is tetragonal with a = 0.404 nm, c =

NUCLEATION IN SOLIDS

6.45

0.580 nm; q is the incoherent equilibrium phase nucleated at the surface of q¢ and is bct: a = 0.607 nm, c = 0.487 nm. For alloys with a low Cu/Mg ratio (1.5 < Cu/Mg < 4), the precipitation sequence can be described as a s Æ GPB1 zones Æ GPB2 zones S ¢¢ Æ S ¢ (coherent semicoherent, orthorhombic, lath-shaped, Al 2CuMg) phase Æ S phase

(incoherent, orthorhombic, Al 2CuMg) The GPB1 and GPB2 represent Mg- and Cu- rich (Bagaryatski) GP1 and GP2 zones (or Cu-Mg co-clusters) and form as thin rods on {110} matrix planes; and S¢ plays a key role in age hardening at higher temperatures.90 The formation of these intermediate precipitates may be influenced by trace elements. For example, Si additions restrict the vacancy condensation and favor more homogeneous distribution of S¢ precipitates89 and produce an increase in hardness. The equilibrium S phase forms either by loss of coherency of the S¢ or by heterogeneous nucleation. Like Al-Cu alloys, a small cold deformation after solution treatment is more potent in refining and in improving the precipitate distribution. In an experimental Al-4.2Mg-0.6Cu alloy, Ratchev et al.92 have reported the heterogeneous formation of S≤ phase on dislocation loops and helices and proposed that this is the predominant cause of strengthening in the early stages of precipitation hardening. They also observed S≤ precipitation at the later aging stage, which suggests its significance for precipitation hardening. Chopra et al.93 have observed cubic Z phase as the primary strengthening precipitate in Al-1.5Cu-4Mg-0.5Ag alloy aged at 473 and 513 K. This phase has two orientation relationships with the a-matrix phase, such as (100)Z // (100)a; [010]Z // [010]a at shorter aging time, and (011)Z // (111)a; [011]Z // [011]a. Gao et al.94 have found that the addition of > 0.2 wt% Si to the Al-4Cu-0.3Mg0.4Ag alloy causes complete suppression of W phase that dominates the microstructure of the quaternary alloy. However, the hardness of the Si-modified alloys can still be enhanced with an appropriate amount of Si addition, even in the absence of W phase. 6.7.2.3 Al-Mg-Si (or 6xxx) Alloys. In the Cu-free Al-Mg-Si alloys, the needlelike precipitate plays a main role of age hardening. The precipitation sequence in these alloys is given by a s Æ Co-clusters of Si and Mg atoms Æ GP1 zones Æ GP2 zones /b ¢¢

( needle-shaped precipitates or zones along {100}a containing a high concentration of vacancies) Æ B¢ (lath-shaped precipitates) + b ¢(rod-shaped, Mg 2 Si ) Æ b (plate-shaped, fluorite structure, Mg 2 Si ) The Mg/Si ratio in the Mg-Si co-clusters, the small precipitates, the b≤ precipitates, and B¢ precipitates are close to 1 : 1.95 The zones are oriented parallel to the direction of the Al matrix and contain coherency strain around the needles.90 When ⬃1% Cu is added to the Al-Mg-Si alloy, q (CuAl2, bct), J¢ (Mg2Si), and a third hardening Q phase (Al4Cu2Mg8Si7, hexagonal with a unit cell: a = 1.04 nm and c = 0.405 nm), being possibly precursor to Q¢ phase (Al4CuMg5Si4), appear. When present as very fine precipitates after age hardening, these phases combine to produce an alloy (e.g., 6013) with a higher yield strength than 6061, 6009, and 6010.96

6.46

CHAPTER SIX

Among the Al-Mg-Si(-Cu) system, 6011-T4 has become the material of choice as an autobody (outer panel) sheet alloy due to greater demands for lightweight vehicles to overall improve fuel efficiencies and reduce vehicle emissions. In 6011 alloy, the Mg/Si ratio in the Q phase is ⬃1.0.96a The strengthening of Al-Mg-Si(-Cu) alloys in automotive applications by precipitation hardening is achieved during the paint bake cycle with an average temperature of 175°C for about 20 min. The 6016 Al alloys, aged under the automotive paint bake (180°C for 30 min) and peak-strength conditions (180°C for 11 hr), contain only one recognized precipitating phase, b≤ with the monoclinic structure. In the 6111 Al alloy, aged at 180°C for 30 min and 11 hr, two phases b≤ (monoclinic) and Q (hexagonal) appear together. In 6016 alloy, the Mg/Si ratio increases from 0.85 to 1.01 with extended aging time in the b≤ phase and attains 1.35 in the Q phase.96a In a 6022 alloy containing small amounts of Cu, the precipitation sequence involves as Æ GP zones Æ needlelike b≤ Æ rodlike b¢ + lath-shaped Q¢ Æ b + Si. However, with increased Cu content to 0.91%, the precipitation sequence becomes96b a s Æ GP zones Æ needlelike b ¢¢ Æ lath-shaped Q¢¢ Æ Q( hexagonal ) + Si 6.7.2.4 Al-Zn-Mg(-Cu) (or 7xxx) Alloys. The Al-Zn-Mg(-Cu) systems are widely used in the aerospace industry due to their high specific strength. These alloys attain their strength by precipitation through a complex sequence.96c In Al-Zn-Mg alloy, the role of GP zones and of vacancy-rich clusters in the nucleation of h¢ phase offers a greater response to precipitation-hardening treatments with substantially enhanced strength. In addition to the major alloying elements, and impurity and dispersoid additions, some alloys (e.g., 7010 and 7050) contain Zr for more efficient grain refinement, decreased quench sensitivity, and improved strength and toughness.89 The precipitation sequence in these alloys is summarized as follows: For higher Zn/Mg ratio, as Æ GP zones Æ h¢ [semicoherent, hcp, MgZn2 platelets nucleated on (111) plane] Æ h (incoherent, hcp, MgZn2 plates or rods)96d and for lower Zn/Mg ratio, as Æ GP zones Æ T¢ [semicoherent, metastable, hcp, irregular morphology, Mg32(Al, Zn)49 with high Mg/Zn ratio] Æ T (incoherent, cubic, Al2Zn3Mg3, with high Mg/Zn ratio). The additions of Fe and Si to 7050 Al alloy retard the formation of GP zones and h¢ precipitates and decrease the mechanical properties.96e In 7075, strengthening occurs by h¢ precipitates which are regarded as partially ordered intermetallic compound with a formula Mg4Zn11Al, hcp structure.96f An addition of 0.7 wt% Li in 7075 alloy results in the following precipitation sequence: a s Æ GP zones (vacancy-rich) Æ T ¢ Æ T( Al, Zn) 49 Mg 32 According to this mechanism, the Li-vacancies (Li-v) aggregates act as nuclei for subsequent clustering of Zn and Mg atoms, result in the limited formation of vacancy-rich GP zones, and produce narrow PFZs. This results in an early inhibited nucleation and reduced hardening which can be improved by using increased solution heat treatment or two-step aging.97 6.7.2.5 Al-Li Base Alloys. Al-Li alloys have been developed primarily to reduce the weight of aircraft and aerospace structures. Quaternary alloys are used in a wide range of demonstrator parts on civil and military aircraft; and Weldalite finds applications in the large welded cryogenic fuel tanks in the space launch systems for the NASA program and in the welded structure for the ESA space lab project.89

NUCLEATION IN SOLIDS

6.47

Like other age-hardenable alloys, Al-Li alloys achieve precipitation strengthening by artificial aging after a solution heat treatment. The precipitate structure is a function of quenching rate, the following solution heat treatment, the extent of plastic deformation before aging, and the aging temperature and time. Minor alloying additions have also a major effect on the aging process by altering the interfacial energy of the precipitate, by increasing the vacancy concentration, and/or by increasing the critical temperature for homogeneous precipitation. Furthermore, heterogeneous precipitation at interfaces and grain boundaries has a deleterious effect on fracture behavior. In Al-Li alloys the homogeneous precipitation of coherent, ordered, L12, spherical d ¢ (Al3Li) particles may advance via many transformation steps, including shortand long-range ordering and decomposition. Both ordering and decomposition may occur by nucleation and growth or by spinodal (decomposition) mechanism, increasing thereby the number of possible combinations of consecutive transformation steps.98 In the presence of grain boundaries and dislocations, coarsening of d ¢ is enhanced.99 The precipitation of equilibrium d (AlLi) at the grain boundaries can lead to PFZs which can produce further strain localization and promote intergranular failure. Consequently, dispersoids (Mn, Zr) and semicoherent / incoherent precipitates such as T1 (Al2CuLi), q ¢ (Al2Cu), or S (Al2LiMg) introduced by Cu or Mg additions and thermomechanical processing have been used to minimize the formation of PFZs and optimize Al-Li microstructures, thereby achieving the best combinations of strength and toughness.100 The important Al-Li alloys are 1420 (Al-5Mg-2Li-0.5Mn), 2020, 2090, 2091, 8090, 8091, and Weldalite 049, which are grouped into two categories: those with small Li additions to Al-Cu to enhance strength (e.g., 2020 and Weldalite) and those with high Li contents such as 1420, 2090, 2091, and 8090 alloys to offer the maximum density reduction. Among the various precipitation sequences for Li-containing aluminum alloys, the one for Al-Li-Cu-Mg(-Zr), e.g., 8090 alloy, is very complicated. These are given as89,90,99 Al-Li alloys: Al-Li-Zr: Al-Li-Mg: Al-Cu-Li (low Li/Cu ratio): or Al-Li-Cu (high Li/Cu ratio): Al-Li-Cu-Mg (high Mg/Cu ratio): Al-Li-Cu-Mg-Zr:

as Æ d ¢ (Al3Li) Æ d (AlLi) as Æ d ¢ +a ¢ (Al3Zr) Æ d + a ¢ as Æ d ¢ Æ Al2MgLi (rod-like precipitate, cubic: a = 1.99 nm) as Æ GP1 zones Æ GP2 zones / q ≤ Æ q ¢ Æq (CuAl2) as Æ GP zones Æ T1 [thin plates with {111}a habit plane, hcp (a = 0.497 nm, c = 0.934 nm), Al2CuLi] as Æ T1¢(Al2CuLi) Æ T1(Al2CuLi) as Æ GP zones Æ S¢ Æ S (Al2CuMg) as Æ d ¢ + S¢ (Al2CuMg) + T1(Al2CuLi) + d + a ¢ (metastable or stable Al3Zr)

The role of Zr is considerably more complex. The Al-Li-Zr alloys achieve a peak yield strength of about 360 MPa in 2 to 3 hr whereas corresponding Al-Li-Mn alloys exhibit a maximum strength of ⬃250 MPa after ⬃50 hr at 200°C.99 In Al-Li-Mg alloy, since Mg reduces the solubility of Li, for a given Li content, an increased volume of d ¢ forms. The precipitation of incoherent Al2MgLi occurs as rods with a a growth direction, either by overaging or by heterogeneous

6.48

CHAPTER SIX

precipitation, primarily at grain boundaries, and leading to significant deleterious effects upon ductility and toughness.99 Primarily, the Al-Cu-Li system provides the likelihood of three precipitates such as d ¢, q ¢, and T1. The balance between them depends on the Cu content and stretching: High Cu content leads to increased q ¢ precipitation while stretching and low Cu content promotes T1 at the expense of d ¢.99 In the Al-Li-Cu system, the precipitation of the T1 phase takes up a portion of the Li, and further d ¢ nucleation on q ¢ occurs so that the volume fraction of d ¢ available for strengthening is significantly decreased. In a typical composition with 2.5% Li, 1.4% Cu, and Mg > 0.5%, S¢ dominance prevails.99 In Al-Li-Cu-Mg alloys, the T1 (Al2CuLi) phase still forms at low Mg levels. However, as the Mg/Cu ratio increases, S (Al2CuMg) phase progressively replaces the T1 phase.99 Hirosawa et al.101 have reported that a small addition of Mg to the Al-Li-Cu-Zr alloy (e.g., 8090 alloy) accelerates the formation of GP1 zone, not d ¢ Al3Li phase, leading to improved age hardening; this is attributed to the decreased activation energy for the GP1 zone nucleation due to the Mg/Cu/vacancy complexes. It is believed that dislocations are essential for the nucleation of T1 phase in Al-Li-Cu such as 2090 alloy and S¢ phase in Al-Li-Cu-Mg-Zr such as 8090 alloy. The addition of Ag and Mg induces a very high age-hardening response in Al-Cu-Li alloys due to the promotion of a finer and more uniform dispersion of T1 phase precipitate without recourse to cold work prior to aging.102 In the Al-Li-Cu-Mg-Zr system, Zr precipitates coherently and homogeneously as a ¢, which controls the grain size and shape and acts as a nucleation site for d ¢; consequently a composite a ¢/d ¢ precipitate containing a shell of d ¢ phase around a ¢ precipitate forms, which is more resistant to planar slip than d ¢ and thereby improves the strength of the alloy. That is, these composite precipitates cannot be sheared by moving dislocations, thereby leading to dispersed slip band and subsequent high tensile ductility and fracture toughness, even at cryogenic temperatures.103 6.7.2.6 Al-Ag Alloys. The Al-rich Al-Ag system is useful for the study of heterogeneous nucleation of precipitates because of (1) small differences in size between Al and Ag atoms, resulting in an extremely small strain energy interaction between Ag atoms, vacancies, and dislocations; (2) very small vacancy-solute binding energy; and (3) simplicity of the crystallographic nature of the g ¢ precipitation.104 Figure 6.29 is the Al-Ag phase diagram including coherent miscibility gaps for GP zones.105 If Al-rich Al-Ag alloys containing up to ⬃23 at% Ag are (1) solutionized above the solvus temperature, (2) quenched, and (3) given a low-temperature aging treatment, decomposition of the metastable supersaturated solid solution occurs104 in the sequence a s Æ GP zones Æ g ¢ Æ g (Ag 2 Al ) GP zones in this system are Ag-rich spherical clusters that tend to form by homogeneous nucleation and growth. On aging, the GP zones increase in size but decrease in number, which is revealed by the X-ray pattern which exhibits decrease in ring diameter. Faceting of GP zones in an Al-5.07 at% Ag alloy has been found predominantly at a lower aging temperature (160°C) than at a higher aging temperature (350°C), as shown in Fig. 6.30A and B, respectively.106 These zones also contain ⬃68% Ag.107 On prolonged aging, GP zones gradually dissolve with the nucleation and growth of intermediate g ¢ platelets. However, GP zones do not take part in the nucleation

NUCLEATION IN SOLIDS

At. % Ag 20

10

30

6.49

50

70

K

°F Liq.

1000

89% Liq. +

800

88.5% 1000K Liq. +Ag2Al 70% 839K

Al

55.6%

1200

85%

Al Ag2Al

ary nd

ta

ou

tas Me

800

pb ga

600

ty bili sci mi ble

Al + Ag2Al 400

400

Al

20

40

60

80

Ag

Wt. % Ag FIGURE 6.29 London.)

Al-Ag phase diagram.102 (Reprinted by permission of Butterworths,

of g ¢. This fact has been established by the existence of short streaks in the X-ray pattern. The g ¢ has an hcp structure with a = 2.858 Å and c = 4.607 Å, and the orientation relationship with the matrix is

(0001)g ¢ ||(111)a ; [1120]g ¢ || [1 1 0]a where g ¢ forms as thin coherent platelets (Ag-rich). Initially they form as rodshaped precipitates and later, by growth, become plates. Figure 6.30C shows an electron micrograph of an Al-Ag alloy aged at 160°C for 5 days, illustrating g ¢ precipitates with stacking fault contrast.108 The steps involved in the formation of Ag-enriched stacking faults are (1) Ag enrichment of helical dislocations by the Suzuki mechanism, thereby decreasing the stacking fault energy (SFE); (2) the climb of dislocation onto {111} planes; and (3) subsequent dissociation of the dislocation into two partials separated by a ribbon of a stacking fault (g ¢ precipitate) by the reaction28 Ê a ˆ 110 Æ Ê a ˆ 111 + Ê a ˆ 112 ] Ë 3 ¯[ ] Ë 6 ¯[ ] Ë 2 ¯[ The equilibrium g phase Ag2Al is hexagonal with a = 2.879 Å and c = 4.573 Å and has the same orientation relationship with the matrix as g ¢. It forms either by discontinuous precipitation (i.e., cellular mechanism) at grain boundaries or from g ¢ into g by acquiring misfit dislocations.

6.50

CHAPTER SIX

FIGURE 6.30 Electron micrographs of Al-5.07 at% Ag alloy illustrating (A) dominant faceted GP zones at lower aging temperature T = 160°C (T/Tc = 0.59) and (B) less faceted GP zones at higher aging temperature T = 350°C (T/Tc = 0.85). Note that the T = 160°C specimen has more “angular” zones. Also note that the electron diffraction pattern exhibits diffuse scattering in the {111} and {100} directions around the fundamental reciprocal.103 (C) Electron micrograph of AlAg alloy aged at 160°C for 5 days showing g ¢ precipitate within stacking fault contrast.105 [(A) and (B) Reprinted by permission of Pergamon Press, Plc; (C) reprinted by permission of The Institute of Metals, England.]

NUCLEATION IN SOLIDS

6.51

6.7.3 Cu-Base Alloys Precipitation-hardening wrought Cu-Be alloys find wide applications in electronic components, electrical equipment, control bearings, housings for magnetic sensing devices, and resistance welding systems. The wrought high-strength alloys (C17000 and C17200) contain 1.6 to 2.0% Be and nominal 0.25% Co. The traditional wrought high-conductivity alloys (C17500 and C17510) contain 0.2 to 0.7% Be and nominal 2.5% Co (or 2% Ni). The newer version of high-conductivity alloy is C17410, which contains 2T° (i.e., if q becomes 90° prior to reaching Fm ), the dislocation bypasses the particle by the Orowan process. If we assume that this is the onset of plastic deformation, the following equation for the increase in CRSS, denoted by t0, is obtained: Dt 0 =

2T∞ bd

for Fm > 2T∞

(6.64)

Since the applied stress, t0, acts only on the free distance l = (d - x) between particles, which also represents the average separation between two particles in the slip plane, Eq. (6.64) should be written as Dt 0 =

2T∞ 2T∞ = b(d - x) bl

for Fm > 2T∞

(6.65)

Since the applied shear stress Dt0 causes a dislocation to bow into a radius r(= l/2) for it to extrude between the particles, we may write Dt 0 =

T∞ rb

(6.66)

where the line tension is assumed to be independent of dislocation character. Again it can be shown from Fig. 6.32 that 2r sin q = l

(6.67)

We can also rewrite Eq. (6.61) as Dt =

2T sin q bl

(6.68)

6.8.2 Interparticle Spacing Figure 6.33 shows the Friedel process for dislocation-point-obstacle interaction, which illustrates that each time a glide dislocation breaks free from the obstacle pinning it and encounters one new obstacle, it sweeps out an area A of the slip plane.130 For a regular square array of obstacles, if there are ns obstacles per unit

FIGURE 6.33 The Friedel process for dislocation-pointobstacle interaction.130 (Reprinted by permission of Pergamon Press, Plc.)

6.60

CHAPTER SIX

area of the slip plane, the spacing between them, called square lattice spacing ls, is defined as l s = ns-1 2

(6.69)

2 s

where

(6.70)

A=l

From the geometry in Fig. 6.33, we have A ⬵ hl ⬵ ls2 and

(6.71)

r = l + (r - h) 2

2

2

(6.72)

which yields l2 ⬵ 2 hr

for r >> h

(6.73)

Combining Eqs. (6.67), (6.71), and (6.73), we get l = lf =

ls sinq

(6.74)

where lf is the effective Friedel obstacle spacing. This parameter is a function of square lattice spacing and obstacle strength. For strong obstacle strength, effective obstacle spacing becomes smaller. Combining Eqs. (6.68) and (6.74) yields Dt =

2T (sin q ) bl s

3 2

(6.75)

Equation (6.75) is based upon the following assumptions: (1) a straight dislocation [i.e., a weak obstacle (or soft particle)] and (2) a square array of obstacles. However, for strong obstacles where Orowan looping occurs, Brown and Ham128 have suggested the equation Dt =

0.8(2T sin q ) bl s

(6.76)

6.8.3 Effect of Finite Particle Size The above approach deals with the idealized situation involving point obstacle or smaller particle size relative to the planar particle spacing (i.e., when the volume fraction is small). However, these assumptions do not hold good for real systems, where we have a large volume fraction and a large particle size. For the Orowan process, with finite particles, Ardell131,132 has summarized the following relationships among average particle (or dispersion) radius , the average planar radius , and the volume fraction of precipitates f: p< rs > 2 =f l s2 p < rs > = 4 32 < rs > 2 2 < rs > = 3p 2

(6.77a) (6.77b) (6.77c)

NUCLEATION IN SOLIDS

6.61

Substituting Eq. (6.77c) into Eq. (6.77a) results in 12

Ê 2p ˆ Ê 32 ˆ ls = Á ˜ < rs > = Á ˜ Ë 3f ¯ Ë 3pf ¯

12

(6.77d)

For cuttable particles, an allowance must be made for the finite-sized particles when the bow-out area is measured. This depends on the specific dislocation configuration at breakthrough.130

6.8.4 Coherency Strain Hardening In most cases, coherent particles (formed particularly in the early stages of precipitation) are surrounded by an elastic strain field as a result of small differences in the average atomic volume of the precipitate and matrix. The precipitation of coherent particles—very strong and resistant to cutting and with slight misfit in the matrix—gives rise to strain fields, which hinder the dislocation movement in the matrix.135 Recent theories of coherency strengthening have been developed by Gerold and Haberkorn,136 Gleiter,137 and Brown and Ham,128 based upon treatments using isotropic linear elasticity theory. This theory considers an elastic interaction between a spherical coherent particle of radius r and misfit strain e with an infinite straight-edge dislocation where e=

1 Ê 1 + v ˆ Ê Da ˆ 2 @ d 3 Ë 1- v¯Ë a ¯ 3

(6.78)

where d is the fractional difference in lattice parameter between the particle and matrix. Based on this theory, Gerold and Haberkorn found that the interaction force F between a coherent precipitate and a straight-edge dislocation, gliding on a slip plane and located at a distance z from the center of the particle (see Fig. 6.34), is given by

and

3 2

z2 3 mb e r 3 for 2 > 2 z r 4 12 z2 ˆ z2 3 Ê for 2 < F = 8 mb e zÁ 1 - 2 ˜ Ë r ¯ r 4

3 F =Ê ˆ Ë 2¯

(6.79) (6.80)

FIGURE 6.34 Schematic illustration of the interaction between a spherical coherent precipitate, d < 0, and a positive edge dislocation on a slip plane at a distance z from the center of the particle. In the hatched regions, the elastic interactions are repulsive for this combination of d and b. Reversing the sign of either also reverses the quadrant of repulsive and attractive interaction.131 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

CHAPTER SIX

6.62

where m is the shear modulus of the matrix. This expression yields maximum interaction force Fm Fm = 4 mb e r

at

z2 1 = r2 2

(6.81)

Note that for z2/r2 > 3/4, Fm occurs for an edge dislocation on a slip plane at a distance z from the center of the particle; for z2/r2 < –34 , Fm lies at the matrix-particle interface, and the slip plane is located at the center plane of the particle; for z = 0 (i.e., slip plane located at the center of the particle), Fm = 0. For coherency strain hardening, the flow stress is derived by substituting the average value of the interaction force F of all of the precipitates in the microstructure over a distance z of the central position of the particles with respect to the slip plane (Fig. 6.34)131 into Eq. (6.62). Once a suitable method of averaging is chosen, an average value of Fm is measured. For underaged alloys (small coherent particles), the flow stress Dt0,e, which is equivalent to the CRSS, is given by Dt 0 , e = 4.1e 3 2 m Ê Ë

f ˆ b ¯

12

(6.82)

For large coherent particles a different formula is used, namely, Ê eb3 ˆ Dt 0 , e = 0.7 mf 1 2 Á ˜ Ë < r >3 ¯

14

(6.83)

where m, e, and f have the usual meanings and is the averaged value of the radius of spherical precipitate particles. Thus for small particles, coherency strain hardening increases with an increase in average particle size , whereas for larger particles, the coherency strain hardening decreases with an increase in average particle size , leading to maximum in the strengthening at a critical par1 ticle size which is of the order of /b = /4 e-1, irrespective of the volume fraction.70 The systems in which the coherency strengthening mechanism is expected to predominate are Cu-Co, Cu3Au-Co, Cu-Fe, and Cu-Mn. However, it has been concluded, on the basis of experimental results, that the theories of coherency strengthening are quite inadequate for estimation of the expected strengthening in the underaged, peak-aged, and overaged hardening alloy regimes.131

6.8.5 Orowan Strengthening When (coherent or incoherent) precipitate particles or obstacles are impenetrable (i.e., nonshearable) by the glide dislocation and Fm > 2T°, the contribution of the precipitates (i.e., the CRSS, Dt0, required to make the dislocation pass between the impenetrable obstacles) is determined by the Orowan equation Dt 0 =

2T∞ bl

(6.65)

By assuming T° = 1/2 mb2 in Eq. (6.65), this equation can be simplified to the form Dt 0 =

mb l

(6.84)

This theory was modified by Ashby and others by introducing the relation between the dislocation line tension T and dislocation line energy E for both a long straight

NUCLEATION IN SOLIDS

6.63

dislocation [(Eq. (6.85)] and a curved segment of dislocation [(Eq. 6.86)] lying in the slip plane of an isotropic crystal:128 E(q ) =

mb2 È 1 - v cos 2 q ˘ Ê r1 ˆ ln 4p ÍÎ 1 - v ˙˚ Ë r0 ¯

(6.85)

d 2 E(q ) dq 2

(6.86)

T (q ) = E(q ) +

where q is the angle between the dislocation line or its tangent and its Burgers vector for straight and curved segments of dislocations, respectively, and r1 and r0 are the outer and inner cutoff distances used in the calculation. Combining Eqs. (6.85) and (6.86), we obtain, for an isotropic medium, an effective dislocation line tension T T=

mb2 4p

Ê 1 + v - 3v sin 2 q ˆ Ê 2r ˆ Á ˜ ln Ë ¯ Ë r0 ¯ 1-v

(6.87)

where 2r is the averaged particle diameter. The dislocation under an applied stress no longer lies on a circular arc, but it takes up a position such that its radius of curvature at any position is r=

T Dt b

(6.88)

Since for a pure screw q = 0, whereas for a pure edge q = 90°, the line tension is different for edges and screws. The line tension is lower for an edge dislocation. As a result, edge dislocations will be more flexible, bowing out with ease under an applied stress and encountering more obstacles than screws. Note that Orowan stress is independent of the nature of dislocation due to smaller curvature of a screw dislocation (than that of an edge) associated with a larger average interparticle spacing for the screw dislocation; this causes a compensating effect of the difference in interparticle spacing for the difference in line tension.118 In fact, the appropriate line tension is the geometric mean for edges and screws. Based on all these factors, the final expression for the Orowan stress for a random array of noncuttable, hard spherical particles can be given as138–140 Dt =

0.84 mb Ê 2r ˆ ln 12 2p (1 - v) l s Ë r0 ¯

(6.89)

where 2r is the average particle diameter intersecting the slip plane, the cutoff radius r0 ⬵ 4b, and ls is the effective interparticle spacing [i.e., Eq. (6.69)] due to the large size of the noncuttable particle. For nondeformable precipitates (e.g., incoherent oxides such as silicon oxide and beryllium oxide particles in dispersion-strengthened copper and carbides and coherent g ¢ in both the overaged Co-based alloys and Ni-based superalloy Nimonic PE16), the strength is independent of the properties of the precipitates but is strongly dependent on their dispersion and size.133,141 This can also be concluded from Eq. (6.89). For a constant f, coarsening of particle size is linked with the increase in effective interparticle spacing ls, which leads to decreased yield stress and overaging (Fig. 6.35). This has also been experimentally observed in Co-based alloys containing g ¢ precipitates, where the lowering of strength at larger particle radii is consistent with the Orowan strengthening.130

CHAPTER SIX

6.64

Strength

ttable Noncu s le Partic

articles

Cuttable P

FIGURE 6.35 Variation of strength with particle size defining cuttable and noncuttable particle regimes.72 (Reprinted by permission of Butterworths, London.)

Particle Size

FIGURE 6.36 Shearing of a coherent and soft particle by dislocation across the glide plane.16 (Reprinted by permission of John Wiley & Sons, New York.)

6.8.6 Chemical (or Surface) Strengthening Chemical strengthening arises when a dislocation actually passes through the coherent precipitate particle, causing the particle to be sheared by one Burgers vector across the glide plane by breaking favorable bonds within the particle, thus creating a new lunar-shaped precipitate particle/matrix interface area (Fig. 6.36).16 Using the same schematic diagram (Fig. 6.32) as in particle looping, the CRSS, Dt0, for the cutting mechanism is Dt 0 =

Fm bd

for Fm < 2T∞

(6.90)

It is thus noted that the stress required to shear the obstacles is much less than that necessary to force dislocation loops between precipitate particles.142 Since this process creates two ledges of new precipitate/matrix interface of specific energy gs, interaction force Fm is given by Fm = 2gsb

(6.91)

g sb T

(6.92)

Substituting into Eq. (6.62), we get sinq =

NUCLEATION IN SOLIDS

6.65

Combining Eq. (6.92) and Eq. (6.75), we obtain the theoretically predicted CRSS due to chemical hardening Dt0,c: Dt 0 ,c =

2g 3s 2 b1 2 T 1 2ls

(6.93)

When the value of ls is inserted in Eq. (6.93), the final expression becomes128,131,132,143 12

Ê 6g 3s bf ˆ Dt 0 ,c = Á ˜ < r > -1 Ë pT ¯

(6.94)

This theory predicts that for a given volume fraction f, the CRSS occurs at minimum particle size. In most cases, in which precipitate nucleation is relatively easy, strength increases with the increase in particle size in the initial stages of aging. This is opposite to Eq. (6.94).130 It thus appears that the chemical strengthening mechanism will be of interest in the early stages of the precipitation processes. This is not a principal contributor to the strength in the aged alloys, except for the steadily increasing volume fraction of precipitates with aging and for very small-sized precipitates. This mechanism seems to be important in Al-Cu and Al-Cu-Mg alloys where GP1 zones, q≤, and S¢ are coherent platelike precipitates,65 and in Al-Ag alloys containing Ag-rich spherical particles. Brown and Ham128 interpreted strengthening of copper by Be-rich zones as being due to chemical strengthening. Figure 6.35 shows the variation of strength with particle size for cuttable and noncuttable (nondeformable) particles.72

6.8.7 Order Strengthening Strengthening by ordered coherent precipitate particles takes place when a matrix dislocation shears (cuts) through an ordered precipitate and creates disorder, for example, antiphase boundary (APB) on the slip plane within the precipitate particle (Fig. 6.37).72 In such materials, dislocations move in pairs to restore the perfect long-range order on the glide plane of the precipitate where the trailing dislocation is attracted to the leading dislocation and removes the antiphase boundaries created by the first dislocation (Fig. 6.38).57,128,144,145 In contrast to chemical strengthening where the planar fault is limited to the periphery of the particle, the antiphase boundary extends over the entire glide-plane intersection.133 The APB energy per unit area of the slip plane gapb, which is the force per unit length, opposes the motion

FIGURE 6.37 Ordered precipitate particle (a) sheared by a dislocation in (b) to produce an antiphase boundary (APB) on the slip plane.72 (Reprinted by permission of Butterworths, London.)

CHAPTER SIX

6.66

FIGURE 6.38 Schematic illustrations of the shearing of ordered precipitates by a pair of dislocations where dII is finite in (a) and is equal to zero in (b). It is assumed here at the critical cutting configuration that the spacing LI of the precipitates along the leading dislocation I is equal to LF. The trailing dislocation II interacts with a different set of precipitates, which have been previously sheared by dislocation I. The obstacle spacings LII and LI are different because sheared, ordered precipitates are attractive obstacles to dislocation II. The distance D between the two dislocations changes along the length of the pair.132 (Courtesy of A. J. Ardell.)

of dislocation as it enters the particle. Gleiter and Hornbogen146 developed this theory based on the models of Castagné,147 Ham,148 Raynor and Silcock,144 and Brown and Ham.128 The maximum force of interaction for a single dislocation shearing an ordered precipitate is given by Fm = 2g apb< rs > =

pg apb< r > 2

(6.95)

where is the average planar radius and is the avarage radius of a spherical precipitate particle. The CRSS, Dt0,O, which arises from the interaction of a single dislocation with a random row of ordered precipitate particles by creating antiphase boundaries within them, can be given by substituting Eq. (6.95) into Eqs. (6.62) and (6.75) and using Eq. (6.77).131,132 Dt 0 ,O =

g apb Ê 3p 2g apb f < r > ˆ ˜ Á ¯ b Ë 32T

12

(6.96a)

For a pair of dislocations, the CRSS is one-half of the value given in Eq. (6.96a) which becomes Dt 0 ,O =

g apb Ê 3p 2g apb f < r > ˆ ˜ Á ¯ 2b Ë 32T

12

(6.96b)

Figure 6.38 depicts the schematic illustration of the shearing of ordered, coherent precipitates by a pair of leading and trailing dislocations.132 It is normally assumed that the trailing dislocation is straight. If the trailing dislocations cannot be entirely pulled through the sheared precipitates due to repulsive force exerted by the leading dislocations, Eq. (6.96b) becomes Dt 0 ,O =

g apb 2b

ÈÊ 3p 2g apb f < r > ˆ 1 2 ˘ ˜ - f˙ ÍÁË ¯ 32 T Î ˚

(6.97)

This equation is valid when the cutting mechanism operates. For alloys aged to peak strength, one obtains

NUCLEATION IN SOLIDS

Dt 0 ,O =

g apb Ê 3pf ˆ 2b Ë 8 ¯

6.67

12

(6.98)

where Dt0,O is constant for fixed f and Eq. (6.98) does not make complete allowance for particles of finite size. In complex disordered particles, more than two dislocations may be needed to eliminate the disorder; consequently, several dislocations can be able to move as a coupled unit.149 Order strengthening is the relevant mechanism in Al-Li; Ni-base alloys such as Ni-Al, Nimonic PE16, Nimonic 105, Nimonic 80A, etc.; Co-Ni-Cr superalloys; and Fe-base alloys such as stainless steels and A-286 where strengthening occurs by g ¢ precipitates. In all these systems, the matrices are fcc, and the coherent precipitates have the L12 superlattices. Other alloys strengthened by L12 ordered precipitates are Fe2TiSi in Fe-Ti-Si alloys: Al3Sc in Al-Sc alloys, metastable Al3Zr in Al-Zr alloys, and Pb3Na in Pb-Na alloys.132 Some other systems strengthened by long-range ordered precipitates with different crystallographic structures are NiAlTi, B2 in FeNi-Al-Ti; Ni4Mo, D1a in Ni-Mo; Cu4Ti, D1a in Cu-Ti; Ni3Nb, DO22 in Co-Ni-Cr-NbFe; NiAl, B2 in Fe-Cr-Ni-Al; various intermetallic phases in maraging steels; and magnesium ferrite, spinel in magnesium oxide–iron oxide systems.133 Unless they are very large, they are spherical. In Al-Li alloys and in superalloys, the lattice mismatch is very small, below E1•, the force F can be written as F Ê E12 ˆ = Á1 - 2 ˜ 2T Ë E2 ¯

12

(6.99)

with E1 Ê E2• - E1• ˆ ln(r r0 ) = Á1 ˜ E2 Ë E1• ¯ ln(R r0 )

(6.100)

where E1 and E2 are the dislocation line energies in the particle and matrix, respectively,130 and R and r0 are the outer and inner cutoff radii, respectively, for the elastic strain field of the dislocation. As the particle radius r increases, its strength increases logarithmically.This equation suggests a maximum strength at very small particle sizes. There is a clear evidence that modulus hardening may be an important mechanism of overaging of precipitates growing at a constant volume fraction in several alloy systems. It is unfortunate that the modulus of metastable precipitates are mostly unknown but can be measured indirectly, thus making comparison between theory and experiment, in underaged and peak-aged alloys, more difficult.130

CHAPTER SIX

6.68

6.8.9 Stacking-Fault Strengthening When the stacking-fault energies gsfp and gsfm of the precipitate and matrix phases, respectively, are different, the separation of the partial dislocations will also be different. In addition, when they are both fcc or both hcp structures and when the dislocations move from the matrix into the particle, the glide of the dislocation is hindered and consequently stacking-fault strengthening occurs. A large difference in SFE (denoted by Dg = |gsfm - gsfp|, where gsfp gsfp involving different combinations of relative values of and partial dislocation separation (or ribbon width), Wm and Wp, in matrix and precipitate, respectively (Fig. 6.39); and they arrived at the following equation for maximum force exerted by the split dislocation:131 Fm = Dg l

(6.101)

where Dg has the usual meaning and l is the length of the chord within the particle at the critical breaking condition, as shown in Fig. 6.39c. In the underaged alloy when 2 ˆ Dt 0 ,g = Dg 3 2 Á ˜ Ë 32Tb2 ¯

12

(6.102)

This equation agrees well with the strengthening by g ¢ in dilute Al-Ag alloys151 and by GP zones in Al-Zn-Mg alloys.154

Wm

rs (a)

(b)

l

(c)

FIGURE 6.39 Representation of some of the possible configurations encountered in dislocation-precipitate interactions during stacking-fault strengthening when gsfm > gsfp (Wm < Wp). (a) Wm > 2 , l = 2 ; (b) Wm ⬵ 2 , l = 2 ; (c) Wm < 2 , l < 2.131 (Reprinted by permission of The Metallurgical Society, Warrendale Pennsylvania.)

NUCLEATION IN SOLIDS

6.69

In the Al-Ag alloy system, stacking-fault strengthening is more dominant. The Ag content of the coherent precipitates was observed to be higher for the lower aging temperature. It is observed that SFE decreases significantly with increasing Ag content of the precipitate phase; this increases Dg, thereby increasing the strength of the precipitate phase.129 There are initially no coherency strain fields around these precipitates, and no strong ordering is found. For large particles, Hirsch and Kelly152 found a decrease in stacking-fault strengthening, with an increase in particle size, which explains the overaging behavior.130

6.8.10 Hardening by Spinodal Decomposition This mechanism occurs due to the periodic variation in composition that produces periodic variation in elastic strain, which can result in a significant increase in strength of the alloys. One theory of hardening by spinodal decomposition has been given by Cahn,155 who predicted an increment in the CRSS by assuming the decomposition process to be perfectly periodic in nature: Dt 0 ,s = =

( AhY )2 b 3 6 bT

( AhY )2 b 2 bT

for screws

(6.103a)

for edges

(6.103b)

where b is the wavenumber of the composition modulation, A is the amplitude of that modulation (in atom fraction), h is defined as (1/a)(da/dc) = d(ln a)/dc, where da/dc is the variation of the lattice constant a with composition c (in atom fraction), Y is a function of the elastic constants of the alloy, and T is the dislocation line tension. For elastically isotropic materials, we have Y = E/(1 - n). However, for cubic crystals of normal anisotropy, we have Y=

(C11 + 2C12 )(C11 - C12 ) C11

(6.104)

where C11 and C12 are the elastic constants. Another theory of spinodal decomposition was put forward by Kato et al.,156 who found it much more difficult to move a mixed dislocation than either a pure edge or pure screw dislocation in a periodic stress field, and found this in good agreement with theory.2 Their expression for the flow stress is Dt 0 ,s =

AhY 6

(6.105)

The predictions involved in Eqs. (6.103) and (6.105) are quite different. In Cahn’s theory, the increment in CRSS is linearly dependent on l = 2p/b; however, in the theory of Kato et al., CRSS is independent of l (i.e., independent of b and T). In fcc spinodal alloys, it appears that the periodic distribution of coherent internal stress due to composition variation results in an increase in the yield stress;155,156 however, in bcc spinodal alloys, the periodic variation of elastic moduli arising from very large amplitude of the composition variation plays an important role in the hardening.157 Kato158 has found that both the misfit strain effect due to coherent

6.70

CHAPTER SIX

internal stress and the modulus effect due to periodic fluctuation of elastic modulus are the major contributors to the increased yield strength. 6.8.11 Conclusions In spite of many attempts, there is no precise agreement between the theoretical and experimental results on precipitation-strengthening mechanisms. This situation arises partly due to the shortcoming of the theories and partly due to their experimental inaccuracies in the measurement of small-sized particles and their distribution and in accounting for various shaped particles deviating from the spherical shape. It is also very difficult to calculate the precipitate properties of metastable phases such as misfit, surface energy, modulus, and so forth, in order to measure the contributions of individual strengthening mechanisms. In the case of noncuttable particles, particularly in the overaged and dispersion-hardened conditions, experimental and theoretical Orowan stresses are in good agreement. In other cases any specific mechanism may operate entirely. Moreover, in many commercial alloys, a superposition of strengthening mechanisms prevails because of the formation of different types of precipitates, for example, a combination of coherent hardening and modulus hardening in the Cu-based alloys and a combination of coherency strengthening and order strengthening in Ni-base superalloys.131 In some cases, an additional strengthening mechanism, for example, grain boundary strengthening, that is not associated with the particles may also operate. However, the strengthening mechanisms so far discussed provide only a semiquantitative picture in our understanding.

REFERENCES 1. K. C. Russell, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 3263–3267; The Encyclopedia of Advanced Materials, Pergamon Press, Oxford, 1994, pp. 1807–1810. 2. J. W. Gibbs, Scientific Papers, vol. 1, Dover, New York, 1961. 3. J. W. Christian, The Theory of Transformations in Metals and Alloys, Part I, Pergamon Press, Oxford, 1975. 4. W. A. Soffa and D. E. Laughlin, in Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS-AIME, Warrendale, Pa., 1982, pp. 159–183. 5. L. Farkas, Z. Phys. Chem., vol. 125, 1927, p. 236. 6. M. Volmer and A. Weber, Phys. Chem, vol. 119, 1926, p. 227. 7. R. Becker and W. Doring, Ann. Phys., vol. 24, 1935, p. 719. 8. K. C. Russell, Adv. Colloid Interface Sci., vol. 13, 1980, p. 205. 9. J. E. Burke and D. Turnbull, Progress in Metal Physics, vol. 3, Pergamon Press, Oxford, 1952, p. 220. 10. R. D. Doherty, in Physical Metallurgy, 4th ed., eds. R. W. Cahn and P. Haasen, Elsevier Science BV, Amsterdam, 1996, chap. 15, pp. 1363–1505. 11. Y. Wang, L. Q. Chen, and A. G. Khachaturyan, Scripta Met., vol. 25, 1991, pp. 1969–1974. 12. F. Laszlo, JISI, vol. 164, 1950, p. 5. 13. K. Robinson, J. Appl. Phys., vol. 22, 1951, p. 1045. 14. J. D. Eshelby, Proc. Roy. Soc. London, vol. A241, 1957, p. 376.

NUCLEATION IN SOLIDS

6.71

15. D. M. Barnett, J. K. Lee, H. I. Aaronson, and K. C. Russell, Scripta Met., vol. 8, 1974, pp. 1447–1450. 16. J. D. Verhoeven, Fundamentals of Physical Metallurgy, Wiley, New York, 1975. 17. W. C. Johnson, C. L. White, P. R. Marth, S. M. Tuominen, K. D. Wade, K. C. Russell, and H. I. Aaronson, Met. Trans., vol. 6A, 1975, p. 911. 18. K. C. Chan, J. K. Lee, G. J. Shiflet, K. C. Russell, and H. I. Aaronson, Met. Trans., vol. 9A, 1978, pp. 1016–1017. 19. J. K. Lee, D. M. Barnett, and H. I. Aaronson, Met. Trans., vol. 8, 1977, p. 963. 20. S. R. Speich, L. J. Cuddy, C. R. Gordon, and A. J. DeArdo, in Proceedings of the International Conference on Phase Transformations in Ferrous Alloys, eds. A. R. Marder and J. L. Goldstein, TMS-AIME, Warrendale, Pa., 1984, pp. 361–389. 21. H. I. Aaronson and K. C. Russell, in Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS-AIME, Warrendale, Pa., 1982, pp. 371–397. 22. P. J. Clemn and J. C. Fisher, Acta Metall., vol. 3, 1955, p. 70. 23. W. Huang and M. Hillert, Met. and Mats. Trans., vol. 27A, 1996, pp. 480–483. 24. J. W. Cahn, Acta Metall., vol. 4, 1956, p. 449. 25. A. H. Cottrell, Report on Strength of Solids, The Physical Society, London, 1948. 26. J. S. Koehler and F. Seitz, J. Appl. Mech., vol. 14, 1947, p. 217. 27. F. C. Larche, in Dislocations in Solids, vol. 4, ed. F. R. N. Nabarro, North-Holland Physics Publishing, Amsterdam, 1979, pp. 135–153. 28. G. W. Lorimer, in Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMSAIME, Warrendale, Pa., 1982, p. 613. 29. R. M. Allen, in Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMSAIME, Warrendale, Pa., 1982, p. 655. 30. H. I. Aaronson, in Lectures on the Theory of Phase Transformations, TMS-AIME, Warrendale, Pa., 1975. 31. D. M. Vanderwalter and J. B. Vander Sande, in Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS-AIME, Warrendale, Pa., 1982, pp. 371–397. 32. J. W. Cahn, Acta Met., vol. 5, 1957, p. 169. 33. J. N. Hobstetter, in Decomposition of Austenite by Diffusional Processes, Interscience, New York, 1962, p. 138. 34. R. G. Ramirej and G. H. Pound, Met. Trans., vol. 4, 1973, p. 1563. 35. J. Friedel, Dislocations, Pergamon Press, Oxford, 1964. 36. G. C. Dollins, Acta Metall., vol. 18, 1970, p. 1209. 37. H. F. Merrick, in International Symposium on Superalloys, ASM, Metals Park, Ohio, 1980, pp. 161–190. 38. H. Brooks, Metal Interfaces, ASM, Metals Park, Ohio, 1982, p. 20. 39. R. G. Baker, D. G. Brandon, and T. Nutting, Philos. Mag., vol. 4, 1959, p. 1339. 40. G. C. Weatherly, Philos. Mag., vol. 17, 1968, pp. 791–799. 41. J. W. Cahn, Acta Met., vol. 9, 1961, pp. 795–801. 42. J. W. Cahn, Acta Met., vol. 10, 1962, pp. 179–183. 43. J. W. Cahn, Acta Met., vol. 10, 1962, pp. 907–913. 44. M. Hillert, Acta Met., vol. 9, 1961, pp. 525–535. 45. C. M. F. Jantzen and H. Herman, in Phase Diagrams: Materials Science and Technology, vol. 5, Academic Press, New York, 1978, pp. 128–184. 46. J. E. Epperson, in Encyclopedia of Materials Science and Engineering, ed. R. W. Cahn, suppl. vol. 1, Pergamon Press, Oxford, 1988, pp. 311–318.

6.72

CHAPTER SIX

47. K. B. Rundman, in Metals Handbook, vol. 8, ed. T. Lyman, ASM, Metals Park, Ohio, 1975, pp. 184–185. 48. J. W. Cahn, TMS-AIME, vol. 242, 1968, pp. 166–180. 49. J. W. Cahn and J. E. Hilliard, J. Chem. Phys., vol. 31, 1959, p. 688. 50. J. W. Cahn and J. E. Hilliard, J. Chem. Phys., vol. 28, 1958, p. 258. 51. F. K. LeGoues, Y. W. Lee, and H. I. Aaronson, Acta Met., vol. 32, 1984, pp. 1837–1843. 52. J. E. Hilliard, in Phase Transformations, ASM, Metals Park, Ohio, 1970, pp. 497–560. 53. T. N. Bartel and K. B. Rundman, Met. Trans., vol. 6, 1975, pp. 1887–1893. 54. E. L. Huston, J. W. Cahn, and J. E. Hilliard, Acta Met., vol. 14, 1966, p. 1053. 55. D. E. Laughlin and W. A. Soffa, in Metals Handbook, vol. 9, 9th ed., ASM, Metals Park, Ohio, 1985, pp. 652–654. 56. K. G. Kubarych, M. Okada, and G. Thomas, Met. Trans., vol. 9A, 1978, pp. 1265–1272. 57. K. Binder, in Materials Science and Technology, vol. 5: Phase Transformations in Materials, vol. ed. P. Haasen, VCH, Weinheim, 1991, pp. 405–471. 58. A. M. Mebed and T. Miyazaki, Met. and Mats. Trans., vol. 29A, 1998, pp. 739–749. 59. R. K. Ham et al., Acta Met., vol. 15, 1967, p. 861. 60. J. L. Murray, Int. Met. Rev., vol. 30, no. 5, 1985, pp. 211–233. 60a. ASM Handbook, vol. 3, 10th ed., ASM International, Materials Park, Ohio, 1992. 61. K. C. Russell and H. I. Aaronson, J. Met. Sci., vol. 10, 1975, pp. 1991–1999. 62. J. B. Cohen, in Solid State Physics, eds. H. Ehrenreich and D. Turnbull, Academic Press, Orlando, Fla., vol. 39, 1986, pp. 131–196. 63. J. W. Martin, Precipitation Hardening, 2d ed., Pergamon Press, Oxford, 1988. 64. R. J. Rioja and D. E. Laughlin, Acta Met., vol. 20, 1980, pp. 1301–1313. 65. S. M. Copley and J. C. Williams, in Alloy and Microstructural Design, eds. J. K. Tien and G. S. Ansell, Academic Press, New York, 1976. 66. Metals Handbook, vol. 2, 10th ed., ASM International, Materials Park, Ohio, 1991. 67. Stainless Steels, ASM International, Materials Park, Ohio, 1994. 68. R. A. Lula, Stainless Steel, ASM, Metals Park, Ohio, 1986. 69. Heat Treater’s Guide, 2d ed., ASM International, Materials Park, Ohio, 1995. 70. T. Gladman, Materials Sc. and Tech., vol. 15, 1999, pp. 30–36. 71. J. M. Silcock, T. J. Heal, and H. K. Hardy, J. Inst. Met., vol. 82, 1953–1954, p. 239. 72. R. E. Smallman, Modern Physical Metallurgy, 4th ed., Butterworths, London, 1985. 73. J. W. Martin and R. D. Doherty, Stability of Microstructure in Metallic Systems, Cambridge University Press, Cambridge, 1976. 74. W. F. Smith, Principles of Materials Science and Engineering, McGraw-Hill, New York, 1986. 75. A. Guinier, Nature, London, vol. 142, 1938, p. 142. 76. G. D. Preston, Proc. Roy. Soc., London, vol. A167, 1938, p. 526. 77. R. J. Rioja and D. E. Laughlin, Met. Trans., vol. 8, 1977, pp. 1257–1261. 78. E. Hornbogen, Aluminum, vol. 43, 1967, p. 9. 79. V. Gerold, Z. Metallk., vol. 45, 1954, pp. 593–599. 80. X. Auvray, P. Georgeopoulos, and J. B. Cohen, Acta Metall., vol. 29, 1981, p. 1061. 81. H. Yoshida, D. J. H. Cokayne, and M. J. Whelan, Philos. Mag., vol. 34, 1976, p. 89. 82. T. Matsukara and J. B. Cohen, Acta Metall., vol. 33, 1985, pp. 1945–1955. 83. G. W. Lorimer and R. B. Nicholson, The Mechanism of Phase Transformations in Crystalline Solids, Institute of Metals, London, 1968, p. 36.

NUCLEATION IN SOLIDS

84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 96a. 96b. 96c. 96d. 96e. 96f. 97. 98. 99. 100. 101.

102. 103. 104. 105. 106.

6.73

G. C. Weatherly and R. B. Nicholson, Philos. Mag., vol. 17, no. 148, 1968, pp. 801–831. I. J. Polmear, Trans. TMS-AIME, vol. 230, 1964, pp. 1331–1339. B. C. Muddle and I. J. Polmear, Acta Metall., vol. 37, 1989, pp. 777–789. I. S. Suh and J. K. Park, in Solid Æ Solid Phase Transformations, eds. W. C. Johnson et al., The Metallurgical Society, Warrendale, Pa., 1994, pp. 159–164. W. E. Bensen and J. M. Howe, in Solid Æ Solid Phase Transformations, eds. W. C. Johnson et al., The Metallurgical Society, Warrendale, Pa., 1994, pp. 1109–1114. P. J. Gregson, in High Performance Materials in Aerospace, ed. H. M. Flower, Chapman & Hall, London, 1995, pp. 49–84. Y. Murakami, Materials Science and Technology, vol. 8: Structure and Properties of Alloys, vol. ed. K. H. Matucha, VCH, Weinheim, 1996, pp. 213–276. A. M. Zahra, C. Y. Zahra, C. Alfonso, and A. Charai, Scripta Mater., vol. 39, no. 11, 1998, pp. 1555–1558. P. Ratchev, B. Verlinden, P. De Smet, and P. Van Houtte, Acta Mater., vol. 46, no. 10, 1998, pp. 3523–3533. H. D. Chopra, B. C. Muddle, and I. J. Polmear, Philos. Mag. Letters, vol. 73, no. 6, 1996, pp. 351–357. X. Gao, J. F. Nie, and B. C. Muddle, Materials Sc. Forum, vol. 217–222, pt. 2, 1996, pp. 1251–1256. G. A. Edwards, K. Stiller, G. L. Dunlop, and M. J. Couper, Acta Mater., vol. 46, no. 11, 1998, pp. 3893–3904. R. A. Jeniski, Jr., B. Thanboonsombut, and T. H. Sanders, Jr., Met. and Mater. Trans., vol. 27A, 1996, pp. 19–27. A. Perovic, D. D. Perovic, G. C. Weatherly, and D. J. Lloyd, Scripta Mater., vol. 41, no. 7, 1999, pp. 703–708. W. F. Miao and D. E. Laughlin, Met. and Mats. Trans., vol. 31A, 2000, pp. 361–371. A. Deschamps, X. Bréchet, and F. Livet, Mat. Sc. & Technol., vol. 15, 1999, pp. 993–1000. S. K. Maloney, K. Hono, I. J. Polmear, and S. P. Ringer, Scripta Mater., vol. 41, no. 10, 1999, pp. 1031–1038. H. Tsubakino, A. Yamamoto, and T. Ohnishi, Thermec ’97, The Metallurgical Society, Warrendale, Pa., 1997, pp. 1059–1064. J. H. Auld and S. M. Cousland, J. Aust. Inst. Met., vol. 19, 1974, p. 194. Z. W. Huang, M. H. Loretto, R. E. Smallman, and J. W. White, Acta Met. et Mater., vol. 42, no. 2, 1994, pp. 549–559. J. Lendvai, Mater. Sci. Forum, vol. 217–222, pt. 1, 1996, pp. 43–46. R. Grimes, in The Encyclopedia of Advanced Materials, eds. D. Bloor et al., Pergamon Press, Oxford, 1994. R. S. James, ASM Handbook, vol. 2, 10th ed., ASM International, Materials Park, Ohio, 1990, pp. 178–199. S. Hirosawa, T. Sato, A. Kamio, K. Kobayashi, and T. Sakamoto, Materials Sc. Forum, vol. 217–222, pt. 2, 1996, pp. 839–844; S. Hirosawa, T. Sato, and A. Kamio, Materials Sc. and Engineering A, no. 1–2, February 1998, pp. 195–201. S. P. Ringer, K. Hono, I. J. Polmear, and T. Sakurai, in Solid Æ Solid Phase Transformations, eds. W. C. Johnson et al., TMS, Warrendale, Pa., 1994, pp. 165–170. E. W. Gayle and J. B. Vander Sande, Scripta Met., vol. 18, 1984, p. 473. D. E. Passoja and G. S. Ansell, Acta Metall., vol. 19, 1971, pp. 1253–1261. L. F. Mondolfo, Aluminum Alloys: Structure and Properties, Butterworths, London, 1976. K. B. Alexander, F. K. LeGoues, H. I. Aaronson, and D. E. Laughlin, Acta Metall., vol. 32, 1984, p. 2241.

6.74

CHAPTER SIX

107. M. E. Fine, Met. Trans., vol. 6, 1975, pp. 625–630. 108. R. B. Nicholson, G. Thomas, and A. Nutting, J. Inst. Met., vol. 87, 1958–1959, p. 431. 108a. M. Miki, Y. Ogino, and S. Ishikawa, Thermec ’97: Int. Conf. on Thermomechanical Processing of Steels and Other Materials, eds. T. Chandra and T. Saki, TMS, Warrendale, Pa., 1997. 109. J. H. Harkness, W. D. Spiegelberg, and W. R. Cribb, ASM Handbook, vol. 2, ASM International, Materials Park, Ohio, 1990, pp. 403–427. 109a. S. A. Lockyer and F. W. Noble, Mat. Sc. and Technol., vol. 15, October 1999, pp. 1147–1153. 110. J. M. Oblak and B. H. Kear, Trans. ASM, vol. 61, 1961, pp. 519–527. 111. M. McLean, in High Performance Materials in Aerospace, ed. H. M. Flower, Chapman & Hall, London, 1995, pp. 135–154. 112. Metals and Alloys in the Unified Numbering System, 8th ed., SAE-ASTM, Warrendale, Pa., 1999. 113. E. F. Bradley, ed., Superalloys, ASM International, Materials Park, Ohio, 1988. 114. R. A. Ricks, A. J. Porter, and R. C. Ecob, Acta Metall., vol. 31, 1983, pp. 43–53. 115. W. L. Mankins and S. Lamb, ASM Handbook, vol. 2, 10th ed., ASM International, Materials Park, Ohio, 1990, pp. 428–445. 116. I. Kirman and D. H. Warrington, J. Inst. Met., vol. 94, 1971, pp. 197–199. 116a. R. M. Forbes Jones and L. A. Jackman, JOM, January 1999, pp. 27–31. 116b. Y. Rong, S. Chen, G. Hu, M. Gao, and R. P. Wei, Met. and Mats. Trans., vol. 30A, 1999, pp. 2297–2303. 116c. Adv. Mater. & Process., January 2001, pp. 51–54. 117. Alloy Data: Carpenter Specialty Steels, 1998. 117a. J. R. Crum, E. Hibner, N. C. Farr, and D. R. Muriasinghe, in Casti Handbook of Stainless Steels and Nickel Alloys, Casti Publishing Inc., Edmonton, Alberta, Canada, 1999, pp. 259–324. 117b. G. Aggen, Phase Transformations and Heat Treatment Studies of a ControlledTransformation Stainless-Steel Alloy, Ph.D. Thesis, Renssalaer Polytechnic Institute, Troy, N.Y., August 1963, pp. 30–31. 118. A. M. Beltran, in The Encyclopedia of Advanced Materials, vol. 1, eds. D. Bloor et al., Pergamon Press, Oxford, 1994, pp. 437–441. 118a. Haynes International, Inc. Publication, 1996. 118b. J. F. Grubb, in Uhlig’s Corrosion Handbook, 2d ed., ed. R. W. Revie, John Wiley, & Sons, Inc., New York, 2000, pp. 667–676. 119. S. Paetke and A. R. Waugh, in Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS-AIME, Warrendale, Pa., 1982, p. 769. 120. H. J. Rack and D. Kalish, Met. Trans., vol. 5, 1974, pp. 1595–1605. 121. G. T. Murray, Met. Trans., vol. 12A, 1981, pp. 2138–2141. 122. G. Krauss, Steels: Heat Treatment and Processing Principles, ASM International, Materials Park, Ohio, 1990. 123. F. B. Pickering, Int. Met. Rev., 1976, pp. 227–268. 124. J. M. Wright and M. F. Jordan, Met. Technol., vol. 7, 1980, pp. 473–482. 124a. J. F. Grubb, private communication, 1999. 125. A. Wilm, Metallurgia, vol. 8, 1911, p. 225. 126. P. D. Merica, R. G. Waltenberg, and H. Scott, Trans. AIME, vol. 64, 1920, p. 41. 127. A. Kelly and R. B. Nicholson, in Strengthening Methods in Crystals, eds. A. Kelly and R. B. Nicholson, Applied Science Publishers, London, 1971, pp. 1–8.

NUCLEATION IN SOLIDS

6.75

128. L. M. Brown and R. K. Ham, in Strengthening Methods in Crystals, eds. A. Kelly and R. B. Nicholson, Applied Science Publishers, London, 1971, p. 9. 129. V. Gerold, in Dislocations in Solids, vol. 4, North-Holland Physics Publishing, Amsterdam, 1979, pp. 221–260. 130. D. J. Lloyd, in Strength of Metals and Alloys, eds. H. J. McQueen, J. P. Bailon, J. I. Dickson, J. J. Jonas, and M. G. Akben, Proceedings, 71st International Conference on the Strength of Metals and Alloys, Montreal, Canada, August 12–16, 1985, pp. 1745–1778. 131. A. J. Ardell, Met. Trans., vol. 16A, 1985, pp. 2131–2165. 132. A. J. Ardell, in Intermetallic Compounds, vol. 2, eds. J. H. Westbrook and R. L. Fleischer, John Wiley, & Sons, New York, 1994, p. 257. 133. E. Nembach, Particle Strengthening of Metals and Alloys, John Wiley, & Sons, Inc., New York, 1997. 134. F. B. Pickering, Physical Metallurgy and the Design of Steels, Applied Science Publishers, London, 1978. 135. N. F. Mott and F. R. N. Nabarro, Bristol Conference on the Strength of Solids, Physical Society, London, 1948. 136. V. Gerold and H. Haberkorn, Phys. Status Solidi, vol. 16, 1966, p. 675. 137. H. Gleiter, Z. Angew. Phys., vol. 23, 1967, p. 108. 138. P. B. Hirsch and F. J. Humphreys, Proc. Roy. Soc., London, vol. A318, 1970, p. 45. 139. D. J. Bacon, O. F. Kocks, and R. O. Scattergood, Philos. Mag., vol. 28, 1973, p. 1241. 140. I. D. Embury, Met. Trans., vol. 16A, 1985, pp. 2191–2200. 141. R. B. Nicholson, in Strengthening Methods of Crystals, eds. A. Kelly and R. B. Nicholson, Applied Science Publishers, London, 1971, pp. 535–613. 142. J. W. Martin, Micromechanisms in Particle-Hardened Alloys, Cambridge University Press, Cambridge, 1980. 143. A. Kelly and M. E. Fine, Acta Metall., vol. 5, 1957, p. 365. 144. D. Raynor and J. M. Silcock, Met. Sci. J., vol. 4, 1970, pp. 121–130. 145. H. Gleiter and E. Hornbogen, Acta Metall., vol. 13, 1965, pp. 576–578. 146. H. Gleiter and E. Hornbogen, Phys. Status Solidi, vol. 12, 1965, p. 235. 147. J. L. Castagné, J. Phys. (Paris), vol. 27, 1966, pp. C3–C33. 148. R. K. Ham, Trans. Jpn. Inst. Met., vol. 9, Suppl., 1968, p. 52. 149. J. Greggi and W. A. Soffa, in Strength of Metals and Alloys, Proceedings of the 5th International Conference on the Strength of Metals and Alloys, eds. P. Haasen, V. Gerold, and G. Kestorz, Pergamon Press, Oxford, 1979, p. 651. 150. K. C. Russell and L. M. Brown, Acta Metall., vol. 20, 1972, p. 969. 151. G. Knowles and P. M. Kelly, in The Effect on 2nd Phase Particles on the Mechanical Properties of Steel, BSC/ISI Conference, Scarborough, England, 1971. 152. P. B. Hirsch and A. Kelly, Philos. Mag., vol. 12, 1965, p. 881. 153. V. Gerold and K. Hartmann, Trans. Jpn. Inst. Met., vol. 9, Suppl., 1968, p. 509. 154. I. Kovas, J. Lendvai, T. Uangar, T. Urmezey, and G. Groma, Acta Metall., vol. 25, 1977, p. 673. 155. J. W. Cahn, Acta Metall., vol. 11, 1963, p. 1275. 156. M. Kato, T. Mori, and L. H. Schwartz, Acta Metall., vol. 28, 1980, p. 285. 157. D. Chandra and L. H. Schwartz, Met. Trans., vol. 2, 1971, p. 511. 158. M. Kato, Acta Metall., vol. 29, 1981, pp. 79–87.

CHAPTER 7

PEARLITE AND PROEUTECTOID PHASES

7.1 INTRODUCTION We shall now consider the diffusion-controlled austenite decomposition that occurs in the eutectoid region of the phase diagram. Pearlite, ferrite, and cementite are the main constituents of the microstructure of slowly cooled and isothermally reacted steels, whereas interphase boundary carbides and fibrous carbides embedded in ferrite often occur in low- and high-alloy steels containing carbide formers in addition to the foregoing constituents. In this chapter, important features of various diffusional transformations are discussed, such as nucleation and growth; interlamellar spacing; overall kinetics and mechanisms of pearlite transformation; interphase precipitation; fibrous carbides; proeutectoid phases; and structure-property relationships of ferrite, and mediumand high-carbon ferrite-pearlite, and fully pearlitic microstructures. Finally, the applications of some important medium- and high-carbon steel products are briefly described.

7.2 PEARLITE Pearlite is a two-phase lamellar product of eutectoid decomposition which can form in steels and various nonferrous alloys (such as Ti-Al, Ag-Ga, Cu-Al, and Zn-Al alloys) during transformations under isothermal, continuous-cooling, or forcedvelocity (directional) growth conditions below the eutectoid temperature due to cooperative and synchronized growth of two constituent phases from a parent phase.1–3 A pearlite nodule is composed of multiple colonies; each colony has parallel lamellae which are oriented differently with respect to lamellae in adjacent colonies. Figure 7.1 shows the lamellar structure of pearlite formed isothermally from austenite in a plain carbon eutectoid steel.4 This also exhibits a wide range of apparent interlamellar spacings in different colonies because of intersection of pearlite colonies at different angles with the plane of polish. Table 7.1 lists some ferrous and nonferrous alloy systems that contain eutectoid transformations and produce numerous pearlite morphologies. 7.1

7.2

CHAPTER SEVEN

FIGURE 7.1 Lamellar pearlitic structure formed isothermally at 705°C in a plain carbon eutectoid steel.4 (Reprinted by permission of ASM International, Materials Park, Ohio.)

7.2.1 Mechanism of Pearlite Reaction According to Hillert,5 the possible sites for nucleation of pearlite may be either ferrite or cementite on an austenite grain boundary, depending on the composition and reaction temperature. In hypereutectoid steel, cementite will usually nucleate first; in hypoeutectoid steel, the ferrite will nucleate first. Nicholson,6 in analyzing kinetic data for the formation of pearlite, has also emphasized that ferrite can nucleate first in lower-carbon hypoeutectoid steels. Mehl and Hagel7 proposed the formation of pearlite nodules by sidewise nucleation/growth and edgewise growth into the austenite. Modin8 and Hillert5 first showed that the proeutectoid phase can be continuous with the same phase present in the pearlite. Later work by Dippenaar and Honeycombe9 on 13% Mn-0.8% C (hypereutectoid) steel showed conclusively that nucleation of pearlite occurred on grain boundary cementite and there is a continuity of grain boundary and pearlitic cementite. They have, therefore, confirmed the earlier observation of Hillert 5 that in this alloy sidewise growth occurred as a result of a branching mechanism rather than by repeated nucleation or multiple nucleation events of cementite at the proeutectoid ferrite/austenite interface. That is, all the cementite lamellae were branches from one single lamellae or nucleus of cementite which had grown from the cementite network, and all the ferrite lamellae also joined together to form a continuous crystal. It is thus thought that a pearlite unit is regarded as a bicrystal comprising two interwoven crystals of ferrite and cementite.10 During further growth, more branching occurred until the interlamellar spacing characteristic of the transformation temperature was obtained (Fig. 7.2).5 Considerable evidence has been found from repeated sectioning studies favoring the branching mechanism.5

TABLE 7.1 Eutectoid Transformations in Nonferrous and Ferrous Alloys51,54

Alloy

Eutectoid composition, wt%

Eutectoid temperature °C

°F

High-temperature phase and crystal structure

Low-temperature phases and crystal structures

15.3 Ga

380

716

Cu-Al

11.8 Al

565

1049

b-hcp structure type hP2 Æ a-fcc (cF4) + b¢ ordered hcp (hP9 crystal structure) b-bcc

Cu-Be

6 Be

605

1121

b-bcc

Cu-In

574

1065

b-bcc

Cu-Si

31.4 In (20.15 at% In) 5.2 Si

555

1031

k-hcp

Cu-Sn

22–26 Sn

520

968

b-bcc (e/a = 3.2, Cu5Sn)

Cu-Sn

27.0 Sn

520

968

g-bcc

a-fcc e (e/a = 21/13, gamma brass)

Cu-Sn

32.5 Sn

350

662

d (gamma brass)

Fe-C

0.8 C

723

1333

g-fcc (interstitial C)

Fe-N

2.35 N

590

1094

g-fcc (interstitial N)

Fe-O

23.3 O

560

1040

Wüstite cubic (NaCl)

Ni-Zn

56 Zn

675

1247

b-cubic (CsCl)

Ti-Cr

15 Cr

680

1256

b-bcc

a-fcc e-orthorhombic (Cu3Sn) a-bcc Fe3C (orthorhombic) a-bcc g ¢-fcc (interstitial N) a-bcc Fe3O4 cubic (spinel) b1-tetragonal (Cu-Au) g (gamma brass) a-hcp TiCr2-fcc (MgCu2) a-hcp + intermetallic phase

7.3

Ag-Ga

Zr alloy

b-phase

Reactions observed

a-fcc b¢¢ metastable (9R or hR3 crystal structure)

Lamellar pearlite

a-fcc g2 (gamma brass) a-fcc b¢ (bcc, CsCl) a-fcc d (deformed gamma brass) a-fcc g-cubic (b-Mn) a-fcc g (complex cubic with a DO3 type superlattice)

Lamellar pearlite; granular pearlite Lamellar pearlite Lamellar pearlite; granular pearlite Granular pearlite Non-lamellar; needles of a about which g precipitates Lamellar pearlite; needles of a about which d precipitates Lamellar pearlite Lamellar pearlite Lamellar pearlite; granular pearlite Lamellar pearlite; granular pearlite Lamellar pearlite Lamellar pearlite; granular pearlite Lamellar pearlite

Source: After A. R. Marder and J. A. Kowalik, in Metals Handbook, vol. 9: Metallography and Microstructures, ASM, Metals Park, Ohio, 1985, p. 659.

7.4

CHAPTER SEVEN

FIGURE 7.2 Schematic branching mechanism for the growth of fine pearlite at a lower temperature formed by partial transformation from coarse pearlite at a higher temperature showing sideways growth of pearlite.5 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

It has been observed that the formation of pearlite requires the establishment of cooperative growth of ferrite and cementite.5 This cooperation gradually develops and increases with time, and finally lamellar pearlite forms. The necessary criterion for this cooperative growth is that both phases should have an incoherent interface with the austenite. In some hypereutectoid steels, a low degree of cooperation between the two phases resulted in nonlamellar growth of ferrite and cementite, thereby producing the so-called degenerate pearlite structure. Partially coherent interphase boundaries seriously inhibit cooperation. When the two phases form in a noncooperative mode, it is called a divorced eutectoid transformation (DET) in which divorced or spheroidal cementite particles grow directly from the austenite phase along a cellular g /a reaction front, without encasement in ferrite shells. Use of DET permits the development of fine, equiaxed microstructure with spheroidized carbide particles.11 DET mode occurs at lower undercoolings compared to the pearlite mode observed at higher undercoolings.12 Hackney and Shiflet13,14 presented direct experimental evidence, in an Fe-C-Mn alloy, similar to that of Dippenaar and Honeycombe, for the existence of mobile growth ledges associated with both phases of pearlite at the advancing pearlite/ austenite interface. This, in turn, implies that the advancing pearlite/austenite interface, in Fe-C-Mn alloy, is partially coherent with the austenite and migrates by the lateral movement of steps. Edgewise growth is attributed to ledgewise growth mechanism which takes place despite the lack of a reproducible orientation relationship between the constituents of pearlite and the austenite grain into which the growth is occurring. Later, Hackney15 proposed a linear stability theory which supports Hillert’s branching mechanism by assuming (1) morphological instability of each phase (say of ferrite which produces branching of cementite and vice versa) and (2) ledgewise growth.10,13 This is in contrast to Hillert’s mechanism of unhindered incoherent pearlite interface growth, i.e., uniform attachment. Based on the HackneyShiflet results, it may be inferred that the pearlite reaction occurs by ledgewise cooperative (or shared) growth, and the essential conditions to be encountered are Ga = Gcm and ha /la = hcm /lcm where G’s are growth rates, h’s are ledge heights, and l’s are interledge spacings for the respective phases.16 Currently, the ledge growth mechanism has been experimentally confirmed by many researchers, and it is now widely accepted. This ledge mechanism of pearlite growth was also verified by Whiting and Tsakiropoulos17 for the Cu-Al lamellar eutectoid growth. Recently, Lee and Park10 have suggested a new model in which sidewise growth occurs by time-

PEARLITE AND PROEUTECTOID PHASES

7.5

sequential branching (and not random branching) via the lateral movement of growth ledges, which also causes edgewise growth.

7.2.2 Theories of Pearlitic Growth Figure 7.3a shows a portion of the Fe-Fe3C phase diagram. The region outlined by the extrapolated Ae3 and Aecm phase boundaries corresponds to the area, or range,

FIGURE 7.3 Schematic illustrations of (a) a portion of the Fe-Fe3C phase diagram and (b) the austenite/pearlite growth front.2 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

CHAPTER SEVEN

7.6

of compositions and temperatures for which the austenite compositions can be simultaneously saturated with respect to both ferrite and cementite. As the reaction temperature falls from just below Ae1 toward the nose temperature (of the TTT diagram), the pearlite formed becomes finer in grain size with smaller interlamellar spacing. Figure 7.3b shows a schematic representation of the austenite/pearlite reaction front. Since ferrite and Fe3C have low ( 1. The relation between t, Dp, and S is given by:32 t=

0.12S Dp - 0.12

(7.10)

In a pearlitic steel wire, the relation between fp, initial interlamellar spacing S, and cementite thickness t can be expressed, according to O’Donnelly’s results, by t=

0.15( wt% C)S fp

(7.11)

The increased interlamellar spacing leads to the large size globular cementite particles in colonies.33

7.2.4 The Overall Kinetics of Pearlite Formation The pearlite reaction is a typical nucleation and growth process.34 That is, the rate of pearlite transformation depends on (1) the nucleation rate of pearlite nodules N and (2) the growth rate of these nodules G. Studies by Mehl et al. have shown that N increases with time and with decreasing temperature down to the knee temperature of the TTT curve for a eutectoid steel, as shown in Fig. 7.5.7 N˙ is a structuresensitive parameter and is influenced by austenite grain size, inhomogeneities, and impurities. Like N, G increases rapidly with decreasing reaction temperature or increasing degree of undercooling DT below the A1 temperature, reaching a maximum at the nose of the TTT curve and decreasing again to a lower temperature (Fig. 7.5). Though G is structure-insensitive (independent of grain size), it is a strong function of both temperature and alloying elements. The external morphology of pearlite nodules strongly depends on the ratio N˙ /G, which, in turn, increases with increase in supersaturation, as will now be described.

CHAPTER SEVEN

7.10

725

TEMPERATURE (°C)

700

RATE OF GROWTH RATE OF NUCLEATION

650

600

550 –5 10

10 –4 10 –3 RATE OF GROWTH

10 –4

10 –2

10 –2 10 0 RATE OF NUCLEATION

102

10 –1 mm/sec

104 NUCLEI/mm3/sec

(a)

TEMPERATURE (°C)

800

g

A1

700 600 γ+

500

α + Fe3C α

+F

e

3

C

400 300

10

102

103 TIME (s)

104

105

(b)

FIGURE 7.5 (a) Variation of nucleation rate and growth rate of pearlite with temperature.7 (b) Isothermal transformation curve for plain carbon eutectoid steel. [(a) Reprinted by permission of Pergamon Press, Plc.; (b) reprinted by permission from R. E. Smallman, Modern Physical Metallurgy, 4th edition, Butterworths, London, 1985.]

Pearlite nodules usually nucleate on austenite grain boundaries and grow at a roughly constant radial velocity into the surrounding austenite grains. At temperatures immediately below A1, where the ratio N˙ /G is small, the reaction proceeds very slowly, and therefore relatively few pearlite nodules nucleate; these nodules grow as hemispheres or spheres without interfering with one another (i.e., before impinging on other growing nodules). Though the nodules nucleate at the austenite grain boundaries, the distribution of nuclei may be taken effectively as randomly distributed throughout the austenite because the ratio of the nodule diameter to the

PEARLITE AND PROEUTECTOID PHASES

7.11

austenite grain diameter is large. “Accordingly, the following time-dependent reaction equation due to Johnson and Mehl35 applies: ˙ 3t 4 ˆ Ê pNG f (t ) = 1 - expÁ ˜ Ë 3 ¯

(7.12)

where f(t) is the volume fraction of the pearlite formed isothermally at a given time t; N˙ is the nucleation rate, assumed to be constant; and G is the growth rate, also assumed to be constant. As shown in Fig. 7.6a, this expression yields a typical sigmoidal curve for fixed values of N˙ and G.36 When f(t) is plotted against 4 NG ˙ 3t 4 , a sigmoidal master curve is obtained which illustrates the fundamental kinetic behavior of a nucleation and growth process in a particular alloy (Fig. 7.6b).7 At large undercoolings (i.e., lower reaction temperatures), the nucleation rate becomes much higher (that is, N˙ /G is quite large) and site saturation occurs;37 that is, the boundaries become covered with pearlite nodules prior to the transformation of a significant fraction of austenite. Transformation simply proceeds by the thickening of these pearlite layers into the grains37,38 (Fig. 7.7). Thus the morphology changes as the cooling rate increases, predominantly from a spherical nodule to a hemispherical one, and Eq. (7.12) no longer holds because:9 (1) random nucleation does not prevail; instead, grain boundary nucleation predominates; (2) N˙ is not a constant but a function of time; (3) G also varies from nodule to nodule with time;39 and (4) the nodules are not really spheres. Consequently, Avrami applied the following time-dependent reaction equation: f (t ) = 1 - exp( - kt n ) ln ln

and

(7.13)

1 = ln k + n ln t 1 - f (t )

(7.14)

where k is a temperature-dependent parameter related to constant growth rate, nucleation frequency, and shape factor; and n, varying between 1 and 4, is a quan-

0.8

N = 1000cm3/sec G = 3  10–5/sec

1.0

0.6 0.4 0.2

100

200 Time

400 (a)

800 sec

f '(t), Fraction transformed

1.0

0.8 0.6 0.4 0.2 0 0.1

0.2

0.4 0.6 0.8 1.0 4 . √N G 3.t

2.0

(b)

FIGURE 7.6 (a) Theoretical fraction of austenite transformed to pearlite as a function of time for specific N˙ and G. (b) Master reaction curve for general nucleation, abscissa scale logarithmic.7 (Reprinted by permission of Pergamon Press, Plc.)

7.12

CHAPTER SEVEN

FIGURE 7.7 A partially transformed eutectoid steel at 550°C. Nucleation of pearlite on grain boundaries and subsequent thickening into grains, 120X. (Courtesy of Paul G. Shewmon; after Aaronson.)

tity dependent on the dimension of growth.40 Christian has tabulated possible values for the n exponent [(Eqs. (7.13) and (7.14)] for both the interface-controlled growth including pearlite reaction and the diffusion-controlled growth (early stages of reaction only), which are listed in Table 7.2.41 Cahn and Hagel37 have further developed the Avrami equation into the following forms, assuming site-saturated transformations on various nucleation sites, namely, grain faces, grain edges, and grain corners (see Fig. 6.5): f (t ) = 1 - exp( -2 AGt ) f (t ) = 1 - exp( -pLG t

2 2

(7.15a)

)

4p hG3 t 3 ˆ f (t ) = 1 - expÊ Ë 3 ¯

(7.15b) (7.15c)

Equation (7.15a) corresponds to grain face nucleation, where A is the grain boundary area; Eq. (7.15b) applies to nucleation at grain edge where L is the grain edge length; and Eq. (7.15c) corresponds to grain corner nucleation,37 where h is the number of grain corners per unit volume. [Equations (7.15a) to (7.15c) can be applied only as limiting cases; they have never been verified experimentally for the austenite Æ pearlite reaction.42] The nucleation rate N˙ is irrelevant in all three situations because all grain boundaries at which pearlite can form have already been transformed. Transformation by site saturation is completed when the pearlite nodule migrates halfway across the grain. Hence the time for completion of the reaction is given by:37 t f = 0.5

d G

(7.16)

where d = average grain diameter, G = average rate of pearlite growth, and d/G is the time taken for one nodule to transform one austenite grain. In eutectoid plain carbon steels, N˙ would have no effect below about 660°C, while in hypereutectoid steels, the overall rate would be independent of N˙ below 720°C.37 At high temper-

PEARLITE AND PROEUTECTOID PHASES

7.13

TABLE 7.2 Possible Values of n in Kinetic Law 41 f(t) = 1 - exp(-kt -n) Conditions

n

(a) Polymorphic changes, discontinuous precipitation, eutectoid reaction, interface-controlled growth, etc. Increasing nucleation rate Constant nucleation rate Decreasing nucleation rate Zero nucleation rate (saturation of point sites) Grain edge nucleation after saturation Grain boundary nucleation after saturation

>4 4 3–4 3 2 1

(b) Diffusion-controlled growth (early stages of reaction only) All shapes growing from small dimensions, increasing nucleation rate All shapes growing from small dimensions, constant nucleation rate All shapes growing from small dimensions, decreasing nucleation rate All shapes growing from small dimensions, zero nucleation rate Growth of particles of appreciable initial volume Needles and plates of finite long dimensions, small in comparison with their separations Thickening of long cylinders (needles), e.g., after complete edge impingement Thickening of very large plates, e.g., after complete edge impingement Segregation of dislocations (very early stage only)

>21/2 21/2 11/2–21/2 >11/2 1–11/2 1 1 /2 2.3 1

Source: Courtesy of Pergamon Press, Plc., Oxford; after J. W. Christian.

atures, N˙ remains very low and site saturation will not occur. In this situation, N˙ , where measured, appears to increase with time according to the following equation: N˙ = at m

(7.17)

where a and m are constants. Cahn has found the time exponent n in the Avrami equation to be just 4 + m and k to be a function of grain boundary area for grain boundary nucleation.7

7.2.5 Measurement of Growth Rate of Pearlite In the isothermal reaction technique, the rate of growth of pearlite nodules can be determined by allowing pearlite to grow isothermally for progressively increasing times, by measuring the radius of the largest pearlite nodule from each polished and etched specimen, and by plotting radii of these nodules versus time.5,9 A straight line is obtained; its slope is taken as G, the growth rate of pearlite. This method provides the maximum growth rate rather than an average. It cannot be applied with ease to rapidly transforming specimens and where impingement of growing nodules has become considerable. Despite these limitations, this technique provides results similar to those of the average growth rate method developed by Cahn and Hagel.43

7.14

CHAPTER SEVEN

A hot-stage technique is an alternative method in which cinephotomicrography and thermal analysis are used under continuous cooling conditions, which enables us to perform in situ measurements of growth rate on several rapidly growing nodules.44,45 In the forced-velocity growth method, a high-purity Fe-C specimen, in the form of rod, is moved at a constant velocity with respect to a high-temperature gradient, which causes the pearlite/austenite front to move unidirectionally along the rod in a manner similar to that observed in the directional solidification technique. It is based on the assumption that the pearlite/austenite interface temperature for growth at a given velocity corresponds to the temperature of isothermally transformed pearlite producing the same growth velocity.46,47

7.2.6 Orientation Relationships Based on selected area diffraction (SAD) pattern analysis of pearlite nodules nucleated on the austenite grain boundary, it is found that pearlitic ferrite bears a K-S or N-W orientation relationship (OR) to that of the austenite grain into which it is not growing and has no simple OR with the austenite grain into which it is growing.48 The ORs existing between pearlitic ferrite and cementite are represented by:49 Isaichev OR: (1¯03)cm // (110)a; [010]cm // [11¯1¯]a; [311]cm 0.91° from [11¯1]a Bagaryatskii OR: (001)cm // (112¯)a; [100]cm // [01¯1]a; [010]cm // [11¯1¯]a Pitsch-Petch OR: (001)cm // (52¯1¯)a; [100]cm 2.6° from [131¯]a; [010]cm 2.6° from [113]a where the subscripts cm and a denote cementite and ferrite, respectively. In pearlite four new ORs between pearlitic ferrite and cementite have recently been identified by Zhang and Kelly,49 employing the more accurate CBKLDP method. In pearlitic steel the pearlite follows the New-2, New-3, and New-4 ORs; in hypoeutectoid steels the Isaichev OR is observed. These authors were not able to find the two widely accepted orientation realtionships, namely, the Pitsch-Petch and Bagaryatskii ORs; and in hypereutectoid steels the New-4 OR is found. These are expressed as follows:49 New-2 OR: (1¯03)cm // (1¯01)a; [010]cm 8.5° from [131]a; [31¯1]cm // [11¯1]a New-3 OR: (02¯2)cm // (1¯01)a; [1¯01]cm 2.4° from [13¯1]a; [311]cm // [11¯1]a New-4 OR: (210)cm // (101)a; [001]cm // [131]a; [1¯21]cm 5.95° from [101]a New-4 OR can also be represented by (100)cm 1.4° from (5¯12)a; [010]cm 5.4° from [11¯3]a; [001]cm // [131]a. New-5 OR: (1¯03)cm // (1¯01)a; [010]cm // [131]a; [311]cm 8.5° from [11¯1]a

7.2.7 Pearlite in Nonferrous Alloys Unlike the pearlite reaction in interstitial Fe-C alloys, the eutectoid transformation in a substitutional (nonferrous) solid solution involves diffusion of solute atoms either through the matrix or along the ledges or boundaries.50 In the Cu-Al system, the austenite b phase (bcc) transforms into the a phase (fcc) and the g2 phase (complex bcc) at 565°C (838 K) at a composition of 11.8 wt% Al, 88.2 wt% Cu (Table 7.1). Recently, the eutectoid transformation in this system, previously reported to be volume-diffusion-controlled by Asundi and West (1970), has been recognized to undergo a ledge-growth mechanism assisted by boundary diffusion.50 In the Cu-(22–26)Sn systems, the b phase (bcc electron compound with e/a = 3/2 at

PEARLITE AND PROEUTECTOID PHASES

7.15

the eutectoid composition with Cu5Sn formula) decomposes into the eutectoid product of a (fcc) and g (complex cubic with a DO3 type superlattice) phases by nucleation and growth process when held at temperatures between 520 and 586°C.51 Although the equilibrium phases at temperatures between 480 and 520°C are a and d, the b phase initially decomposes to a and g phases in this temperature range, followed by d precipitation at the a/g interface. In the Cu-In system, the eutectoid transformation occurs as b matrix Æ a + d at 20.15 at% In below Teu = 847 K by volume-diffusion-controlled decomposition or interphase boundary diffusioncontrolled process (Table 7.1). Here the eutectoid colonies nucleate at the b/b grain boundaries and grow into the untransformed b matrix.52 In the fcc Al-78% Zn system, Cheetham and Ridley53 have shown that the eutectoid transformation occurs by boundary diffusion. The eutectoid products in many Ti- and Zr-base alloys consists of a mixture of a (hcp) phase and an appropriate intermetallic phase. In some of these so-called “active” eutectoid systems, the decomposition of the b (bcc) phase into a lamellar aggregate of two phases cannot be suppressed in alloys of near eutectoid composition, even by rapid quenching. Some of these alloys include Ti-Cu, Ti-Ni, Zr-Cu, Zr-Ni, and Zr-Fe. Active eutectoid decomposition in Zr-3 wt% Fe alloys occurs on water quenching from the b phase fields. Sympathetic nucleation and growth may be regarded as a possible mechanism for the straight to wavy morphological transition during this decomposition process.54 Livingston and Cahn (1974) have observed the discontinuous coarsening reaction in polycrystalline eutectoids in Co-Si, Cu-In, and Ni-In when the microstructures were annealed near the eutectoid temperature. Chuang et al. (1988) reported that the transport mechanism in both discontinuous precipitation and subsequent coarsening reaction in Ni-7.5 at% In alloy was attributed to the grain-boundary diffusion.24 Doherty (1982) described the formation of similar coarsened lamellar structure by deformation-induced boundary migration (recrystallization) in Nibased alloys containing coherent Ni3 Al precipitates. In this case the driving force for boundary movement was the stored energy of the dislocations.24 In Ag-Ga systems, two different pearlite morphologies occur. The first one is the coarse pearlite occurring at high reaction temperatures near the eutectoid composition and constitutes equilibrium phases a and b¢, with fcc and hcp (hP2) crystal structures, respectively. The second one is blocky or fine pearlite occurring at a lower reaction temperature and constitutes a and metastable b ¢¢ with a 9R (hR3) crystal structure. Both types of pearlite undergo coarsening by a discontinuous or cellular reaction arising from the lower interfacial energy required to form this type of pearlite compared to traditional pearlite. The lower interfacial energy linked to blocky/fine pearlite results from the planar feature of the interlamellar boundaries and the similar structures of its product phases.3 The unusual presence of both a and b¢¢ found within each lamella of blocky pearlite has not been fully understood, but may be attributed to the low concentration in the metastable parent phase in front of the b¢ (or b¢¢) product lamellae.3

7.3 INTERPHASE PRECIPITATION In 1964, Mannerkoski55 and Relander56 identified two new mechanisms of eutectoid decomposition of austenite containing appreciable amounts of strong carbideforming elements such as Nb, Ti, V, Cr, Mo, and W with different alloy carbide morphologies.57 These are called interphase precipitation (or sometimes interphase boundary carbide precipitation) and fibrous carbide precipitation.58 In both struc-

7.16

CHAPTER SEVEN

FIGURE 7.8 Fe-1.04% V-0.20% C after 5 min at 725°C. Thin-foil electron micrograph showing very fine banded dispersions of vanadium carbide within the ferrite.64 (Courtesy of A. D. Batte and R. W. K. Honeycombe.)

tures, nucleation and growth of alloy carbides occur only on the partially coherent region of the g /a boundaries; their spatial distribution and orientation tend to be strictly controlled; and the carbides are observed to be crystallographically related to the ferrite into which they are growing.57 In the interphase boundary carbide structure, the precipitation of alloy carbides takes place periodically or repeatedly during the g Æ a transformation at the advancing allotriomorphic a /g interface boundaries within a ferrite matrix. Further growth of allotriomorphic ferrite alternates with carbide precipitation; both the distribution and the crystallography of the carbides are influenced by the interphase boundary. This results in a finely banded structure comprising three-dimensional sheets or parallel layers of separate, small carbide particles (typically VC) embedded in a ferrite matrix58–62 (Fig. 7.8). The bands are associated with either planar or curved interfaces which depend on the nature of the mobile g /a interface. The interband spacing is dependent on the temperature of transformation and on the composition. As the temperature of transformation is decreased, the band spacing decreases and extremely fine-scale and rapid transformation occurs. Detailed investigations have indicated that the fine carbide precipitates are often much less than 100 Å (10 nm) in diameter and that the sheet and fibrous interband spacing may vary between 50 Å (5 nm) and 500 Å (50 nm).58 Classically, interphase precipitation reaction is observed in Ti, Nb, or V microalloyed steels (TiCXN1-X, NbCXN1-X, VCXN1-X) (see Chap. 15 also) and in alloy steels (such as Fe-12Cr-0.2C, Fe-10Cr-0.4C, and Fe-4Mo-0.2C alloy steels) with Cr23C6, (FeMo)23C6, or (MoCrFe)23C6 precipitation, as well as in medium- and high-carbon steels, with predominantly or entirely pearlitic microstructure.31,57,63

PEARLITE AND PROEUTECTOID PHASES

7.17

FIGURE 7.9 (a) Fe-12% Cr-0.2% C transformed after 30 min at 650°C. Interphase precipitation of Cr23C6 at g /a interface. (a) Bright field showing ledges in g /a interface. (b) Dark field showing white areas corresponding to precipitate particles.9 (Courtesy of K. Campbell and R. W. K. Honeycombe.)

The planar interphase boundary carbide structure occurs usually at the lower transformation temperature and is produced by the thickening of ferrite allotriomorphs by the ledge mechanism.62 This mechanism of interphase precipitation is operative on the g /a interface, as clearly shown in electron micrographs of isothermally transformed thin foils of Fe-12Cr-0.2C alloy steel at 650°C in Fig. 7.9. Figure 7.9a is a bright field image showing the ledges on the interface (i.e., stepped region of the interface), and Fig. 7.9b is a dark field image showing white areas corresponding to the precipitate particles. For the most part, the nucleation and growth of carbide particles occur only on the immobile, low-energy, partially coherent broad faces of ledges, while the mobile, incoherent risers of the ledges are free of precipitate. The size of the carbide particles becomes smaller as each ledge riser is approached.58,64 The process is repeated for each ledge. Figure 7.10 schematically represents the nucleation and growth of carbides on the g /a interface. Davenport and Honeycombe showed that the relationship between the carbides and ferrite in the interphase precipitated condition (sheets of finely distributed carbides) corresponds to that of the Baker-Nutting orientation relationship: (001)carbide // (001)a; [010]carbide // [110]a. This is also valid for the case of V(CXN1-X). The occurrence of curved type of interphase precipitation has been observed predominantly at the higher transformation temperature in Fe-C-Cr, Fe-C-V, Fe-CMo, and Fe-C-Ni alloys. Based on both the occurrence of this type of transformation mainly at grain boundaries and the absence of ledged planar arrays of carbide particles, some workers have proposed that it is the absence of a low-energy orientation relationship (and thus the ledge mechanism) which causes the g /a interface to bow out between coarsely spaced carbide particles and move between them in the form of a regular sequence of curved interfaces. This produces curved rows of

7.18

CHAPTER SEVEN

(a)

(b)

FIGURE 7.10 Schematic planar interphase precipitation showing (a) uniform steps and (b) irregular steps.64 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

interphase precipitates. However, migration of the g /a interface by ledge mechanisms has still been noticed in such structures, which indicates that partially coherent facets may not be completely absent.65 Based on direct observation of b/a interface in a Ti-Cr alloy, Furuhara and Aaronson66 proposed that curved interphase interfaces are, in fact, partially coherent and can be explained by closely spaced growth ledges lying at an adequate angle to each other or terrace planes stepped up or down by structural ledges. Hence, the ledge mechanism theory can qualitatively describe the interphase boundary precipitation at both planar and curved interfaces.66,67 The sheet and fibrous morphologies of erbium-rich particles in the Ti-1.7 at% Er alloy are surprisingly similar to those found in isothermally transformed alloy steels.67 Note that in low-carbon steel, the pearlitic transformation temperature range extends from 500 to 550°C up to A1 temperature, and the interphase precipitation occurs usually in ferritic microstructure. However, during isothermal pearlitic transformation in vanadium alloyed medium- and high-carbon steels, interphase precipitation of linear or curved arrays of uniformly sized and shaped particles of VC or VCN occurs within both the proeutectoid ferrite and lamellar pearlitic ferrite in the entire temperature and composition ranges of their formations. Figure 7.11a shows a typical example of interlamellar interphase precipitation.68 These interphase precipitates form curved arrays of uniformly sized and shaped VC (or VCN) particles (white) whose alignment is normal to the cementite long axis.63,69 In a mediumcarbon steel, dark field electron micrographs often exhibit interphase precipitation in both the ferrite and pearlitic ferrite (as shown in Fig. 7.11b). Note that the same vanadium carbonitride reflection will illuminate the interphase precipitates (under dark field conditions) in both the ferrite and contiguous pearlite regions, which implies the presence of some continuity in the ferrite orientation between the polygonal ferrite and interlamellar ferrite. However, VC does not precipitate within the pearlitic cementite; moreover, the pearlitic cementite is not supersaturated in vanadium.63,68

PEARLITE AND PROEUTECTOID PHASES

(a)

(b)

FIGURE 7.11 (a) Vanadium carbonitride particles having the interphase precipitate morphology and distribution in the interlamellar ferrite of the pearlite eutectoid structure (dark field TEM). (b) Vanadium carbonitride precipitation produced in a mediumcarbon (Fe-0.1C-0.1SV) steel (by isothermal transformation at 600°C for 1 hr) having a ferrite-pearlite structure. Note that the VCN in both the ferrite and contiguous pearlite is illumined by the same diffracted beam, suggesting a common variant of the same orientation relationship in a common ferrite orientation dark field TEM.63 [(a) Courtesy of T. Gladman; (b) courtesy of G. Fourlaris.]

7.19

CHAPTER SEVEN

7.20

7.4 FIBROUS CARBIDES Unlike the interphase precipitations that usually form periodically in bands and have planar, irregular, curved g /a interfaces, fibrous carbides involve colonies of fine (~20 to 100 nm in diameter and up to 10 mm in length), parallel, alloy carbide needles, rods, or fibers with little or no branching. The rare occurrence of branching, the fibrous morphology, and growth nearly normal to the reaction front are characteristic features which distinguish this mode of transformation from that of pearlite. In heat-treated low-alloy steels containing carbide-forming alloying elements, fcc alloy carbides (e.g., NbC, TiC, VC, and Cr23C6), hcp carbides (e.g., Mo2C and W2C), and rhombohedral carbide Cr7C3 can form fine fibrous aggregates. In isothermally transformed steels, fibrous carbide can form in association with interphase precipitation. For example, within one allotriomorphic ferrite crystal formed at an austenite grain boundary, the fibrous carbides (Mo2C or VC) can form on one side and interphase precipitation on the other (Fig. 7.12a).39,64 On the fibrous carbide side of the boundary, the ferrite does not show a low-energy orientation relationship with the austenite in which it is growing, and the g /a interface does not correspond to a rational plane, thereby possibly resulting in a high-energy incoherent interface. In contrast, on the interphase precipitation side, the ferrite bears a Kurdjumov-Sachs orientation relationship with the austenite, and the bands are parallel to the reaction front such that (111)g // (110)a. Figure 7.12b is a TEM micrograph of the Fe-5Cr-0.2C alloy specimen isothermally transformed at 650°C for 30 min. This figure illustrates both the carbide

(a)

(b)

FIGURE 7.12 TEM micrographs. (a) The formation of fibrous carbide phase on one side and interphase precipitation on the other side of the allotriomorphic ferrite. Specimen Fe-4% Mo-0.2% C steel isothermally transformed at 700°C after 0.5 hr.64 (b) The Fe-5% Cr-0.2% C alloy specimen isothermally transformed at 650°C for 0.5 hr, illustrating both the fibrous and interphase precipitation.70 [(a) Courtesy of F. G. Berry and R. W. K. Honeycombe; (b) courtesy of K. Campbell and R. W. K. Honeycombe.]

PEARLITE AND PROEUTECTOID PHASES

7.21

morphology—fibrous—and interphase precipitation, exhibiting their characteristic features but incorporating the same Cr7C3 phase.70 The parameters that determine the relative proportion of fibrous and interphase carbides are alloying additions, transformation temperature, and the crystallography of the a /g boundary from which these structures develop. However, studies made on Fe-C-Mo alloy partially reacted at 850°C and then further reacted at 725°C show that an interphase boundary carbide structure obtained at the higher temperature could be changed into a fibrous one at the lower temperature. Associated with this conversion was a change in the shape of the g /a interface from ledged to smoothly curved, which suggested that there were significant influences of both lattice orientation and boundary orientation in the determination of the g /a interfacial structure and mechanism of eutectoid reaction.64 For isothermal transformation in the temperature range of 600 to 900°C, fibrous carbide does not occur in Fe-1V-0.2C steel, while it is predominant in Fe-4Mo-0.2C steel due to favorable reaction kinetics;64 the extent of precipitation of fibrous Mo2C increases when the transformation temperature falls from 850 to 600°C, which implies that the slower transformation enhances the growth of fibrous carbides.71 It has been noted that when a sufficient proportion of an austenite-stabilizing element (e.g., Mn) is present in Fe-13Mn-2V-0.8C steel, the vanadium carbide phase occurs profusely in the fibrous mode.72 It may be inferred again that the slow reaction favors the g /a interface movement toward the fibrous mode, while rapid reaction causes abundant formation of interphase precipitation.

7.5 PROEUTECTOID PHASES As discussed in Chap. 1, within the a + g region or below the A1 temperature, the separation of proeutectoid ferrite precedes pearlite formation in hypoeutectoid steels, and the separation of proeutectoid cementite precedes pearlite formation in hypereutectoid steels. The proportion of proeutectoid constituents decreases as the carbon content of steel approaches the eutectoid composition (0.8%C) and as the reaction temperature decreases. The proeutectoid constituents (especially ferrite) are identical in crystal structure to those present in pearlite, but their distribution in the microstructure is quite different from their lamellar arrangement in pearlite.

7.6 FERRITE MORPHOLOGY Precipitation of proeutectoid ferrite forming from austenite, at the austenite grain boundary, occurs in shapes that have been classified as allotriomorphs, primary and secondary sideplates, and primary and secondary sawteeth. Ferrite morphologies forming in the interior of austenite grains include intragranular plates and equiaxed idiomorphs.73–75 The so-called massive structure results from the impingement of crystals of other morphologies. Figure 7.13 shows the Dubé morphological classification system for proeutectoid ferrite in steels which includes six components. As in the Fe-C system, Widmanstatten plates and needles have been found in nonferrous systems, for example, Cu-Zn, Cu-Cr, Al-Cu, Al-Mg-Zn, Al-Ag, and Al-Au systems. Most of these systems also exhibit other Dubé morphologies, particularly those of grain boundary allotriomorphs.

CHAPTER SEVEN

7.22 (a)

(1)

(2)

(1)

(2)

(1)

(2)

(b) (c)

FIGURE 7.13 Dubé morphological classification system for ferrite crystals: (a) grain boundary allotriomorphs; (b) Widmanstätten sideplates, (1) primary, (2) secondary; (c) Widmanstatten sawteeth, (1) primary, (2) secondary; (d) idiomorphs; (e) intragranular Widmanstätten plate; and ( f) massive structure.73 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

(d)

(e)

(f)

7.6.1 Allotriomorphic Ferrite The major ferrite morphology that forms at relatively high transformation temperature near A3 in hypoeutectoid plain carbon and low-alloy steels is that of grain boundary allotriomorphs.73 Allotriomorphic ferrite crystals nucleate first during the decomposition of austenite, at the prior austenite grain boundaries (Fig. 7.14a and b),76,77 and grow preferentially along these boundaries (i.e., more rapidly than thickening), and have shapes approximating a double spherical cap or an oblate ellipsoid. The g /a interface is thought to be of the disordered type; however, increasing evidence appears to indicate this to be in favor of an interface comprising coherent and disordered areas.73 They form at small, as well as at large, undercoolings below A3 and also below A1. Their appearance at small undercooling requires the prolonged isothermal reaction or slow cooling through a + g region. 7.6.1.1 Growth Kinetics Thickening Kinetics. This section deals with the planar boundary growth problem. This approximates growth of a grain boundary allotriomorph in thickness direction, that is, migration of the broad faces of a grain boundary allotriomorph. During the formation of ferrite, carbon diffuses in the austenite ahead of the a/g interface. Figure 7.15a shows a portion of the Fe-Fe3C diagram, and Fig. 7.15b shows a plot of carbon concentration normal to the a/g boundaries. If we assume complete local equilibrium at the a /g interface, the carbon concentrations in a and g at the a/g interface, denoted by Caag and C ga g , respectively, correspond to those in the equilibrium phase diagram (Fig. 7.15a). If a unit area of the a/g interface advances a distance ds, the amount of carbon removed from a and piled up in front of the ag advancing a/g boundary is ds (C ga g - C a ). In order to maintain equilibrium at the a/g interface, this amount of carbon must be equal to the product J dt, where J is the flux of carbon atoms diffusing away from the a/g interface in the austenite in a direction perpendicular to this interface. That is, ∂Cˆ - J dt = Dcg Ê dt = ds(Cgga - Caag ) Ë ∂ S ¯ ga

(7.18)

PEARLITE AND PROEUTECTOID PHASES

7.23

FIGURE 7.14 0.5% C-0.06% Si-0.7% Mn steel austenitized for 1 hr at 1350°C, cooled at 300°C/hr (etchant; Picral). (A) Different morphologies of ferrite. 100X (reduced 66%).27 Observation of separate ferrite morphology in isothermally reacted 0.29%C steel at (B) 13 s at 750°C (grain boundary a); (C) 2 min at 700°C (sideplate); (D) 20 s at 600°C (intragranular plate); (E) 5 min at 725°C (intragranular idiomorph).76 (Reprinted by permission of ASM International, Materials Park, Ohio.)

CHAPTER SEVEN

7.24

FIGURE 7.15 (a) A portion of the Fe-Fe3C phase diagram. (b) Carbon concentration profile along a direction perpendicular to the g /a interface. (c) Carbon concentration profile in front of growing ferrite at different time intervals or different volume fractions of ferrite formed. Shaded regions below and above C0 in (b) and (c) are the same.

The growth rate G of ferrite is ds Dcg -J Ê∂Cˆ = G = ga = ga ag Ë ag dt Cg - Ca Cg - Ca ∂ s ¯ interface

(7.19)

The concentration gradient ∂C/∂s at the interface in Eq. (7.19) can be approximated, using Zener’s linearized concentration gradient expression, as DC/L, where DC is the difference of carbon concentration in austenite near to, and remote from, the g /a interface and L is the effective diffusion distance: Cgga - C0 Ê∂Cˆ = Ë ∂ s ¯ interface L

(7.20)

It is clear from Fig. 7.15b that the shaded region below C0 represents the amount of carbon which diffused away from the ferrite as it grows a distance s. All of this carbon content piled up near the a/g interface must diffuse further into the austenite. Assume the same cross-sectional area of the two phases; then conservation of solute requires that the two shaded areas (above and below C0) must be equal, that is, A = A¢, A = s(C0 - C aag ), and A¢ = (L/2)(C gg a - C0). Substituting Eq. (7.20) into Eq. (7.19), we obtain ds Dcg = ga dt Cg - Caag

Cgga - C0 Ê Cgga - C0 ˆ Á ˜= Ë ¯ 2 s(C0 - Caag )(Cgga - Caag ) L

(7.21)

On integration of this equation, we get s=

(Cgga - C0 ) Dcg t (C0 - Caag )(Cgga - Caag )

(7.22)

or s = at1/2

(7.23)

PEARLITE AND PROEUTECTOID PHASES

7.25

where a=

(Cgga - C0 ) Dcg (C0 - Caag )(Cgga - Caag )

(7.24)

where s is the half-thickness of the allotriomorph and a is the parabolic rate constant for thickening. This equation is the parabolic law of thickening and states that the thickening of ferrite layer s varies with the square root of the growth time t. Since the extent of the carbon diffusion field, that is, effective diffusion distance, increases with time, the growth rate of ferrite must decrease with time (Fig. 7.15c). That is, G µ t -1/2 because L @ Dcg t . Note that this equation is based on three assumptions: (1) The interface boundary is planar and disordered, (2) the migration kinetics are volume diffusion-controlled, and (3) Dgc is composition-invariant.76 As already noted, the foregoing derivation of allotriomorph thickening kinetics is based upon Zener’s linearized gradient approximation. An exact expression for the parabolic rate constant for thickening a,78 though still based on the assumption that Dcg is independent of carbon concentration, has been derived by Dubé79 and Zener:80 Cgga - C0 Ê Dcg ˆ Á ˜ Cgga - Caag Ë p ¯

12

=

˘ È a a Ê a2 ˆ expÁ ˙ ˜ erfc Í 1 2 g Ë 4Dc ¯ 2 ÍÎ 2(Dcg ) ˙˚

(7.25)

where Dcg is the weighted average diffusivity of carbon in austenite81 and t is the growth time. 7.6.1.2 Experimental Study of Growth Kinetics of Allotriomorphs. An experimental study of the kinetics of lengthening and thickening of grain boundary allotriomorphs in the Fe-C system consists of reacting isothermally a number of similar samples for different times and then quenching. The half-thickness (or halflength) of the largest allotriomorph on the plane of polish in each specimen is measured and plotted as a function of the square root of the reaction time. Provided that the austenite grain boundaries are perpendicular to the plane of polish, the allotriomorph length and thickness should correspond to the dimensions occurring in three-dimensional space of those first nucleated. In this way, Bradley and Aaronson determined the values of a and b, the parabolic rate constants for thickening and lengthening of allotriomorphs, respectively. When these values of a and b were compared with the theoretical values [assuming allotriomorphs to be oblate ellipsoid with the experimentally observed aspect ratio (~1/3) and the variation of diffusivity with concentration82], they found experimental allotriomorph growth kinetics data to be considerably lower (i.e., less than an order of magnitude) than those calculated, although mainly at lower undercooling and carbon concentration. This was attributed to the existence of a variable proportion of partially coherent facets at the broad, disordered-type faces of the allotriomorphs, as proposed by Kinsman and Aaronson, which resulted from the low-energy orientation relationship of the allotriomorphs with at least one adjacent austenite grain.83 These facets migrate (and grow) by a ledge mechanism at a slower rate compared to those of incoherent g /a boundaries where migration occurs by essentially uniform attachment of Fe atoms and removal of carbon atoms. Measurements of growth kinetics of ferrite allotriomorphs in Fe-C-X alloys (where X = Mn, Ni, Si, or Cr) reacted over the widest experimentally accessible temperature range were made by Bradley and Aaronson84 utilizing the Hillert-

7.26

CHAPTER SEVEN

Staffansson85 regular solution treatment of ternary-phase equilibrium, and they compared their results on the basis of three different models (hypotheses) of the growth process. These models included (1) full local equilibrium, also called orthoequilibrium, at a/g boundaries with bulk partition of both X and C between g and a above a critical temperature. This is called X partition under local equilibrium (PLE); (2) full local equilibrium at a /g boundaries with localized “pile-up” of X in front of the advancing a /g boundaries below the critical temperature, called X negligible partition under local equilibrium (NP-LE); and (3) paraequilibrium at g /a boundaries without partitioning of X in a and g (i.e., with partitioning only of carbon in a and g). The best overall agreement between experimental and calculated growth kinetics was achieved with the paraequilibrium hypothesis. It has been demonstrated by Aaronson and Domian86 that the bulk partitioning of alloying elements during ferrite growth in Fe-C-X alloys (X = Mn, Ni, Pt) occurs at low undercooling but not at higher undercooling. No partitioning was found for X = Si, Mo, Co, Al, Cr, and Cu. They also found, in some cases, a significant deviation of the ratio of the corrected experimental parabolic rate constant for allotriomorph thickening acorr to the rate constant calculated from the paraequilibrium model apara from unity. Shiflet et al.87 suggested interphase boundary carbide precipitation and segregation of either Si or Ni to disordered g /a boundaries to be responsible for this discrepancy in Fe-C-Si and Fe-C-Ni alloys. Mainly, however, these discrepancies are attributed to a solute draglike effect (SDLE). Elements that lower the activity of carbon in g segregating to disordered areas of a/g boundaries diminish the activity gradient of carbon in g driving ferrite growth.78,84 In case of three quaternary Fe-C-Mn-X2 alloys (where X2 = Si, Ni, and Co), the parabolic rate constant a for thickening was observed to be an order of magnitude higher than the amount predicted by the P-LE mode in the alloying element diffusion-controlled regime in the applicable temperature range, whereas the opposite was found to be true in the carbon-diffusion-controlled regime. Likewise, the calculated paraequilibrium constant was normally quite larger than the experimentally determined value. Substantial discrepancies between the measured and calculated curves were attributed to the synergistic effects of Mn and X2 upon growth.88 7.6.1.3 Interface Characteristics and Orientation Relations. Based on the foregoing discussion, the allotriomorph interface does not move uniformly at all elements of the interface. In fact, partially coherent facets present along the disordered-type interface tend to slow down the growth rate. At lower isothermal transformation temperatures, the disordered segments tend to decrease while the slowly moving partially coherent interface becomes more dominant. The ferrite allotriomorphs nucleate with a Kurdjumov-Sachs orientation relationship with (at least) one bounding austenite grain g1:

{111}g // {110}a ; < 110 > g // < 111 > a But they also grow into the neighboring austenite grain g2 with which they have a random orientation relationship. However, recent TEM observations of the interfacial structure of the broad faces of a allotriomorphs precipitated from a hypoeutectoid Ti-Cr alloy—where retention of the b matrix is permitted by an Ms far below room temperature—have shown that even in the absence of a low-energy orientation relationship (Burgers relationship in this bcc-to-hcp transformation), both the rationally and the irrationally oriented interfaces of proeutectoid a allotriomorphs are partially coherent in b Ti-Cr alloys.89 Presumably the same situation applies to ferrite allotriomorphs.

PEARLITE AND PROEUTECTOID PHASES

7.27

7.6.1.4 Relief Effects. Using scanning tunneling microscopy, Bo and Fang90 have reported the existence of surface relief effects associated with the formation of grain boundary allotriomorphs in Fe-C alloys. According to them, the interface between the grain boundary allotriomorph and austenite matrix at one side of the grain boundary is partially coherent, and the interface between the grain boundary allotriomorph and adjacent austenite matrix at another side of the same grain boundary may be disordered. The sympathetic nucleation-ledgewise growth mechanism is suggested to explain the formation of grain boundary allotriomorphs. Note that the formation of massive ferrite does not produce martensite-like surface relief effects. However, growth of both sideplates and intragranular plates is associated with relief effects at the free surface similar to those produced by bainite and martensite plates.

7.6.2 Widmanstätten Ferrite At lower transformation temperature, the ferrite precipitates adopt a plate morphology called Widmanstätten ferrite aw. Widmanstätten plates nucleate mainly from prior g grain boundaries or develop from existing grain boundary allotriomorphs. The broad faces of aw plates are believed to be partially or fully coherent with the matrix phase.73 Figure 7.16a shows a schematic TTT diagram for a plain carbon steel for the precipitation of grain boundary and Widmanstätten ferrite from austenite. An approximate temperature Tw, at which Widmanstätten ferrite appears from allotriomorphic ferrite, is also shown in the figure. Widmanstätten ferrite becomes dominant below this temperature, but ferrite allotriomorphs do not disappear completely (Fig. 7.14A). Figure 7.16b73 is a portion of the Fe-Fe3C diagram showing temperature and composition regions in which various morphologies are dominant. Above the temperature Tw, the ferrite forms as grain boundary allotriomorphs, and below this temperature it is mainly in the form of Widmanstätten ferrite plates. It is apparent that Tw is roughly parallel to the A3 temperature. Lowering the isothermal transformation temperature causes: (1) a decrease of the proportion of grain boundary allotriomorphs, (2) an increase in the number of Widmanstätten ferrite sideplates, (3) appearance of sideplates in groups into several plates growing from the same allotriomorph, (4) a decrease in an average spacing between parallel sideplates, and (5) physical impingement of Widmanstätten sideplates.91 Recently, Enomoto has used finite difference techniques to develop a rigorous model for estimating Tw temperature in Fe-C alloys, based on the growth of the ledge mechanism.92 In fact, increasing supersaturation causes changes in various Widmanstätten morphologies.93 At moderate undercooling below A3, primary sideplates, the so-called Widmanstätten ferrite primary sideplates, grow directly from the preexisting grain boundary as platelets in the matrix phase. However, these morphologies are scarce. They become finer with increasing undercooling. The important factors which contribute to the fineness of these sideplates are the increasing aspect ratio (i.e., thickness/length) with high undercooling and the fact that the radius of curvature at plate tip is inversely proportional to the undercooling. This is encouraged by large austenite grain size. According to Aaronson, they form at low-angle austenite grain boundaries provided the matrix is fcc and the precipitate has another structure;93,94 their rate of nucleation is large. It has been indicated by many workers that these sideplates lengthen by a ledge mechanism.95 This morphology has also been found in Al-Mg-Zn-type alloys.96

CHAPTER SEVEN

7.28

T

A3 γ

Grain boundary a

TW

Widmanstatten a

Log time (a)

1000 γ

900 TW

Temperature °C

800

GBA

M 700 600 500

GBA

W

Pearlite

Pearlite and Bainite

Pearlite and Bainite

Bainite

400 0

W

0.1

0.3

0.5

0.7

0.9

1.1

1.3

1.5

Weight percent carbon (b)

FIGURE 7.16 (a) Typical TTT curve for proeutectoid ferrite transformation. (b) Fe-Fe3C diagram showing temperature versus composition regions in which various morphologies are dominant at late reaction times in specimens with ASTM grain size nos. 0 to 1. GBA, grain boundary allotriomorphs; W, Widmanstätten sideplates or intragranular plates; M, massive ferrite.73 [(a) Reprinted by permission of Van Nostrand Reinhold International; (b) reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.]

Secondary sideplates develop from grain boundary precipitates (usually allotriomorphs) formed earlier (Fig. 7.14A and C). These secondary ferrite sideplates (also called Widmanstätten ferrite secondary sideplates), have been reported to occur much more frequently than the primary variety.93 They usually form in group morphologies which display a large degree of regularity, and consist of several evenly spaced plates growing such that their leading edges define an approximately planar

PEARLITE AND PROEUTECTOID PHASES

7.29

front.93 They form most readily in plain carbon steels containing less than 0.3 to 0.4% carbon under conditions of large austenite grain size and intermediate temperature range of transformation. Another mechanism of secondary sideplate formation involves a process in which primary sideplates initially nucleate directly from the grain boundary, followed by rapid lateral impingement along their bases to form apparent grain boundary allotriomorphic film; hence, at very early stages in their growth, they resemble secondary sideplates. Face-to-face sympathetic nucleation of additional primary sideplates may also contribute to this process. On the second mechanism, sympathetic nucleation of ferrite sideplates takes place atop preexisting grain boundary ferrite allotriomorphs. Thus, small misorientations at the a : a boundaries are noticed between the sideplates and the allotriomorphic regions with which they are associated.97 Primary and secondary sawteeth ferrite can be regarded as morphologies intermediate between grain boundary allotriomorphs and primary or secondary sideplates. Primary ferrite sawteeth are observed at only a few austenite grain boundaries. Although some impingement of primary sawteeth has occurred, it is evident that the majority form with a truncated rather than a sharp tip. (Examples may be observed along the left grain boundary above the center in Fig. 7.14A.) Aaronson and Wells have observed a ferrite/ferrite boundary between sawtooth and the allotriomorph, thereby suggesting sympathetic nucleation to be responsible for their formation.98 Intragranular Widmanstätten plates are needle- or platelike precipitates formed at large undercooling. Large austenite grain size favors the formation of intragranular plates which may form on isolated dislocations or at small-angle grain boundaries in austenite; direct determination of these nucleation sites, however, has not been achieved due to the destruction of the austenite matrix by martensite formation during quenching to room temperature. They are often grouped in starlike and more complex configurations at low transformation temperature in the proeutectoid ferrite region.99 They have a lenslike shape and appear as double-ended isosceles triangles when formed during continuous cooling (Fig. 7.14A and D). At still lower temperatures the precipitation of nonlamellar carbides converts the ferrite plates to bainitic ferrites which often appear in sheaves,100 probably arising from sympathetic nucleation.99 This ferrite is found in Fe-C alloys and in carbon-free Fe10% Cr alloy when transformed at 525°C (977°F).75 7.6.2.1 Mechanism of Widmanstätten Plate Formation. The atomistic mechanism of the formation of Widmanstätten plates has generated great interest. The important characteristic features of Widmanstätten plate formation are as follows: 1. Well-defined double-tilted tentlike or single-tilted surface-relief effects appear at Tw temperature (about 50°C higher than the calculated T0 temperature), upper limiting temperature for the composition-invariant transformation in many ferrous alloys such as ternary Fe-C-X alloy (except in alloys having high Mn or Ni contents). 2. The growth of precipitate plates exhibiting surface relief is associated with the composition variation. 3. Plates exhibiting complex surface reliefs are actually monocrystals. Recently, several studies have been performed on the thermodynamic and kinetic aspects of plate formation.101,102 TEM studies have shown that aw plates consist of a pair of two plates with a different orientation that forms in a mutually accom-

7.30

CHAPTER SEVEN

FIGURE 7.17 (a) Effect of curvature on the phase boundaries. (b) Composition profile along centerline of growing Widmanstätten ferrite plates/needles.

modating fashion. It has been reported that this pairing of plates gives rise to double-tilted surface relief. The subunit in aw exhibits the same habits with a growth direction surrounded by two sets of parallel planes close to {451}a. This suggests that aw forms via a displacive mechanism with respect to substitutional elements. If Widmanstätten plates are truly formed by displacive mechanism, the solute concentration has to be inherited from the matrix, at least initially. The ledgewise lengthening has been shown to provide a good account for reported diffusioncontrolled lengthening kinetics of ferrite plates in Fe-C and Fe-Ni alloys in a wide range of reaction temperatures and supersaturation.103 7.6.2.2 Growth Kinetics of Widmanstätten Plates. There are three important models for the growth kinetics of Widmanstätten plates. Two models for plate lengthening are called the Zener-Hillert treatment and the Trivedi treatment, while the third one for plate thickening is known as the Cahn-Hillig-Sears (CHS) model. Lengthening Kinetics of Plates. Note that when the steel is austenitized at a high temperature T1, the austenite may have a bulk composition C0 (Fig. 7.17a). On cooling to a temperature A3,1, the compositions of austenite and ferrite in equilibrium with each other are C gg a (= Cggar•) and Caag , respectively, assuming the planar, disordered interfaces (i.e., incoherent boundaries with infinite radii of curvature). As described earlier, the concentration difference, DC, within the austenite between a point near the planar growing ferrite interface and a distant point in the bulk austenite is given by DC = C ga g - C0

(7.26)

Let us assume a Widmanstätten ferrite (aw) plate growing into austenite of composition C0 (Fig. 7.17b) where the edge has a constant radius of curvature r. With the growth of aw plate, most of the carbon rejected from the tip diffuses to the sides of the thin plate. As before, if we assume the same cross-sectional area of two phases, the increased amount of carbon atoms piled up ahead of the advancing tip (denoted by area A¢) becomes much smaller than the depleted amount of carbon behind the interface (represented by area A) (Fig. 7.17b). For the curved growing plate interface, Zener18 used the local equilibrium model and took into account the pressure difference between ferrite and austenite across the plate tip and derived the following relationship:

PEARLITE AND PROEUTECTOID PHASES

(DC )tip = DC Ê 1 Ë

r* ˆ r ¯

7.31

(7.27a)

or r* Cggar - C0 = (Cgga - C0 )Ê 1 - ˆ Ë r ¯

(7.27b)

where C ggar is the actual composition of austenite at the tip interface (Fig. 7.17a) and r* is the critical radius of curvature at which the concentration difference (DC)tip is zero and growth ceases. The value of r* can be approximately determined from the following Gibbs-Thomson equation r* =

gVma RT (Cgga - C0 )

(7.28)

where g is the a /g surface tension, V ma is the molar volume of the a phase, R is the universal gas constant, and T is the absolute temperature. If we assume, in addition to the local equilibrium at the interface, that the effective diffusion distance L [in Eq. (7.20)], in front of the moving interface is proportional to the radius of curvature of the plate tip (L = kr), then the equation for the concentration gradient at the tip, on comparing Eq. (7.20), becomes ga Ê ∂ C ˆ = (Cg - C0 )(1 - r* r ) Ë ∂ r ¯ tip kr

(7.29)

This equation indicates the greater steepness of the concentration gradient for the edge of a plate (Fig. 7.17b). Substituting Eq. (7.29) in Eq. (7.19) yields the growth rate of the advancing plate G, given by G=

D(Cgga - C0 )(1 - r* r ) D = W0 kr (Cgga - Caag )kr

(7.30)

where W0 = (C gg a - C0)/(C gg a - C aag ) is the dimensionless supersaturation parameter. Equation (7.30) shows the variation of growth rate with tip radius. On differentiating with respect to r and setting ∂G/∂r = 0, we obtain the maximum growth rate G* at r = 2r*:18 G* =

D W0 4 kr *

(7.31)

Hillert104 has extended this treatment further and found k = 2. Hence the equation of maximum (or steady-state) growth rate of a Widmanstätten plate becomes G* =

D W0 8r*

(7.32)

This equation is called the Zener-Hillert equation for plate lengthening. Hillert105 further modified the above Zener-Hillert equation for the lengthening of a precipitate plate; this modification agrees well with Trivedi’s results (discussed below) for medium and high values of W0: G* =

( D 4r* )W 0 1 - W0

exp[ -5.756(1 - W 0 )]

(7.33)

CHAPTER SEVEN

7.32

The value of r* depends on the composition of the alloy and can be computed from Eq. (7.28). Figure 7.18a shows a comparison between experimental lengthening rates of edges of Widmanstätten plates and calculated lengthening rates employing Hillert and Trivedi growth equations, which suggests the superior accuracy of the Trivedi model.106 The most advanced and rigorous treatment of the volume diffusion-controlled plate-lengthening kinetics is presented by Trivedi.107 His solution is developed from Ivantsov’s treatment108 and includes the following assumptions:106,109 1. The growth of the plate maintains a constant shape, i.e., that of a parabolic cylinder. The steady-state shape of the interface near the growing tip corresponds to the parabolic cylinder for the plate morphology. 2. The elastic strain energy and anisotropy of interface properties may be ignored. 3. The concentration of solute in the parent phase is such that the theory of capillarity applicable to dilute solution can be employed. 4. The diffusivity is concentration-invariant. 5. This is valid provided the concentration variation along the interface is not very large so that the plate shape does not deviate much from that of a parabola. Trivedi obtained the following result of plate thickening: r* 12 È Gl ˘ W 0 = (pp) e perf ( p1 2 )Í1 + W 0 S1 ( p) + W 0 S2 ( p)˙ r Î Gc ˚

(7.34)

where p = Glr/2D (a dimensionless quantity, called the Peclet number), D is the diffusion coefficient of solute in the matrix, S1(p) and S2(p) are mathematically complicated functions of p and are presented graphically, Gl is the lengthening rate of a plate, Gc is the lengthening rate provided that growth was completely controlled by interfacial reaction, r is the radius of curvature at the tip of the plate, r* = radius at which growth stops by capillarity, and the other terms have the usual meanings. The right-hand side of Eq. (7.34) has three terms: The first term is the result due to Ivantsov for the case of isoconcentrate interface, and the second and third terms are correction factors to Ivantsov’s solution due to interface kinetics and capillarity effects, respectively, which grow larger with the decrease of supersaturation.110 Equation (7.34) holds for the case of isotropic energy but does not hold directly for the anisotropic case of Widmanstätten plate.111 This equation, however, can be simplified to low-to-medium values of the supersaturation W0 by the approximation due to Bosze and Trivedi.112 The growth rate equation can now be obtained by using the definition of the Peclet number as G=

8 D 2 27D W BT = W 3BT 9p r 256p r*

(7.35)

9 2 W BT 16p

(7.36)

and p=

The Bosze and Trivedi supersaturation WBT is related to the supersaturation W0 by W BT =

W0 1 - (2W 0 p ) - (W 02 2p )

(7.37)

PEARLITE AND PROEUTECTOID PHASES

Growth rate (µm/s)

100

7.33

Curves are calculated acc. to Zener-Hillert

10

1

Curves are calculated acc. to Trivedi

0.1

680°C 0

0.5 Supersaturation (Ω0)

1.0

(a) 1.0

(J–T) 10–1

(A) P 10–2

10–3

10–4

0

0.2

0.4

0.6

0.8

1.6

Ω0 (b)

FIGURE 7.18 (a) Comparison between experimental lengthening rates of the edges of Widmanstätten plates and the calculated lengthening rates using the Hillert and Trivedi growth equations.105,106 (b) Relationship between p and W0 for the growth of an isolated ledge, as predicted by Jones and Trivedi (J-T) and Atkinson (A).106, 115–116 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

CHAPTER SEVEN

7.34 Semicoherent interface

γ

Riser

Gt

Gl

λ h α FIGURE 7.19. Schematic illustration of a growth ledge on a semicoherent interface.

where r* is the critical radius of nucleation. Hillert111 has shown that Eq. (7.35) holds well with the observed rate of linear lengthening of Widmanstätten a-iron plates in Fe-C alloys. Purdy113 demonstrated similar agreement for the needlelike a-brass precipitates forming from b-brass in Cu-Zn alloys. In both situations, the precipitates grow from grain boundary nucleation sites because, for the small compositional differences between the two phases, inadequate driving force exists for nucleation within grains. Thickening Kinetics of Plates. The growth ledge mechanism involving even kinks was introduced to describe the thickening rates or kinetics aw plates. It is well agreed that the broad faces of aw plates are partially coherent with the g matrix. The migration of such interfaces must advance by a ledge mechanism, as shown in Fig. 7.19. The movement of growth ledges occurs by atom additions at the riser of the ledge. It is usually thought that the thickening of aw plates arises from the movement of growth ledges across the partially coherent interface. The dependence of the thickening rate of Widmanstätten plates Gt on the lateral migration rate of ledges (i.e., broad faces of the ledges are considered entirely immobile and the risers of the ledges are mobile) was first proposed by Aaronson73 and is expressed as:114 Gt =

hG l l

(7.38)

where h is the ledge height, Gl is the lateral velocity of a ledge, and l is the interledge spacing. Jones and Trivedi115 and Atkinson116 used different assumptions and mathematical methods to explain the diffusion-controlled kinetics of ledge movement. There are important differences in the concentration profiles around the ledges as estimated by the two approaches.106 Later they extended their treatments to incorporate the migration of multiple noninteracting and interacting ledges.106,115,116 The following relation was obtained initially by Jones and Trivedi, using the flux balance equation Gl =

DW 0 al

(7.39)

PEARLITE AND PROEUTECTOID PHASES

7.35

where D is the averaged (carbon) diffusivity in the austenite matrix, and a is the effective diffusion distance ahead of the riser, which is given by the relationship W0 = 2pa

(7.40)

Atkinson’s detailed and final treatment for the growth of a single ledge is represented by the following equation for p < 0.1: W0 =

2p (2.954 - ln p) p

(7.41)

Figure 7.18b compares the relationship between p and W0 obtained by Jones and Trivedi (JT) and Atkinson (A). The differences between the two results are negligibly small; however, Atkinson’s analysis provides faster (2 to 3 times) growth rates at low p values and remarkably smaller lengthening rate at high supersaturations than those obtained from the Jones-Trivedi treatment.103,115 Recent finite difference computer simulations of the migration of single ledges have provided more insights about the migration kinetics of the ledge growth mechanism during g Æ a + g transformation which suggest that the ledge mechanism may also play a vital role in controlling the morphological evolution of ferrite plates during growth.92 Experimental evidence of diffusional growth by ledge mechanism has been found by many workers in nonferrous systems such as for g Al-Ag2 plates in Al-Ag alloy, q ¢ Al-Cu plates, and a reactions in Ti-Cr and Ti-Fe alloys; and in ferrous systems such as a Fe-C, Fe-C-Si and Fe-C-Ni alloys in a wide range of reaction temperatures and supersaturation.103,117–119 7.6.2.3 Experimental Procedure to Determine Growth Rate. The experimental methods used to measure the lengthening kinetics of Widmanstätten plates are similar to those described earlier for pearlite and allotriomorphs. The thickening kinetics of Widmanstätten sideplates have been found to be highly variable.120 The problem in measuring is much more difficult, especially when TEM is not used.

7.6.3 Massive Ferrite 7.6.3.1 General Aspects of Massive Ferrite. Massive transformation occurs during rapid cooling and involves a change in crystal structure, similar composition of product and parent phases, and lack of any definite crystallographic orientation relationship between them.121,122 Usually, the massive transformation is observed if the competing equilibrium and other (notably, Widmanstätten) transformation modes are subdued. As a rule, it can occur in any system where: (1) the relevant phase fields overlap in composition and (2) a metastable two-phase field exists below an invariant reaction (e.g., eutectoid) temperature.123–125 Growth is accomplished primarily by noncooperative (random) transfer of atoms across the relatively highenergy incoherent product/matrix interfaces. As a result of these features, once nucleated, the product phase grows at high velocities (up to 1 to 2 cm/s) with approximately the same rate in all directions,126 consuming much of a parent grain, or sometimes several grains of the parent phase. This high growth rate puts the massive transformation in a position between diffusional transformation occurring at very slow rates (Å/sec) and the diffusionless martensitic transformation occurring at a velocity approaching the speed of sound (~105 cm/s) (Fig. 7.20).127,128

7.36

CHAPTER SEVEN

FIGURE 7.20 Schematic a TTT diagram illustrating the region of massive C-curve in between diffusional and martensitic transformations.127,128 (Reprinted by permission of Butterworths, London.)

Another important feature of massive transformation is that transformation temperature displays a “plateau” behavior (Fig. 7.21) when initial transformation temperature (i.e., thermal arrest temperature measured in a continuous cooling experiment) is plotted against the different cooling rate for high-purity iron, mild steel, and nickel steel.128,129 However, some researchers have discounted this as a fundamental feature.121 Evidence of both the presence and the absence of definite crystallographic relationships between the massive product and parent phase exists.123,124 In these situations, it has been debated that massive transformation must always involve special crystallographic relations (between the product and parent), resulting in coherent nucleation followed by growth, where the interface changes from coherent to semicoherent and moves by a ledge mechanism.124 Figure 7.22 shows a free-energy-composition curve illustrating the thermodynamic condition for a massive transformation of b to a phase by quenching from a higher-temperature stable b phase. If the transformation temperature lies below the transition T0 in the region II (i.e., with faster cooling rate than any reaction which produces long-range diffusion) and the matrix composition is lower than the critical concentration C(T0) corresponding to the temperature T0, the free energy of the system can be lowered and the b phase may transform directly to a (that is, am) of the same composition. However, if the matrix composition exceeds the critical concentration C(T0) in the region I of the two-phase field, it will transform only to solute-depleted a phase.130 However, Hillert has noted that massive transformations follow the line starting at the solvus point rather than at the line starting at the T0 point.126 That is, massive transformation requires undercooling below the solvus line. Massive transformations are thermally activated processes that exhibit diffusional nucleation and growth characteristics. The kinetics of these transformations are controlled mainly by (1) interfacial diffusion rather than volume diffusion and (2) other interface features such as lack of coherency between the product and parent phases,122,127 which permit the parent-phase grain boundary crossing. It applies to nonmartensitic composition-invariant reactions, which involve shortrange diffusion at the disordered transformation interface. This transformation does not seem to involve invariant plane-strain surface-relief effects.131 However, further

PEARLITE AND PROEUTECTOID PHASES

7.37

Temperature, C

900 Oriented nucleation and growth(?) 800

Massive Martensitic

700 0

10

20 30 40 Cooling rate, 103 C/sec

50

(a)

Transformation temperature, C

1000 900 800 700

PURE IRCN

600

Fe–1 at.%Ni Fe–2 at.%Ni Fe–3 at.%Ni

500

Fe–5 at.%Ni Fe–7 at.%Ni

400 300

Fe–10 at.%Ni 80 C/sec

5

10

15

20

25

30

35

40

45

50

55

60

Cooling rate, 103 C/sec (b)

FIGURE 7.21 Variation of transformation temperature with cooling rates: (a) pure iron128 and (b) iron and Fe-Ni alloys.129 [(a) Reprinted by permission of ASM International, Materials Park, Ohio; (b) reprinted by permission of The Institute of Metals, England.]

study of surface-relief effects associated with the massive transformations is necessary in alloys, where it is possible to control the rapid growth rate and produce a pronounced anisotropic massive morphology.132 Massive transformations occur in many ferrous and nonferrous systems during quenching within the single-phase field or two-phase field; in the latter, at least one phase must be metastable and can form simultaneously, provided that it has the same composition as the parent phase but a different crystal structure.131,133,134 Figure 7.23 shows schematic phase diagrams for pure metals and three types of alloys, illustrating phase relations which are essential for the occurrence of massive transformation.122 Table 7.3 lists typical massive transformations for pure metals and binary alloys.122 The composition in Fig. 7.23a corresponds to a high-purity metal and dilute

CHAPTER SEVEN

7.38

T β T T0

α

II

β

Cα C(T0) Cβ (a)

α

G

I

C

Cα

C(T0) Cβ (b)

C

FIGURE 7.22 (a) Phase diagram illustrating transition temperature T0 below equilibrium temperature TE. (b) Free-energy–composition diagram showing the thermodynamic condition for a massive or a diffusionless transformation of b to a phase in region II. In region I only, a reaction to give solute-depleted a can decrease the free energy.130 (Reprinted by permission of North-Holland Physics Publishing, Amsterdam.)

FIGURE 7.23 Schematic phase diagram for (a) pure metal and (b–d) three types of alloys, all of which may undergo massive transformations. Critical compositions are indicated by the dashed vertical lines.122 (Reprinted by permission of ASM International, Materials Park, Ohio.)

binary alloy exhibiting a polymorphic transformation. In this case the long-range diffusion of the impurity and solute partitioning corresponding to the critical composition line do not occur during rapid cooling through the two-phase region, which facilitates the massive transformation. Examples are iron, low-carbon steels, and low-nickel steels. Figure 7.24a shows the resulting microstructure in Fe-0.002% C alloy. The critical composition in Fig. 7.23b corresponds to an alloy in which the bcc phase transforms to massive hcp phase on cooling through a congruent point (a junction of two-phase fields). It is similar, in several ways, to a polymorphic element. Examples are

PEARLITE AND PROEUTECTOID PHASES

7.39

TABLE 7.3 Typical Massive Transformations for Pure Metals and Binary Alloys121

Alloy system or metal Silver-aluminum Silver-cadmium Silver-zinc Copper-aluminum Copper-zinc Copper-gallium Iron Iron-cobalt Iron-chromium Iron-nickel Plutonium-zirconium

Amount of solute at which transformation occurs (at%)† 23–28 41–42 50 37–40 19 37–38 21–27 20 — 0–25 0–10 0–6 5–45

Temperature during quenching at which transformation occurs† °C 600 300–450 300 250–350 550 400–500 580 600 700 650–800 600–800 500–700 450

°F 1110 570–840 570 480–660 1020 750–930 1075 1110 1290 1200–1470 1110–1470 930–1290 840

Change in crystal structure‡ bcc Æ hcp bcc Æ fcc bcc Æ hcp bcc Æ fcc bcc Æ fcc bcc Æ fcc bcc ∫ hcp bcc Æ fcc fcc Æ bcc fcc Æ bcc fcc Æ bcc fcc Æ bcc bcc Æ fcc

†

Values listed are approximate. bcc, body-centered cubic; fcc, face-centered cubic; hcp, hexagonal close-packed. Reprinted by permission of ASM International, Materials Park, Ohio. ‡

FIGURE 7.24 (a) Massive ferrite with irregular grains in Fe-0.002% C alloy after quenching in iced brine from an austenitizing temperature of 1000°C (1832°F), 385X (reduced 57%). (b) Partial b Æ a massive transformation in Cu-19.3 at % Al which initially consisted of an equilibrium twophase mixture of a and b phases. (c) Massive hcp phase grains in a quenched Cu-18.4 Ga-5Ge (at %) alloy which are able to cross prior b (bcc) boundaries. 60X (reduced 57%).122 (Courtesy of T. B. Massalski.)

7.40

CHAPTER SEVEN

1. b (bcc) Æ xm (hcp) transformations occurring massively in Al-24.5 at% Ag alloy and Cu-23.7 at% Ga alloy 2. x Æ x¢ in Ag-27 at% Ga alloy. The composition in Fig. 7.23c corresponds to the decomposition of the hightemperature bcc phase on cooling through the two-phase field with an entire or partial transformation to the massive fcc phase. Examples are b (bcc) Æ massive xm (fcc) transformation in Cu-Ga rich in Cu and bcc b Æ massive am (fcc) in the Cu-Zn, Cu-Al, Ag-Cd, and Ag-Zn systems. Figure 7.24b is a partial, b Æ massive transformation in Cu-19.3 at% Al which initially consisted of an equilibrium twophase mixture of a and b phases. The composition in Fig. 7.23d corresponds to an alloy where bcc phase transforms to massive hcp phase on quenching through the two-phase field. Examples are (1) b (bcc) Æ am (hcp) transformation in Ti-Ag, TiAu, and Ti-Si eutectoid systems135 and (2) b (bcc) Æ xm (hcp) transformation in Al26 at% Ag, Cu-Ga, and Ag-Cd systems. Figure 7.24c shows the massive hcp phase grains in a quenched Cu-18.4Ga-5Ge (at%) alloy which are able to cross prior b (bcc) boundaries.122 Mechanism. Based on experimental evidence, in some cases, and theoretical consideration, it may be concluded that, during the nucleation stage of a massive phase, some form of a rational (i.e., with low Miller indices) or nearly rational orientation relationship with one of the matrix crystals seems necessary, in order to lower the activation free-energy barrier for critical nucleus formation DG*. This greatly increases the formation rate of critical nuclei to the extent that critical nuclei enclosed by a smaller proportion of low-energy interface cannot successfully compete and therefore are not experimentally observed.136 In this situation, small areas of their interfaces can be coherent with the matrix crystal which adopts the least coherent growth mode, while the remaining areas of their interfaces, being disordered and incoherent, have high mobility for rapid massive growth.124,132 It appears that the movement of the incoherent boundary is hindered by solute drag. That is, the boundary is pinned by solutes at various points along its length, giving rise to irregular grains with ragged boundaries.137 7.6.3.2 Massive (Ferrite) Transformation in Ferrous Alloys. Ackert and Paar138 have shown that the temperature of massive transformation is sensitive to the interstitial concentration. For massive transformation in ultrapure iron, solute drag does not appear; however, the strain energy DGe due to volume change will exist. The g Æ a transformation in pure iron and in low-carbon (with ~0.002 wt% C) and low-nickel steels can occur massively. The austenite is quenched with a higher cooling rate to avoid the transformation with long-range diffusion (near the equilibrium), but with a slow enough rate to avoid the diffusionless martensite-type transformation.129,139 Massive ferrite can only occur below T0 (Ae3 < T0 < Ae1) but above Ms temperature. (see Fig. 7.22, where b is considered as a parent phase). Figure 7.24a shows the characteristic microstructural appearance of massive patches of ferrite with irregular boundaries. Figure 7.21a shows the plot of temperature versus cooling rate for high-purity iron; the graph exhibits an arrest plateau at about 740°C, identified with massive transformation at cooling rates between 5000 and 35,000°C/s, depending upon composition. Such a distinct plateau may be considered as a manifestation of a special nucleation event.123 Another plateau is found at a lower temperature (~690°C) when the cooling rate exceeds ~35 ¥ 103°C/s, corresponding to the onset of martensitic transformation.128 Figure 7.21b represents the results obtained by Swanson and Parr,129 which demonstrate that both the higher

PEARLITE AND PROEUTECTOID PHASES

7.41

FIGURE 7.25 Optical micrograph of carbon steel quenched at 913 K during cooling from 1523 K. The arrow indicates a typical intragranular ferrite idiomorph. (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

plateau massive temperature and critical cooling rates required to suppress the massive transformation are strongly dependent on the nickel content.105 Their data indicate that the massive transformation does not occur beyond the Fe-7 at% Ni alloy, but Massalski et al. have reported the observation of massive transformation in Fe-8.7 at% Ni alloy.140,141 Later Hayzeldon and Cantor142 observed massive ferrite in melt-spun Fe(up to 25 wt%)-Ni. They attributed this to be due to increased nucleation rate at many austenite grain boundaries and grain corners in the very fine austenite grains of size 3 mm. Recently Chong et al.143 have reported the formation of both the massive ferrite and equiaxed ferrite in the two-phase field in Fe-9.14 wt% Ni alloy when furnace (or continuously) cooled from 1000°C to the temperature range of 575 to 558°C (i.e., below T0 and A3 temperatures). They tend to nucleate at grain corners and grain boundaries with at least one coherent boundary. TEM examination showed occasional massive ferrite grains with bulged boundaries of heavy dislocation density and equiaxed grains associated with variable dislocation density. They attributed the presence of paraequilibrium during transformation to massive ferrite and equiaxed ferrite.143 7.6.4 Idiomorphic Equiaxed Ferrite Idiomorphic ferrite can form intragranularly (within the austenite grains) or at austenite grain boundaries at small undercooling (Figs. 7.14C and 7.25). These idiomorphs are thought to be nucleated heterogeneously at nonmetallic inclusions and dislocations. They exhibit a roughly equiaxed morphology. In medium-carbon vanadium steels, intragranular ferrite idiomorphs nucleate at vanadium nitride precipitates which, in turn, form at MnS particles during cooling in the austenite matrix. It is observed that intraganular ferrite idiomorph has the Baker-Nutting orientation relationship (B-N OR) with VN which precipitated at MnS. In the B–N OR, the lattice mismatch across the (001)a // (001)VN atomic habit planes is probably very small.144

7.7 PROEUTECTOID CEMENTITE Proeutectoid cementite morphologies obtained on isothermally transforming hypereutectoid steels are quite similar to those discussed above for proeutectoid

7.42

CHAPTER SEVEN

ferrite, which include grain boundary precipitates; long, thin Widmanstätten sideplates; intragranular Widmanstätten sideplates; and intragranular idiomorphs.

7.7.1 Cementite Morphology Grain boundary allotriomorphs usually form at high temperatures in hypereutectoid carbon and alloy steels. The allotriomorphic cementite phases nucleate with a Pitsch relationship to one austenite grain and grow into the adjacent grain with which they have no orientation relationship. It seems that the interfacial structures of the allotriomorphic carbide phases in steels are similar to allotriomorphic ferrite and other interfaces in both steels and some nonferrous alloys.48,89 The volume fraction of these thin-film carbides decreases with increasing grain size and increasing cooling rate. In high-purity Fe-C alloys and plain carbon steels, Heckel and Paxton145 found the thickening growth rates of cementite allotriomorphs that were an order of magnitude lower than the expected equilibrium thickness based on carbon-diffusion-controlled growth. The thickening kinetics were observed to be insensitive to grain size and transformation temperature. It was suggested that in high-purity Fe-C alloys the thickening of cementite allotriomorphs occurred presumably by a diffusion-interface-controlled growth mechanism and that in plain carbon steels the thickening took place by silicon-diffusion-controlled growth.145,146 The morphology of cementite sideplates is usually similar to that of ferrite sideplates. The primary cementite sideplates nucleate at only a few austenite grain boundaries. Most cementite sideplates are of the secondary type rather than of the primary type. Both primary and secondary sideplates form at twin boundaries. Decreasing the transformation temperature (1) increases the tendency to form cementite sideplates, until interrupted at lower temperature by the pearlite or bainite reaction (7.16b), (2) increases the average length/width ratio, and (3) decreases the average sideplate spacing. Increasing the austenite grain size under given conditions of carbon content and reaction temperature increases the tendency to form more cementite sideplates. Intragranular cementite plates usually nucleate shortly after cementite sideplates; and, in this morphology, apart from the increase in the average length/width ratio, there is not much variation with the transformation temperature. Intragranular Widmanstätten cementite plates also occur at lower heat treatment temperatures and are observed to follow the Pitsch orientation relationship with the austenite matrix.48 The main difference between sideplate and intragranular Widmanstätten cementites which form at lower temperature is that the former can develop from the grain boundary allotriomorphs, while the latter precipitates within g grain by sympathetic nucleation and growth.73,147 In alloy steels (such as Fe-0.96C-12.85Mn-0.78Si steel), a sawtooth morphology of fcc M23C6 (M = Fe, Mn) carbide forms at the grain boundaries and exhibits a cubecube orientation relationship with respect to one of the g grains (A2) into which it grows. These findings resemble the recent work on high-Mn steels. The formation of facets and macroscopic steps at the growth interface of the grain boundary carbide was also reported. Planar faults noticed in the carbide are recognized as stacking faults lying along {111} planes, and these are associated with linear features at the planar interfaces.48

PEARLITE AND PROEUTECTOID PHASES

7.43

7.8 FERRITE-PEARLITE AND PEARLITIC STEELS The most popular microstructure in the context of structural steel has been a mixture of ferrite and pearlite. These steels are usually characterized by their higher strength compared to low-carbon steels in both normalized and heat-treated conditions. Here formulations of quantitative relationships between microstructural parameters and mechanical properties, particularly yield or flow stress sy, tensile or ultimate tensile strength sT, impact transition temperature (ITT), and Charpy shelf energy (CSE) of steels containing ferrite and pearlite are described. In the following equations, quantities of dimensions of length are expressed in millimeters, stress in megapascals (meganewtons per square meter), energy in joules, and temperature in degrees Celsius. Important applications including rail steels, wire rod steels, wire ropes, bridge ropes, spring wires, tire bead wire and tire cord, wire for prestressed concrete, cold heading wire, and microalloyed forging steels are briefly described.

7.8.1 Structure-Property Relationships 7.8.1.1 Ferrite in Low-Carbon Ferrite-Pearlite (or C-Mn) Steels. The typical empirical strengths and impact (or ductile-brittle) transition temperature of C-Mn steels up to 0.25%C are given by148–150 s y (MPa) = K + 37(%Mn ) + 83.2(%Si ) + 2918(% N f ) + 15.1d -1 2

(7.42)

s T (MPa) = 294.1 + 27.7(%Mn) + 83.2(%Si) + 3.9(% pearlite) + 7.7d -1 2 (7.43) ITT( ∞ C ) = -19 + 44(%Si) + 700(%N f )

12

+ 2.2(% pearlite) - 11.5d -1 2

(7.44)

Equation (7.42) can be used to plot the yield strength as a function of ferrite grain size, as seen in Fig. 7.26a.148 In the above equations, K is 88 MPa for normalized steel and 62 MPa for fully annealed (i.e., furnace-cooled) steel; Mn and Si are elements dissolved substitutionally in the ferrite solid solution; Nf is the free nitrogen content dissolved interstitially in the ferrite lattice (i.e., not present as a stable nitride); and d is the average linear intercept in polygonal ferrite. Both Mn and Si increase the yield and tensile strengths by solid solution strengthening of ferrite. Pearlite does not contribute significantly to the yield strength unless it is present in large amounts; but carbon lowers the transformation temperature, thereby decreasing the ferrite grain size. Grain refinement increases yield strength and toughness (i.e., lower ITT) while the strengthening without the grain refinement still reduces the toughness. Thus, an increase in pearlite content increases the tensile strength and ITT and decreases the CSE value. Figure 7.27 shows the exponential decrease of upper shelf energy Cv with the increase in pearlite fraction.155 All solid solution strengtheners except Mn and Ni raise the ITT. Mn and Ni lower the ITT by 30 and 13°C per wt%, respectively. The behavior of Al is different. In small amounts, it lowers ITT by eliminating nitrogen from the solution. Similar equations for yield and tensile strengths have been listed by Baird and Preston.151 However, one of the more sophisticated equations for yield strength, relating to microstructure for a plain C-Mn steel, is152

CHAPTER SEVEN

7.44

Predicted yield stress (MN M–2)

400

300 Ferrite grain size 200 Nitrogen Silicon Manganese

100

Friction stress

0

5 10 15 Ferrite grain size d–1/2 (mm–1/2) (a)

400 ed serv

yield

s

stres

Yield stress (MN m–2)

Ob

Pearlite

300

200

Ferrite grain size

100

Manganese + friction stress Silicon + nitrogen

0

0

0.2

0.4

0.6

0.8

Volume fraction of pearlite (b)

1.0

FIGURE 7.26 (a) Effect of ferrite grain size on the yield stress of normalized 0.2% C-1.0% Mn-0.2% Si0.01% N steel.148 (b) Contributions of various strengthening mechanisms to the yield stress of plain carbon steels.148 (Reprinted by permission of Pergamon Press, Plc.)

s y (MPa) = 70 + 32(%Mn ) + 84(%Si ) + 680(%P) - 30(%Cr ) + 33(%Ni ) + 11(%Mo) + 38(%Cu ) + 5000(%N f ) + 18.1d -1 2

(7.45)

It is clear from the above expressions that the grain size of steel must be controlled if consistent mechanical properties are required. 7.8.1.2 Pearlite in Ferrite-Pearlite (C-Mn) Steels. Regression analysis on a wide range of medium- to high-carbon steels by Gladman et al.72 has demonstrated that the yield strength sy, tensile strength sT, and impact transition temperature ITT can

PEARLITE AND PROEUTECTOID PHASES

7.45

FIGURE 7.27 Effect of the pearlite fraction on the shelf energy in normalized ferritepearlite structures.155 (Reprinted by permission of VCH, Weinheim.)

be related to the compositional and microstructural variation by sy = f nsa + (1 fn)sp, where sy is the yield stress of the aggregate (in megapascals), f is the volume fraction of ferrite, sa is the yield strength of the ferrite (in megapascals), sp is the yield strength of the pearlite (in megapascals), and n is an index representing the nonlinear contributions of the ferrite and pearlite. Both yield strength and tensile strength showed an index of n = –13 , suggesting that the ferrite fraction contributes more to the yield and tensile strength than would be expected on a pro rata basis. On the other hand, the ITT showed an index of n = 1, suggesting a simple law of mixtures.72 The following yield strength sy and tensile strength sT relationships have been found to be applicable for ferrite-pearlite steels with pearlite content ranging between 20 and 100% (eutectoid composition):32 s y (MPa) = f a1 3[35 + 58.5(%Mn) + 17.4d -1 2 ] + (1 - f a1 3 )(178 + 3.8S -1 2 ) + 63.1(%Si) + 425(%N f )

[

s T (MPa) = f a1 3 246 + 1142(%N f )

12

12

(7.46)

]

+ 18.2d -1 2 + (1 - f a1 3 )

(719 + 3.56S -1 2 ) + 97(%Si)

(7.47)

where fa is the volume fraction of ferrite, d is the average linear ferrite grain size intercept (in millimeters), and S is the pearlite interlamellar spacing (in millimeters). The index n = –13 in the volume fraction sa in both expressions for sy and sT is used to show the nonlinear variation of yield and tensile strengths with ferrite and pearlite contents. This also suggests that the ferrite fraction contributes more to the yield and tensile strengths than would be expected on a pro rata proportion basis. As the volume fraction of pearlite increases, the influence of ferrite grain size on yield strength becomes progressively less significant (Fig. 7.26b). As the pearlite content approaches 100% (eutectoid composition), the pearlite becomes the main contributor to the strength; this is controlled by S. The contribution to the yield strength Dsy from the interlamellar spacing can be predicted from the regression equation as:153 Ds y (MPa) = 3.86S -1 2 (1 - fa1 3 )

(7.48)

CHAPTER SEVEN

7.46

Increased yield strength can be achieved by an increased cooling rate from the austenitizing temperature so that pearlite forms at a lower temperature, resulting in smaller values of S. Also, for a fully pearlitic structure Eq. (7.46) reduces to s pe (MPa) = 178 + 3.8S -1 2 + 63(%Si ) + 425( %Nf )

12

(7.49)

The contributions of other strengthening mechanisms may be added. The notch impact transition temperature in ferrite-pearlite steels with increased pearlite volume fraction can be represented by the following regression equation: ITT(∞ C) = fa ( -46 - 11.5d -1 2 ) + (1 - fa )[ -335 + 5.6S -1 2 - 13.3 p-1 2 + (3.48 ¥ 106 )t ] + 49(%Si ) + 762(%Nf )

12

(7.50)

where ITT is the Charpy V-notch 50% FATT (°C), p is the pearlite colony size (i.e., mean linear intercept) (in millimeters), and t is the pearlitic carbide lamellar thickness (in millimeters); other symbols remain the same as in Eq. (7.46). Since in most commercial steels d-1/2 varies from 4 to 10 mm-1/2, Eq. (7.50) illustrates that ITT decreases with the CSE (impact toughness) and decreases with the increased pearlite volume fraction. A small pearlite colony size lowers the ITT.154 Again, other factors involving the influence of solid solution, precipitation strengthening, and dislocation strengthening on ITT may be added. Also, for a fully pearlitic structure Eq. (7.50) reduces to ITT(∞ C) = -335 + 5.6S -1 2 - 13.3 p-1 2 + (3.48 ¥ 106 )t

(7.51)

again with requisite terms for other strengthening effects. Since the detailed pearlite morphology has a significant effect on ITT, we can incorporate a term for the prior g grain size D. In this situation, the ITT for a pearlitic structure is given by Krauss, Hyzak, and Bernstein as ITT(∞ C) = 218 - 0.83 p-1 2 - 2.98D-1 2

(7.52)

A decrease in interlamellar spacing adversely affects the CSE (impact toughness) because of the increased strength, but a decrease in pearlitic cementite plate thickness improves it. That is, S and t act in opposite directions (as shown in Fig. 7.28);155 therefore, a compromise should be made for an optimum interlamellar spacing which will produce the best impact toughness (i.e., lowest ITT) properties. 7.8.2 Applications 7.8.2.1 Rail Steels. Since the advent of railways, rail steels have been challenged by the continuous increase of wheel loads on rails with the increasing speed and tonnage of traffic.156 Rail steels should fulfill five major properties, viz., plastic deformation (including corrugations) resistance, wear resistance, fatigue (comprising both the rolling contact and the internally initiated type) resistance, residual stresses, and weldability.157 Presently, high-volume rail steels have been based on fully pearlitic microstructures which are characterized by high strength, fatigue resistance, and fracture toughness. Failures in rail track are commonly linked to fatigue cracks of various types, while the rail life depends on the extent of wear and loss of the rail profile.158 The traditional methods for extended rail life include alloying and/or heat treatment and lubrication of the wheel/rail interface, especially the gauge face/flange

PEARLITE AND PROEUTECTOID PHASES

7.47

FIGURE 7.28 The contributions of the pearlite interlamellar spacing S and the carbide lamella thickness t to the ductilebrittle transition temperature, illustrating the occurrence of an optimum value of interphase lamellar spacing.155 (Courtesy of F. B. Pickering.)

interface. Both strengthening and lubrication improve wear and deformation resistance while strengthening provides an additional advantage of enhancement in fatigue resistance.159 However, rail life is determined in millions of gross tons (MGT) of traffic, and current rail life is in excess of 250 MGT. The average rail life is 70 years. It is extremely expensive to replace rails. The wear life of rail in curved track is so heavily dependent upon curvature and lubrication that it is meaningless to mention a life in MGT. However, in tangent track, rail life is mostly controlled by fatigue, and heavy-haul lines can yield rail service lives in excess of 1500 MGT.160 The standard axle load for most heavy-haul railroads is now 33 tons with a transition toward 39 tons. Additionally, significant roles of greater track utilization, better lubrication of the wheel/rail interface, and use of higher-strength rail steels have been emphasized. Consequently, the rail wear has decreased while surfaceinitiated rolling contact fatigue (RCF) has become more prominent. The combination of high axle loads and adequate lubrication of wheels and rails during curving is of special relevance.161 Wear resistance of a rail steel is directly related to both hardness and interlamellar spacing, as shown in Fig. 7.29. Thus, the pearlitic interlamellar spacing, which is a function solely of transformation temperature, becomes the most important microstructural parameter to control hardness and wear resistance.162 A pearlitic microstructure with a finer interlamellar spacing and thinner cementite lamellae produces a higher wear resistance. Table 7.4 lists the values of the bulk average true interlamellar spacing St and hardness for four fully pearlitic rail steels. Clearly, the hardness increases with the decreasing St and, in fact, is a function of St-1/2. In general, hardness is expressed in the form of the Hall-Petch relationship as:163 Hardness(HV, kg /mm 2 ) = 150(kg /mm 2 ) + 2.15(kg /mm 3 2 ) ¥ S t-1 2 (mm -1 2 )

(7.53)

An increase in hardness increases both wear and fatigue resistance. Refinement of pearlitic interlamellar spacing is usually limited to a maximum hardness of about Rc 38 to 40. But fatigue resistance seems to improve with hardness as one changes from pearlitic microstructure below Rc 40 to bainitic microstructure above Rc 40.160,161

CHAPTER SEVEN

7.48

Weight loss, g

1.6

1.2

0.8

0.4

0 200

225

250

275

300

325

350

375

Brinell hardness, HB (a)

Weight loss, g

1.6 1.2 0.8 0.4 0 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 Pearlite spacing, µm (b)

FIGURE 7.29 Relationships between (a) hardness and wear resistance (weight loss) for rail steels and (b) pearlite interlamellar spacing and wear resistance (wt% loss) for rail steels.162 (Courtesy of B. L. Bramfitt.)

The desire to avoid premature failure, to increase resistance to surface damage, and to meet the need for improvement in straightness and line geometry has led to the requirement of tighter control for chemistry, inclusion types, segregation, blister, pipe, hydrogen, and surface condition of rails. Rails containing these improvements would be flatter, straighter, longer, and harder.164 Head Hardening of Rail. To provide a rail steel with the highest hardness and wear resistance, a head-hardening process is commonly practiced which is off-line reheating process or in-line heat treatment. Off-line head hardening is a two-stage controlled induction heating process (comprising preheating and reaustenitizing the rails after rolling and cooling to room temperature) followed by simply an accelerated cooling process using forced-air cooling. This process is slow and costly (when compared to the in-line process). The in-line head hardening utilizes the rail exiting the rolling mill and controlled accelerated cooling based on water spray system, air blasting, or an aqueous polymer solution to produce a fine pearlitic microstructure

PEARLITE AND PROEUTECTOID PHASES

7.49

TABLE 7.4 Values of the Mean True Interlamellar Spacing and Hardness for Rail Steels163

Steel 1 1 1 2 2 2 2 3 2 2 3 3 2 4 3 3 4 4 3 4 4 4 † ‡

Hardness (HV 10 ± 5%)

S¯t (nm)†

S¯t-1/2 (mm-1/2)‡

276 280 285 306 316 325 330 330 335 342 342 350 360 360 365 368 370 378 380 390 398 405

226 ± 20 200 ± 20 204 ± 20 185 ± 13 163 ± 15 178 ± 15 123 ± 13 164 ± 10 145 ± 10 111 ± 12 145 ± 10 126 ± 10 109 ± 13 142 ± 10 118 ± 10 87 ± 12 92 ± 8 106 ± 5 77 ± 10 84 ± 5 77 ± 5 68 ± 5

66 71 70 74 78 75 90 78 83 95 83 89 96 84 92 107 104 97 114 109 114 121

S¯t = 0.5S¯t is the mean true interlamellar spacing. S¯t-1/2 is the reciprocal root of the true interlamellar spacing.

in the rail.164–167 In this process, the head of the as-rolled rail is held (to about 1000°C) for several minutes and is stood head up. Only the head of the rail is allowed to develop a fully, fine pearlitic microstructure with the highest possible hardness (up to 425 HV at the near-surface position at the top and side of the head, down to 280 HV in the more slowly cooled parts of the head, web, and flange with coarse pearlitic microstructure) and wear resistance. This is accomplished, in both processes, by rapidly cooling the rail from the austenite condition to fully transform it to pearlite: (1) at 550 to 600°C (930 to 1100°F) for which a cooling rate up to 1100°C/min is needed or (2) by interrupted cooling at 570°C (i.e., at or near the pearlite transformation temperature), according to the continuous cooling transformation (CCT) diagram.168,169 Figure 7.30a shows the CCT diagram of a rail steel with a cooling rate of 900°C/min (the heavy dashed cooling curve in Fig. 7.30b) with the resulting microstructure consisting of a mixture of pearlite, bainite, and martensite (which causes less than optimum wear properties). Figure 7.30b shows the same CCT diagram illustrating interrupted cooling at about 570°C producing 100% pearlitic microstructure.166 The head-hardened rail thus has a fine pearlite structure, high toughness, and high-wear-resistance head, while the web and flange have the comparable toughness characteristics but lower strength than the standard grades of rail steel. The advantage of head hardening over full hardening is that the rail can be bent, roller-straightened, and drilled more readily.160

(a)

FIGURE 7.30 (a) Continuous cooling transformation (CCT) diagram of a rail steel and (b) same CCT diagram showing uninterrupted (dashed line) and interrupted (solid line) cooling paths.166 (Courtesy of B. L. Bramfitt, R. L. Cross, and D. P. Wirick.)

7.50

PEARLITE AND PROEUTECTOID PHASES

7.51

TABLE 7.5 Rail Steel Grades—Chemical Compositions and Tensile Strengths158 Chemical composition (wt%) Steel grades

S (max.)

P (max.)

Cr

C

Mn

Si

0.40–0.60 0.45–0.60

0.80–1.25 0.95–1.25

0.05–0.35 0.05–0.35

0.050 0.050

0.050 0.050

— —

—

680–730 710 (min.)

0.60–0.80 0.55–0.75 0.65–0.75 0.50–0.70

0.80–1.30 1.30–1.70 0.80–1.30 1.30–1.70

0.10–0.50 0.10–0.50 0.05–0.50 0.05–0.50

0.040 0.040 0.050 0.050

0.040 0.040 0.050 0.050

— — — —

— — — —

880–1030 880–1030 880 (min.) 880 (min.)

Highly wear-resistant— special alloyed (typical analysis and strength) S 1000 0.75 1.00 0.90 S 1100 0.75 1.00 0.70 S 1200 0.75 1.00 0.80

— — —

— — —

— 1.00 1.00

— — 0.10

980–1120 1080–1220 1180–1280

Highly wear-resistant— heat-treated (typical analysis and strength) Heat-treated 0.75 0.90

—

—

—

—

1120–1320

Standard UIC 700 (1986) BS 11 (1978) Wear-resistant UIC 900 A (1986) UIC 900 B (1986) BS 11 A (1978) BS 11 B (1978)

0.25

V

Ultimate tensile strength (MPa)

Rail Steel Specifications. Modern rail steels currently used in Europe and Asia are listed in Table 7.5, and broadly classified into four types according to tensile strength:158, 170 1. Standard grades, ~700-MPa minimum tensile strength 2. Wear-resistant grades, 880-MPa minimum tensile strength 3. Highly wear-resistant grades, 1080 to 1200 MPa 4. Other heat-treated high-strength grades, 1120 to 1320 MPa The tensile strength of these steels can be related to the chemical composition by an empirical relation of the form, due to Kouwenhoven, which is different from Eq. (7.47): s T (MPa) = 227 + 803(%C) + 87(%Si) + 115(%Mn) + 133(%Cr) + 891(%P) + 614(%V)

(7.54)

Equation (7.54) is useful in predicting the effects of alloy additions on the tensile strength of typical rail steel products cooled within a specified range of natural air cooling rates. 1. Standard grades. Typical examples of standard grades include BS 11:1985, Normal and UIC Grade 70. These are high-tonnage grades which are employed in normal service conditions in traditional railways, including high-speed passenger traffic (200 km/hr) and medium-speed (100 km/hr), relatively heavy-axle-load (25ton) freight.

CHAPTER SEVEN

7.52

2. Wear-resistant grades. The hardness and wear resistance of pearlitic steels are increased by refining the pearlite lamellae. This is carried out by increasing the C and Mn concentrations, both of which decrease the transformation temperature from austenite to pearlite. These wear-resisting grades are employed for heavy-axle loads, high-density traffic routes, or tightly curved track. However, the use of wearresisting rails on conventional railways can also engender economic benefits. 3. High-strength grades. For extremely difficult service conditions faced in tightly curved track and under very high axle loads, even higher strengths and further refinement of the pearlitic structure with satisfactory level of hardenability and weldability are needed. This is achieved by head hardening of rails. Increasing the cooling rate of the railhead results in a refinement of the interlamellar spacing of the pearlite and an increase in tensile strength. 4. Other heat-treated grades. These grades include Fe-0.75C-0.90Mn-0.25Si, more alloyed steels such as Fe-0.7C-1.25Mn-0.1Si-1.0Cr, and CrMo steel (Fe-0.71C0.59Mn-0.41Si-0.57Cr-0.21Mo), and microalloyed rail steels containing V or Nb to achieve 1035 to 1140 MPa (or 170 ksi minimum) of tensile strength.170 Austenitic manganese (or Hadfield’s) steel containing 1.2% C and 12% Mn is used in railway trackwork at frogs, switches, and crossings, where wheel impacts at intersections are very severe.170 American Rail Steel Specification. Table 7.6 lists the chemical composition of the standard and alloy (or premium) rail steels according to the AREA 1996 specification. The standard rail has the minimum 300 BHN hardness and minimum TABLE 7.6 Chemical Analysis of Rail Steel (AREA: 1996)171 Product analysis, weight percent allowance beyond limits of specified chemical analysis

Chemical analysis, weight percent Element

Minimum

Maximum

Under minimum

0.72 0.80 (Note 1) — — 0.10

0.82 1.10 (Note 1) 0.035 0.037 0.60 (Note 1) (Note 1) (Note 1) (Note 1)

0.04 0.06

0.04 0.06

— — 0.02

0.008 0.008 0.05

Carbon Manganese Phosphorus Sulfur Silicon Nickel Chromium Molybdenum Vanadium

Over maximum

(Note 1): The manganese and residual element limits may be varied by the manufacturer to meet the mechanical property requirements as follows:

Manganese Minimum

Maximum

0.60 1.11

0.79 1.25

Nickel Maximum

Chromium Maximum

Molybdenum Maximum

Vanadium Maximum

0.25 0.25

0.50 0.25

0.10 0.10

0.03 0.05

Fatigue strength in MPa

PEARLITE AND PROEUTECTOID PHASES

7.53

as-rolled 400

200 corroded

0 600

800 1000 Tensile strength in MPa

1200

FIGURE 7.31 The relationship between bending fatigue strength and the tensile strength of a rail steel.158 (Courtesy of T. Gladman.)

140 ksi tensile strength which lies between wear-resistant grade 2 and highly wear-resistant grade 3 of Table 7.5. The high-strength, Cr-Mo-V alloy and heattreated rail steel lies between highly wear-resistant grade 3 and other heat-treated grade 4 of Table 7.5 and has the 341 to 388 BHN range.171 Recent tests of North American rail suggest that the ratio of endurance strength to tensile strength is greater than 0.4 at the high strength and contrary to the results shown in Fig. 7.31. Life tests indicate that fatigue life is proportional to hardness raised to about the fifth power. The endurance strength of electric flash butt welds approaches that of rail, and the North American welding practices would yield a much higher endurance strength of about 325 MPa (51 ksi) as observed by G. Fowler in 1976.160 To improve fatigue life, cleaner steels (fewer aluminum- or silicon-based oxide inclusions) are required but fracture toughness could not be improved sufficiently in a fully pearlitic steel even with finer interlamellar spacings. To accomplish this, a change from a fully pearlitic microstructure to a bainitic microstructure was made. Rail producers around the world are now studying high-strength bainitic rail steels. A German rail manufacturer has an experimental steel rail in track. Note that over the last 15 years, the introduction of clean steel practices has vastly improved the metallurgical cleanliness (both oxides and sulfides) of rail steels—and this has increased the fatigue (particularly internal fatigue) performance of rails. Usually oxide inclusions promote the development of shells (internally) which then cause a crack to turn from a transversal into a detail fracture, while sulfide inclusions favor dry wear and, in combination with oxide, play a vital role in the development of vertical split head cracks.160 In general, tougher rail being expensive, North American rail steels need not be very tough; 30 to 35 ksi in . is usually adequate. Toughness is a concern if it involves roller-straightened chromium rails (German-made) where toughness is down near 25 ksi in . These rails are prone to catastrophic web cracking.160 Rail Wear. A typical rail profile exhibiting the main regions of rail wear is shown in Fig. 7.32. The rail wear can be described in terms of lateral wear, vertical wear, or the area of rail removed. With both lateral and vertical wear, the extent of wear is directly dependent on the accumulated tonnage supported. Particularly, the lateral wear can be related to the local track curvature, a small radius of curvature inducing a higher wear rate. However, an adequate scatter of wear rate has been

7.54

CHAPTER SEVEN

FIGURE 7.32 Rail wear exhibiting (a) the relationship between the tensile strength and rail wear in square millimeters per 100 million gross tons (MGT) of traffic and (b) a typical railhead wear profile.158 (Courtesy of T. Gladman; after C. Esvald.)

observed from one rail position to another based on the local environmental and geometric features of the track, axle loads, and wheel flange lubrication.172 Dry wear rate of rail steel is a function of pearlitic interlamellar spacing, cementite lamellar fragmentation, MnS inclusion content, chemical composition, and strength parameters. Note that thin cementite lamellae are most ductile and remain relatively intact after deformation.173 Sulfide inclusions are linked also to ductility exhaustion-related processes.159 The dry wear resistance capabilities of commercially available pearlitic rail steels have achieved a maximum limit with an interlamellar spacing near 1000 nm. But the wider use of effective lubrication and the introduction of bainitic steels in some cases can attain the wear demands placed on rail by heavier wheel loads.174 The wear rate can evidently be affected by the tensile strength (or hardness) of the rail material. For a given rail position, the rail head wear rate can decrease from 200 mm2/100 MGT for a rail with 900-MPa tensile strength to 40 mm2/100 MGT for a rail with 1200- to 1300-MPa tensile strength. It is a standard practice to use hightensile-strength rails in tracks with a small radius of curvature (less than 200-m radius), e.g., in curves and turnouts and in track carrying high axle loads; in this manner, it is possible to acquire rail lives in these positions exceeding 500 MGT.158 Rail Corrugation. The formation of rail corrugations is considered to be a periodic wear process which is started by rail roughness and wheel/rail creepage. Current theories to describe the formation of corrugations fall into two categories: large creepage and small creepage. The large creepage is most desirable in heavily curved track; the small creepage is suitable for straight track and, therefore, is likely to be most important for future high-speed railways. Lateral creepage, which

PEARLITE AND PROEUTECTOID PHASES

7.55

appears to be the most damaging form of creepage, is caused by the wheelset with an angle of yaw.175 Residual Stresses, Grinding, and Catastrophic Failure. Residual stresses in rails play a key role which affects the rail performance in modern railway service. Ground rails during service may develop transverse fatigue cracks (detail fractures) of unusual character.176 [Note that not all shell cracks turn to detail fractures. Unground rails and rails ground to fit the wheel profiles (conformally ground) can also produce monoplanar shells, some of which turn into detail fractures.] Proper control of railhead profile at the contact interface by grinding is the most effective means to enhance the fatigue resistance of rail, but the optimum grinding rate tends to decrease with the increase of rail strength.174 Residual stresses due to service and/or roller straightening establish crack propagation and fracture, leading to rail failure. Residual stresses due to wheel contact contribute to economic life limits (gauge corner spalling and subsurface shelling) when lubrication increases wear life and increased axle loads reduce fatigue life.176 The (thermal) residual stresses induced by rail welding act longitudinally and can reach 104 MPa (15 ksi) on cold days.160 The use of welded track demands an additional longer stress cycle period (due to welding stresses). Softer track (less stiff) support of the rail is beneficial, from a fatigue viewpoint, in that it increases base longitudinal tension stresses, but the improved compression stresses in the head are advantageous in preventing head-on-web bending stresses which lead to the development of bending tensile stresses at the bottom of the head when the wheel is directly over the stiff track. Additionally, the soft track minimizes the effects of dynamic loads. But the problem lies in the fact that one cannot produce the uniformity, consistency, and stability (or longevity) of, and track alignment with, the soft track; therefore, soft track is sought primarily to control track geometry.160 Therefore, railroad attempts to attain modestly stiff track which is stable and uniform in character, even though impact loads will lead to higher stresses in the rail and the head on web bending, may cause the occurrence of longitudinal tensile stresses toward the bottom of the railhead directly under the wheel location.160 Fatigue Cracking. One of the major reasons for catastrophic rail failure is the development of rolling contact fatigue (RCF) cracks. The fatigue failure may be displayed in a number of different ways, such as shelling on the upper rail face (in a sharply curved track), black spots, star cracking from fish-plate bolt-holes, and rail foot cracking. In Europe and for wheels in North America, shelling is a surface-initiated fatigue defect that usually develops in heavily deformed metal in the wheel loaded region of the running surface. In North America, for rails, shells are subsurface-initiated cracks that develop in undeformed metal—usually just at the boundary between the worked metal and the unworked metal; this is 1/8 to 1/4 in. beneath the surface. Rail steels can (frequently but not always) turn to detail fractures which break rails. If internally initiated cracks (such as shells, vertical split heads, and horizontal split heads) are not considered as RCF defects, this catastrophic rail failure may not occur. The problem of rail-end bolt-hole cracking (Fig. 7.33a) is usually eliminated, in Europe, by incorporation of residual stresses around the hole by introducing split sleeve cold expansion on existing tracks.177 In North America, bolt-hole cracking is eliminated by replacing bolted rail with continuous welded rail. Squat defect (or dark spot) due to surface initiated rolling contact fatigue is prevalent in Europe and Japan and is characterized by transverse cracks (Fig. 7.33b) with the possibility of causing broken rails and train derailment. Head checking is another surface-

(b)

(c)

FIGURE 7.33 (a) Typical cracks originating at rail-end bolt-hole.177(b) Vertical/longitudinal section through a squat-type rolling contact fatigue defect. Traffic from right to left. (c) Headcheck. Traffic from right to left. [(b) and (c) After M. C. Dubourg and J. J. Kalker, in Rail Quality and Maintenance for Modern Railway Operation, International Conference, Delft, Netherlands, 1992, p. 374.)

7.56

PEARLITE AND PROEUTECTOID PHASES

7.57

initiated RCF in steels which normally takes place on the rail gauge corner, especially on the outer rail in curved tracks (Fig. 7.33c). Deep spalling (to a depth of ~3 mm) due to crack-initiated surface fatigue from the white etching layers masks the details of fractures, and may cause the formation of potentially dangerous transverse cracks. Crushed heads, a defect due to the formation of extensive surface-initiated RCF, occur in older, softer rails with high nonmetallic inclusion contents which can lead to lack of support for the surface material, resulting in its lateral flow.178 The application of a cyclic stress pattern is an integral characteristic of railway track. The specific stress range in the stress cycle will clearly be affected by the axle loading and by residual stresses in the rail itself. It is, therefore, necessary to pay careful attention to correct ballasting of the track, because any local loss of support may result in a marked increase in the stress levels applied to the rail, effectively increasing the unsupported rail span and, thus, the applied stress. The bending fatigue strength of rails for 2 ¥ 106 cycles is a direct function of the tensile strength (Fig. 7.31), ranging from ~300 MPa at a tensile strength of 700 MPa to ~400 MPa at a tensile strength of 1100 MPa, in the as-rolled situation. However, as in all fatigue applications, the fatigue strength is strongly influenced by the presence of surface imperfections and damage, and by the presence of internal defects such as nonmetallic inclusions and hydrogen cracks. Corrosion effects on rails in service can decrease the fatigue strength level mentioned above by about 100 MPa,158 as shown in Fig. 7.31. The fatigue strength of weldments can be lower than that of the parent rail, and the fatigue strength of both flash-butt or thermit welds is of the order of 200 MPa. Note that residual stress adds to the fatigue load, and in this respect, stresses around drilled fish-plate holes should be minimized by reaming.158 Catastrophic web cracking in roller-straightened rail has been linked to high-residual-stress conditions.168,169 7.8.2.2 Wire Rod Steels. Many of the applications of medium- to high-carbon steels (containing 0.30 to 0.85% C, 0.40 to 0.80% Mn, 0.10 to 0.35% Si, 0.01 to 0.045% S, and 0.008 to 0.045% P) in rod form include the conversion of hot-rolled rod to wire by a cold-drawing operation.158 These hot-rolled rods should be free from surface defects due to casting and rolling of steel, segregation, surface imperfections, nonmetallic inclusions, and decarburization. Table 7.7 lists typical estimated tensile strength values for 5.6-mm (7/32-in.) medium-high carbon and high-carbon steel rods rolled on a mill using controlled cooling.179 The microstructure of such a rod is near to that obtained by patenting. The strength usually lies between those obtained by air patenting and lead patenting. Most high-carbon steel wire is drawn from such rods without prior patenting. The wire’s properties depend on the steel chemistry and manufacturing processes such as surface treatment, heat treatment, and drawing, which, in turn, determine the quality of the finished wire rope.180a 7.8.2.3 Wire Ropes. Rope wire is a commodity manufactured mainly with careful control for use in the construction of wire rope.180 Wire ropes are made in various sizes and cover a wide diversity of applications, such as in suspension bridges, as the main load-carrying cables, and as suspension elements connecting the carriage-way with these cables; suspension elements for suspended roofs; in haulage applications such as lifting ropes or cranes, elevators, and mine hoists; load transmission ropes on excavators; and trawl ropes for fishing vessels.158 A standard wire rope consists of three basic components: the wires, the strands, and a core (Fig. 7.34a).181 The selection of wire rope is made based on six factors: (1) strength resistance to breaking, (2) resistance to bending fatigue, (3) resistance

CHAPTER SEVEN

7.58

TABLE 7.7 Tensile Strengths of 5.6-mm (7/32-in.)-Diameter Hot-Rolled MediumHigh-Carbon and High-Carbon Steel Rod179 Data produced from rod produced with controlled cooling Tensile strength for steel with manganese content of 0.60%

0.80%

1.00%

Carbon content of steel, %

MPa

ksi

MPa

ksi

MPa

ksi

0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85

641 689 745 793 848 896 951 1000 1055 1103 1151 1207

93 100 108 115 123 130 138 145 153 160 167 175

676 731 779 834 883 938 986 1041 1089 1138 1193 1241

98 106 113 121 128 136 143 151 158 165 173 180

717 793 820 869 931 972 1020 1076 1124 1179 1227 1282

104 115 119 126 135 141 148 156 163 171 178 186

Reprinted by permission of ASM International, Materials Park, Ohio.

to vibrational fatigue, (4) abrasion resistance, (5) crushing resistance, and (6) reserve strength. There are six grades of wire rope referred to as traction steel (TS), mild plow steel (MPS), plow steel (PS), improved plow steel (IPS), extra improved plow steel (EIPS), and extra extra improved plow steel (EEIPS).181 These steel grades denote the strength of a particular size and grade of rope. The plow steel strength curve is used as the basis for determining the strength of all steel rope wires; the tensile strength of any steel wire grade is a function of the diameter and is highest in the smallest wires. The steel wire has the “bright” or uncoated finish. Steel wires can be galvanized. Drawn galvanized wire has the same strength as the bright wire; however, wire galvanized at finished size has commonly 10% lower strength.181 For more detailed discussion, readers are referred to a recent wire rope user’s manual.181 The strength of the rope depends on the number of individual wires and their arrangement in the rope. A wire rope is produced by helically laying wires about a central axis into strands, followed by helically laying the strands into a rope (Fig. 7.34). The helically wound strands may or may not be wound around an axial member called the core. Individual strands may be based on a centerless grouping principle or comprise a layer of wires wound around a center wire (single-layer principle) or of several layers wound around a center wire (multiple-operation principle). Wire rope is distinguished by its construction, i.e., by the way the wires have been laid to form strands, and by the way the strands have been laid around the core. For example, if the strand in the rope is wound in the opposite direction to the wires in the strand, it is said to be a left or right regular lay rope; and if the strands are wound in the same direction as that of the wires in the strand, it is said to be left or right lang lay rope.158,182 Among all types of wire rope, right regular lay (RRL) is widely used. Or alternate lay consists of alternating regular and lang lay strands. Figure 7.34b shows the cross sections of four basic strand patterns around which standard wire ropes are

CORE

(b)

(a)

WIRE CENTER WIRE

7 Wire

19 Warrington

19 Seale

25 Filler Wire

STRAND 6  19 Classification

(c)

7.59 WIRE ROPE

6  19 Seale IWRC

6  21 Filler Wire FC

6  25 Filler Wire IWRC

FIGURE 7.34 (a) Three basic components of a typical wire rope. (b) Four basic strand patterns. (c) Cross section of commonly used 6 ¥ 19 classification wire rope construction.181 (Courtesy of Wire Rope Technical Board, Maryland.)

6  26 Warrington Seale IWRC

7.60

CHAPTER SEVEN

built. The wire ropes are classified by the number of strands in the rope, the number and arrangement of wires in each strand, and a descriptive word or letter to recognize the type of construction or the geometric arrangement of wires. Some commonly used wire rope constructions are: 6 ¥ 7 classification such as 6 ¥ 7 FC (fiber core); 6 ¥ 19 classification such as 6 ¥ 19 Seale IWRC (independent wire rope core), 6 ¥ 21 filler wire FC, 6 ¥ 25 filler wire IWRC, 6 ¥ 26 Warrington Seale IWRC (Fig. 7.34c); 6 ¥ 37 classification such as 6 ¥ 31 Warrington Seale IWRC, 6 ¥ 36 Seale filler wire IWRC, 6 ¥ 41 Warrington Seale (WS) IWRC, 6 ¥ 46 Seale filler wire IWRC, 6 ¥ 49 filler wire Seale (FWS) IWRC, etc.; 6 ¥ 61 classification such as 6 ¥ 55 (two-operation) filler wire Seale IWRC, 6 ¥ 57 Seale filler wire IWRC, and 6 ¥ 61 filler wire Warrington Seale IWRC. Figure 7.34c represents the commonly used cross sections of wire rope construction according to 6 ¥ 19 classification.181 Rotation-resistant ropes are a special class of wire rope designed to resist the likelihood of spinning or rotating under load. They are available as single-layer or multilayer strand types.181 The rope construction has important bearings on the application. The geometric arrangement of the wires can affect the rope strength, the internal fretting of the wires, the rope stiffness, and the untwisting propensities. The ultimate break strength of a wire rope is by design less than the aggregate strength of all the wires and will depend on the construction of rope and grade of wire used. The proper design factor of a wire rope requires consideration of all loads. These loads should incorporate (if applicable) acceleration, deceleration, rope speed, rope attachments, number and arrangement of sheaves and drums, conditions producing corrosion and abrasion, and length of rope. Usually, the more flexible rope, which contains the largest number of wires and has fiber cores, will stretch more than all-metal ropes with fewer ropes and thus less flexibility.183 In the case of suspension bridge rope, galvanizing treatment is applied to the steel wire to improve corrosion resistance properties. In nonsevere conditions, a light oiling treatment serves the purpose of protective coating. However, in severe service conditions, stainless steel wires are used for very demanding applications. 7.8.2.4 Bridge Ropes. Bridge rope is formed in a similar manner to a helical strand except that strands are formed helically around a center strand or rope, instead of wires. Bridge rope is usually made with a regular lay construction. Bridge strands and ropes are not usually subjected to fluctuating bending stresses over sheaves. The loads are generally in direct tension and comprise mainly a large, constant dead load plus a relatively smaller pulsating live load. The fatigue characteristics of such tension members are, therefore, of interest.182 7.8.2.5 Spring Wires. Mechanical spring wire can be grouped into seven types. Table 7.8 lists the ranges of chemical composition for seven types:180,182,184 1. Hard-drawn spring wire (ASTM A227) is less costly and is used for the manufacture of mechanical springs in applications requiring infrequent stress repetitions or static loads. Its surface quality is relatively low with such imperfections as hairline seams. For hard-drawn spring steel wire (class 1 and class 2), the major requirement is the tensile strength. Class 2 is a higher-strength product. 2. High-tensile hard-drawn spring wire (ASTM A629) is a special-quality, harddrawn carbon steel spring wire with restricted size tolerances. This is used where such restricted dimensional requirements are essential for the production of highly stressed mechanical springs and wire forms. It is used for applications subject to static load and infrequent stress repetitions.182

PEARLITE AND PROEUTECTOID PHASES

7.61

TABLE 7.8 Mechanical Spring Wire (Chemical Composition, %)180,182,184

Alloying element

Uncoated, drawn galvanized at finish size harddrawn wire

High-tensile hard-drawn wire

Carbon Manganese Phosphorus Sulfur Silicon

0.45–0.85 0.30–1.30 0.035 max. 0.045 max. 0.15–0.35

0.65–1.00 0.20–1.30 0.035 max. 0.045 max. 0.15–0.35

OilWire for tempered heat-treated wire components

Music spring steel wire

0.55–0.85 0.30–1.20 0.035 max. 0.045 max. 0.15–0.35

0.70–1.00 0.20–0.60 0.025 max. 0.030 max. 0.15–0.35

0.50–1.03 0.30–1.30 0.035 max. 0.045 max. 0.15–0.35

Valve springquality wire 0.60–0.75 0.60–0.90 0.025 max. 0.030 max. 0.15–0.35

Upholstery spring wire 0.45–0.75 0.60–1.20 0.025 max. 0.030 max. 0.15–0.35

3. Oil-tempered spring wire (ASTM A 229) is a general-purpose wire subjected to static loads or relatively infrequent stress repetitions. This is slightly more expensive and more susceptible to embrittling effects of plating than type 1; however, it is superior in surface smoothness. It is used in automotive and related industries. 4. Spring wire for heat-treated components (ASTM A713) is used for spring steel wire intended for heat-treated parts. The major requirement is a composition suitable for heat treatment. 5. Music spring wire (ASTM A228) is the least subjected to hydrogen embrittlement by electroplating and is similar to valve spring wire in surface quality. This is intended for applications requiring high stresses and good fatigue properties. Final cold drawing is usually accomplished by a wet white liquor method or by phosphating to develop a characteristic smooth bright surface. Specialized coiling tests, twist tests, torsion tests, and bend tests are employed to ensure that the exacting requirements of the uniformity and quality with exceptional high tensile strength of this type of wire are met. 6. Valve spring quality wire (ASTM A230) is used for the manufacture of engine valve springs and other springs requiring dynamic or high-stress repetitions with high-fatigue-strength properties. To meet this requirement, wire will have the highest degree of uniformity with respect to surface imperfections, internal soundness, and definite mechanical property values. The wire is available in both the oil-tempered and hard-drawn conditions or in one of the three conditions available for spring steel for heat-treated components.185 7. Upholstery spring steel wire (ASTM A407) is used in the manufacture of coil spring constructions for mattresses, furniture, beds, and automotive seats and cushions. It is available in sizes from 0.89 to 5.77 mm (0.035 to 0.225 in.) in diameter.185 This is not used for the manufacture of other types of springs. This is drawn from thermally treated or controlled cooled wire rod or wire. 7.8.2.6 Tire Bead Wires and Tire Cords. Tire bead wire is wrapped and reinforced by the rubber-coated plies for pneumatic tires.184 Wire for steel cord is drawn down from 5.5-mm-diameter hot-rolled, high-carbon steel (0.67 to 0.82% C, 0.4 to 0.6% Mn) rods to 0.94-mm (0.037-in.) diameter with a bronze-plated finish. Uniformity in chemical and mechanical properties and a good surface finish for rubber adhesion are essential for satisfactory performance. Mechanical property tests are

7.62

CHAPTER SEVEN

performed on wire samples that have been heated for 1 hr at 150°C (300°F). Minimum breaking load for this wire is 129 kg (285 lb). In torsion tests, the wires must withstand 58 twists minimum in a 203-mm (8-in.) gauge length. Tire cord in the form of strands is used for reinforcing the rubber matrix for automobile radial (ply) tires.184 For tire cord produced from high-carbon hot-rolled steel rods (containing 0.67 to 0.82% C, 0.4 to 0.6% Mn) which is drawn to wire 0.15 to 0.25 mm (0.006 to 0.010 in.) in diameter. The deformation process involves a larger extent of strain (e > 1). The deformability of inclusions is a critical parameter that affects the product performance.184a Elimination or minimization of the size of nondeformable inclusions such as spinel, calcium aluminates, and especially alumina in the steel rod is necessary because they are the most common cause of breakage of filament wire (and wear of the die) during final drawing and bunching into tire cord.184b This extremely fine, high-tensile-strength wire is produced from controlled cooled rods by drawing, followed by drawing, lead patenting (at 500 to 550°C), second drawing and lead patenting, brass plating, wet drawing, and forming into strands to be embedded into the tire rubber.185 The patenting treatments restore an undeformed structure and ensure the necessary refinement of the pearlitic carbide lamellae for the subsequent large amounts of cold work, by shear cracking of the pearlite. Stranding of wire involves laying several wires helically like rope wire; and, again, the strands can be laid to produce the tire cord. The number of filaments in a strand and number of strands in a cord can be varied based on the requirements of reinforcement. Additionally, the cord can include an outer spiral wrap, with a single filament, to stop strands from separating when cord is under an axial compressive load. The length of the lay, i.e., the length of the strand or cord required for one rotation of filament or strand, respectively, is important from the viewpoint of the contact bearing area and fretting, like wire rope. The current conventional tensile strength of the tire cord steel is about 3600 MPa (522 ksi). Steelmakers are attempting to improve the strength to 5000 MPa (725 ksi) to satisfy the requirements of automobile and tire manufacturers. Higher-strength tire cord allows the automobile and tire manufacturers to reduce the weight and roll resistance of tires, leading to significant increase in vehicle fuel economy.184b Accordingly, current developments include small alloy additions such as 0.20 to 0.25% Cr; maintaining ultra-low levels of Mg, Ca, and Al contents to prevent the formation of nondeformable inclusions; adding Wollastonite ladle flux to modify the inclusions during refining of tire cord steels;184b using 0.82 to 0.92% C, controlled cooling for rod coils in the hot mill to eliminate the need for lead patenting treatments; substitution of air patenting for lead patenting for certain products; elimination of a nonlamellar structure out of pearlite to improve the delamination resistance and work-hardening rate during drawing;185a and the use of low alloyed microalloyed medium- to high-carbon steels to provide added strength to the drawn product.158,185a 7.8.2.7 Wire for Prestressed Concrete. There are two types of uncoated round high-carbon steel wire for prestressed concrete applications: cold-drawn and colddrawn and suitably stress-relieved. The wire is used for linear or circular pretensioning or posttensioning structural members (ASTM A416 and ASTM A421). The stress-relieved product can be used as a single wire or as a strand that has been stress-relieved after stranding.185 Stress-relieved uncoated high-carbon steel wire is normally employed for the linear prestressing of concrete structures. It is produced in diameters of 4.88, 4.98, 6.35, and 7.01 mm (0.192, 0.196, 0.250, and 0.276 in.) to tensile strengths of 1620 to

PEARLITE AND PROEUTECTOID PHASES

7.63

1725 MPa (235 to 250 ksi). The wire can also be made in a low-relaxation mode which after drawing is subjected to a continuous thermomechanical treatment to produce the desired mechanical properties. The tensile strength of low-relaxation wire is the same as that of normal-relaxation stress-relieved wire; however, the minimum yield strength is also 90% of the minimum tensile strength (ASTM A421). After stress-relieving the inside diameters of coils are usually greater than those of the drawn wire, in order to prevent stress set or reintroduction of coil stress. Coil inside diameters may be as large as 200 times the wire diameter. High-carbon steel wire for mechanically tensioning is generally used for circular prestressing in the manufacture of concrete pressure pipe.185 7.8.2.8 Cold Heading Wires. Cold heading is a cold forging process where the force, developed by one or more blows of a mechanical hammer or heading tool, is employed to displace or upset a portion of a blank to form a precise section of different contour or configuration than the original blank. Although the process is cold, heat is produced by the work performed.182 The manufacture of fasteners such as high-tensile-strength bolts is done with medium-carbon (0.30 to 0.45%) steels whereas lockwashers or screw drivers are made from high-carbon steel wire rod which is spheroidize annealed either inprocess or after drawing finished sizes. Hot-rolled coiled rod is usually subcritically annealed to furnish low strength and ductility. Cold heading steel wire rod is produced with carefully controlled manufacturing practices and rigid inspection practices to ensure the necessary degree of homogeneity, internal soundness, cleanliness, and freedom from surface imperfections. Decarburization must be held to a minimum for those products that are quenched and tempered. A fully killed fine-grain steel is usually required for the most difficult operations. 7.8.3 Microalloyed Ferrite-Pearlite Forging Steels The concept of simultaneously increasing the pearlite content and making it more dilute has been exploited to advantage in microalloyed medium-carbon ferritepearlite forging steels which lead to an increase of both strength and toughness. This is achieved by increasing Mn content, decreasing C content, and/or increasing the cooling rate through the transformation range. An increase of Si content to 0.6 to 0.7% improves the toughness; microalloying with V and N raises the strength with some sacrifice of toughness; and reduction of g grain size decreases the pearlite and ferrite grain size, which, in turn, improves the toughness without sacrifice of strength.186 Figure 9.21 shows the advantages of manufacturing automotive components such as connecting rods, crank shafts, steering knuckles, truck components, front axle beam, etc., from hot-rolled microalloyed medium-carbon forging steels over the hardened and tempered steels (see Sec. 9.5.2 also).187,188

REFERENCES 1. N. Ridley, Met. Trans., vol. 16A, 1984, pp. 1019–1036. 2. N. Ridley, in Proceedings of the International Conference on Phase Transformations in Ferrous Alloys, eds. A. R. Marder and J. I. Goldstein, TMS–AIME, Warrendale, Pa., 1984, pp. 201–236.

7.64

CHAPTER SEVEN

3. J. K. Chen, S. W. Spencer, M. E. Ekstrand, D. Chen, and W. T. Reynolds, Jr., Met. Trans., vol. 27A, 1996, pp. 1683–1689. 4. J. R. Vilella, G. E. Guellich, and E. C. Bain, Trans. ASM, vol. 24, 1936, pp. 225–252. 5. M. Hillert, in The Decomposition of Austenite by Diffusional Processes, eds. V. F. Zackay and H. I. Aaronson, Interscience, New York, 1961, pp. 197–237. 6. M. E. Nicholson, J. Met., vol. 6, 1954, p. 1071. 7. R. F. Mehl and W. C. Hagel, The Austenite-Pearlite Reaction, Progress in Metal Physics, vol. 6, Pergamon Press, Oxford, 1956, p. 74. 8. S. Modin, Jernkontorets Ann., vol. 1235, 1951, p. 169. 9. R. W. K. Honeycombe and H. K. D. H. Bhadeshia, Steels: Microstructure and Properties, 2d ed., Arnold, London, 1995. 10. D. L. Lee and C. G. Park, Scripta Metall., vol. 32, no. 6, 1995, pp. 907–912. 11. E. M. Taleef, C. K. Syn, D. R. Lesuer, and O. D. Sherby, in Thermomechanical Processing and Mechanical Properties of Hypereutectoid Steels and Cast Irons, eds. D. R. Lesuer, C. K. Syn, and O. D. Sherby, TMS, Warrendale, Pa., 1997, pp. 127–141. 12. J. D. Verhoeven and E. D. Gibson, Met. Trans., vol. 29A, 1998, pp. 1181–1189. 13. S. A. Hackney and G. J. Shiflet, Scripta Metall., vol. 19, 1985, p. 757. 14. S. A. Hackney and G. J. Shiflet, Acta Metall., vol. 35, 1987, pp. 1007–1017, 1019–1028. 15. S. A. Hackney, Scripta Metall., vol. 25, 1991, p. 1453. 16. H. J. Lee, G. Spanos, G. J. Shiflet, and H. I. Aaronson, Acta Metall., vol. 36, 1988, p. 1129. 17. M. J. Whiting and P. Tsakiropoulos, Scripta Metall., vol. 29, 1993, p. 401. 18. C. Zener, Trans. AIME, vol. 167, 1946, p. 550. 19. M. Hillert, Mechanism of Phase Transformations in Crystalline Solids, Monograph no. 33, Institute of Metals, London, 1968, p. 231. 20. M. P. Puls and J. S. Kirkaldy, Met. Trans., vol. 3, 1972, pp. 2777–2796. 21. J. M. Shapiro and J. S. Kirkaldy, Acta Metall., vol. 16, 1968, pp. 579–585. 22. B. E. Sundquist, Acta Metall., vol. 16, 1968, pp. 1413–1427. 23. M. Hillert, in Proceedings of the International Conference on Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS–AIME, Warrendale, Pa., 1982, pp. 789–806; J. W. Cahn and W. C. Hagel, Acta Met., vol. 11, 1963, pp. 561–574. 24. R. D. Doherty, in Physical Metallurgy, 4th ed., eds. R. W. Cahn and P. Haasen, Elsevier Science BV, Amsterdam 1996, chap. 15, pp. 1363–1506. 25. K. Han, G. D. W. Smith, and D. V. Edmonds, Met. Trans., vol. 26A, 1995, pp. 1617–1631. 26. Ref. 12 in A. R. Marder and B. L. Bramfitt, Met. Trans., vol. 7A, 1976, pp. 902–905. 27. N. Ridley, in Proceedings of the International Conference on Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS–AIME, Warrendale, Pa., 1982, pp. 807–817. 28. R. C. Sharma, G. R. Purdy, and J. S. Kirkaldy, Met. Trans., vol. 10A, 1979, pp. 1129–1139. 29. S. K. Tewary and R. C. Sharma, Met. Trans., vol. 16A, 1985, pp. 597–603. 30. J. S. Kirkaldy, Met. Trans., vol. 4A, 1973, pp. 2327–2333. 31. E. Lemaire, J. Copreaux, and F. Roch, Scripta Metall., vol. 35, no. 1, 1996, pp. 83–89. 32. T. Gladman, I. D. McIvor, and F. B. Pickering, JISI, vol. 210, 1972, p. 916. 33. C. M. Bae, W. J. Nam, and C. S. Lee, Scripta Metall., vol. 35, no. 5, 1996, pp. 641–646. 34. R. F. Mehl, C. S. Barrett, and D. W. Smith, Trans. AIME, vol. 105, 1933, p. 215. 35. W. A. Johnson and R. F. Mehl, Trans. AIME, vol. 135, 1939, p. 416. 36. R. F. Mehl and A. Dube, Phase Transformations in Solids, John Wiley & Sons, New York, 1951, p. 574.

PEARLITE AND PROEUTECTOID PHASES

7.65

37. J. W. Cahn and W. C. Hagel, in The Decomposition of Austenite by Diffusional Processes, eds. V. F. Zackay and H. I. Aaronson, Interscience, New York, 1962, pp. 131–192. 38. P. G. Shewmon, Transformations in Metals, McGraw-Hill, New York, 1969. 39. F. C. Hull, R. A. Colton, and R. F. Mehl, Trans. AIME, vol. 150, 1942, pp. 185–207. 40. B. B. Rath, in Proceedings of the International Conference on Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS–AIME, Warrendale, Pa., 1982, pp. 1097–1103. 41. J. W. Christian, Theory of Transformations in Solids, Pergamon Press, London, 1965. 42. P. R. Howell, Materials Characterization, vol. 40, 1998, pp. 227–260. 43. J. W. Cahn and W. C. Hagel, Acta Metall., vol. 11, 1963, pp. 561–574. 44. A. R. Marder and B. L. Bramfitt, Met. Trans., vol. 6A, 1975, pp. 2009–2014. 45. B. L. Bramfitt and A. R. Marder, Met. Trans., vol. 7A, 1976, pp. 902–905. 46. D. D. Pearson and J. D. Verhoeven, Met. Trans., vol. 15A, 1984, pp. 1037–1045. 47. B. L. Bramfitt and A. R. Marder, Int. Metallogr. Soc. Proc., 1968, pp. 43–45. 48. F. A. Khalid and D. V. Edmonds, Acta Metall., vol. 41, no. 12, 1993, pp. 3421–3434. 49. M.-X. Zhang and P. M. Kelly, Scripta Mater., vol. 37, no. 12, 1997, pp. 2009–2015. 50. M. J. Whiting and P. Tsakiropoulos, Scripta Metall. Mater., vol. 30, 1994, p. 1031; Mater. Sci. and Tech., vol. 11, 1995, pp. 717–727. 51. M. B. Cortie and C. E. Mavrocordatos, Met. Trans., vol. 22A, 1991, pp. 11–18. 52. A. Das, S. K. Pabi, I. Manna, and W. Gust, J. Mater. Sci., vol. 34, 1999, pp. 1815–1821. 53. D. Cheetham and N. Ridley, J. Inst. Metals, vol. 99, 1971, p. 371. 54. L. Kumar, R. V. Ramanujan, R. Tewari, P. Mukhopadhyay, and S. Banerjee, Scripta Mater., vol 40, no. 6, 1999, pp. 723–728. 55. M. Mannerkoski, Acta Polytech. Scand., 1964, chap. 26, p. 7. 56. K. Relander, Acta Polytech. Scand., 1964, chap. 34, p. 7. 57. J. V. Bee and D. V. Edmonds, Metallography, vol. 12, 1979, pp. 3–21. 58. R. W. K. Honeycombe, Met. Trans., vol. 7A, 1976, pp. 915–936. 59. A. T. Davenport and P. C. Becker, Met. Trans., vol. 2A, 1971, pp. 2962–2964. 60. P. R. Howell, in Proceedings of the International Conference on Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS–AIME, Warrendale, Pa., 1982, pp. 399–425. 61. R. W. Honeycombe and F. B. Pickering, Met. Trans., vol. 3A, 1972, pp. 1099–1112. 62. A. T. Davenport and R. W. K. Honeycombe, Proc. Roy. Soc., London, vol. A322, 1971, p. 191. 63. T. Gladman, The Physical Metallurgy of Microalloyed Steels, The Institute of Materials, London, 1997. 64. R. W. K. Honeycombe, in Proceedings of the International Conference on Phase Transformations in Ferrous Alloys, eds. A. R. Marder and J. J. Goldstein, TMS–AIME, Warrendale, Pa., 1984, pp. 259–280. 65. T. Obara, G. J. Shiflet, and H. I. Aaronson, Met. Trans., vol. 14A, 1983, pp. 1159–1161. 66. T. Furuhara and H. I. Aaronson, Scripta Metall., vol. 22, 1988, pp. 1635–1637. 67. M. V. Kral, W. H. Hofmeister, and J. E. Wittig, Met. Trans., vol. 28A, 1997, pp. 2485–2497. 68. G. Fourlaris, Mater. Sc. Forum, vol. 284–286, 1998, pp. 427–434. 69. G. Fourlaris, A. J. Baker, and G. D. Papadimitriou, Solid Æ Solid Phase Transformations, eds. W. C. Johnson, J. M. Howe, D. E. Laughlin, and W. A. Soffa, TMS, Warrendale, Pa., 1994, pp. 535–540. 70. K. Campbell and R. W. K. Honeycombe, Met. Sci., vol. 8, 1974, p. 197.

7.66

CHAPTER SEVEN

71. P. G. Berry and R. W. K. Honeycombe, Met. Trans., vol. 1A, 1970, p. 3279. 72. M. H. Ainsley, G. J. Cocks, and D. R. Miller, Met. Sci., vol. 13, 1979, p. 20. 73. H. I. Aaronson, in Decomposition of Austenite by Diffusional Processes, Interscience, New York, 1962, pp. 387–546. 74. C. A. Dubé, H. I. Aaronson, and R. F. Mehl, Rev. Met., vol. 55, 1958, p. 201. 75. H. S. Fong and S. G. Glover, Met. Trans., vol. 15A, 1984, pp. 1643–1651. 76. H. I. Aaronson, C. Laird, and K. R. Kinsman, Phase Transformations, ASM, Metals Park, Ohio, 1970, pp. 313–396. 77. L. E. Samuels, Optical Microscopy of Carbon Steels, ASM, Metals Park, Ohio, 1979. 78. K. R. Kinsman and H. I. Aaronson, Transformation and Hardenability in Steels, Climax Molybdenum Co., Ann Arbor, Mich., 1967, pp. 39–53. 79. C. A. Dubé, Ph.D. Thesis, Carnegie Institute of Technology, 1948. 80. C. Zener, J. Appl. Phys., vol. 20, 1949, pp. 950–953. 81. C. Atkinson, H. B. Aaron, K. R. Kinsman, and H. I. Aaronson, Met. Trans., vol. 4A, 1973, pp. 783–792. 82. J. R. Bradley, G. J. Shiflet, and H. I. Aaronson, in Proceedings of the International Conference on Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS–AIME, Warrendale, Pa., 1982, pp. 819–824. 83. A. D. King and T. Bell, Met. Trans., vol. 6A, 1975, pp. 1419–1429. 84. J. R. Bradley and H. I. Aaronson, Met. Trans., vol. 12A, 1981, pp. 1729–1741. 85. M. Hillert and L. I. Staffansson, Acta Chem. Scand., vol. 24, 1970, p. 3618. 86. H. I. Aaronson and H. A. Domian, TMS–AIME, vol. 236, 1966, p. 781. 87. G. J. Shiflet, H. I. Aaronson, and J. R. Bradley, Met. Trans., vol. 12A, 1981, pp. 1743–1750. 88. T. Tanaka, H. I. Aaronson, and M. Enomoto, Met. Trans., vol. 26A, 1995, pp. 561–580. 89. T. Furuhara and H. I. Aaronson, Acta Metall. et Mater. vol. 39, 1991, p. 2887. 90. X-Z. Bo and H-S. Fang, Acta Mater., vol. 46, no. 8, 1998, pp. 2929–2936. 91. R. G. Kamat, E. B. Hawbolt, L. C. Brown, and J. K. Brimcombe, Met. Trans., vol. 23A, 1992, pp. 2469–2480. 92. E. Enomoto, Met. Trans., vol. 22A, 1991, pp. 1235–1245. 93. H. B. Aaron and H. I. Aaronson, Met. Trans., vol. 2, 1971, pp. 23–37. 94. H. I. Aaronson, in The Mechanism of Phase Transformations in Metals, Institute of Metals, London, 1956, p. 47. 95. E. P. Simonen, H. I. Aaronson, and R. Trivedi, Met. Trans., vol. 4, 1973, pp. 1239–1245. 96. P. N. T. Unwin and R. B. Nicholson, Acta Metall., vol. 17, 1969, p. 139. 97. G. Spanos and M. G. Halls, Met. Trans., vol. 27A, 1996, pp. 1519–1534. 98. H. I. Aaronson and C. Wells, Trans. AIME, vol. 206, 1956, pp. 1216–1223. 99. W. T. Reynolds, Jr., M. Enomoto, and H. I. Aaronson, in Proceedings of International Conference on Phase Transformations in Ferrous Alloys, eds. A. R. Marder and J. I. Goldstein, TMS–AIME, Warrendale, Pa., 1984, pp. 155–200. 100. H. K. D. H. Bhadeshia, Progr. Mater. Sci., vol. 29, no. 4, 1985, pp. 321–386. 101. H. K. D. H. Bhadeshia and J. W. Christian, Met. Trans., vol. 21A, 1990, pp. 776–797. 102. H. I. Aaronson, W. T. Reynolds, Jr., G. J. Shiflet, and G. Spanos, Met. Trans., vol. 21A, 1990. pp. 1343–1380. 103. M. Enomoto, Met. Trans., vol. 25A, 1994, pp. 1947–1955. 104. M. Hillert, Jernkontorets Ann., vol. 14, 1957, p. 757. 105. M. Hillert, Met. Trans., vol. 6A, 1975, pp. 5–19.

PEARLITE AND PROEUTECTOID PHASES

7.67

106. A. Vander Ven and L. Delaey, Progr. Mater. Sci., vol. 40, 1996, pp. 181–264. 107. R. Trivedi, Met. Trans., vol. 1, 1970, p. 921. 108. G. P. Ivantsov, in Growth of Crystals, vol. 3, eds. A. V. Shubnikov and N. N. Sheflat, Consultant Bureau, New York, 1962, p. 53. 109. M. M. Kostic, E. B. Hawbolt, and L. C. Brown, Met. Trans., vol. 7A, 1976, pp. 1643–1653. 110. R. Trivedi, in Proceedings of the International Conference on Solid Æ Solid Phase Transformations, eds. H. I. Aaronson et al., TMS–AIME, Warrendale, Pa., 1982, pp. 477–502. 111. M. Hillert, Met. Trans., vol. 25A, 1994, pp. 1957–1966. 112. W. R. Bosze and R. Trivedi, Met. Trans., vol. 5, 1974, p. 511. 113. G. R. Purdy, Met. Sci., vol. 5, 1971, p. 81. 114. J. W. Cahn, W. B. Hillig, and G. W. Sears, Acta Metall., vol. 12, 1964, p. 1421. 115. G. J. Jones and R. Trivedi, J. Appl. Phys., vol. 42, 1971, p. 4299; J. Cryst. Growth, vol. 29, 1975, p. 155. 116. C. Atkinson, Proc. Roy. Soc., London, vol. 378A, 1981, p. 351; vol. 384A, 1982, p. 167. 117. C. Laird and H. I. Aaronson, Acta Metall., vol. 17, 1969, p. 505. 118. H. I. Aaronson and C. Laird, Trans. TMS–AIME, vol. 242, 1968, p. 1437. 119. R. Sankaran and C. Laird, Acta Metall., vol. 2, 1974, p. 957. 120. K. R. Kinsman, E. Eichen, and H. I. Aaronson, Met. Trans., vol. 6A, 1975, pp. 303–317. 121. S. K. Bhattacharya, J. H. Perepezko, and T. B. Massalski, Scripta Metall., vol. 7, 1973, pp. 485–488. 122. T. B. Massalski, in Metals Handbook, vol. 9, 9th ed., ASM, Metals Park, Ohio, 1985, pp. 655–657. 123. T. B. Massalski, Met. Trans., vol. 15A, 1984, pp. 421–425. 124. M. R. Plichta,W. A. T. Clark, and H. I. Aaronson, Met. Trans., vol. 15A, 1984, pp. 427–435. 125. P. Wang, G. B. Vishwanathan, and V. K. Vasudevan, Met. Trans., vol. 23A, 1992, pp. 690–697. 126. M. Hillert, Met. Trans., vol. 15A, 1984, pp. 411–419. 127. T. B. Massalski, Massive Transformations in Phase Transformations, ASM, Metals Park, Ohio, 1970, pp. 433–486. 128. G. A. Chadwick, Metallography of Phase Transformations, Crane, Russak and Company, New York, 1972. 129. D. Swanson and J. G. Parr, JISI, vol. 202, 1964, p. 104. 130. R. D. Doherty in Physical Metallurgy, eds. R. W. Cahn and P. Haasen, North-Holland Physics Publishing, Amsterdam, 1983, pp. 933–1030. 131. J. H. Perepezko, Met. Trans., vol. 15A, 1984, pp. 437–447. 132. E. S. K. Menon, M. R. Plichta, and H. I. Aaronson, Acta Metall., vol. 36, 1988, p. 321. 133. J. E. Kittl and T. B. Massalski, Acta Metall., vol. 15, 1961, p. 161. 134. G. A. Sargent, L. Delay, and T. B. Massalski, Acta Metall., vol. 16, 1968, p. 723. 135. M. R. Plichta, J. C. Williams, and H. I. Aaronson, Met. Trans., vol. 8A, 1977, pp. 1885–1892. 136. J. C. Caretti, Acta Metall., vol. 34, 1986, pp. 385–393. 137. E. A. Wilson, ISIJ Int., vol. 34, no. 8, 1994, pp. 615–630. 138. R. J. Ackert and J. G. Paar, JISI, vol. 209, 1971, p. 912. 139. M. J. Bilby and J. G. Parr, JISI, vol. 202, 1964, p. 100. 140. Ref. 44 in M. Hillert, Met. Trans., vol. 6A, 1975, pp. 5–19. 141. T. B. Massalski, J. H. Perepezko, and J. Jaklovsky, Mater. Sc. Eng., vol. 18, 1975, p. 193.

7.68

CHAPTER SEVEN

142. C. Hayzeldon and B. Cantor, in Int. J. Rapid Solidification, 1984–1985, p. 237. 143. S. H. Chong, A. Sayles, R. Keyse, J. D. Atkinson, and E. A. Wilson, Mater. Trans., JIM, vol. 39, no. 1, 1998, pp. 179–188. 144. I. Ishikawa, T. Takahashi, and T. Ochi, Met. Trans., vol. 25A, 1994, pp. 929–936. 145. W. Heckel and H. W. Paxton, TMS–AIME, vol. 218, 1960, p. 799. 146. T. Ando and G. Krauss, Acta Metall., vol. 29, 1981, pp. 351–363. 147. W. Heckel and H. W. Paxton, Trans. ASM, vol. 53, 1961, p. 539. 148. F. B. Pickering, in Encyclopedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4621–4632. 149. T. Gladman, D. Dulieu, and I. D. McIvor, in Microalloying ‘75, Union Carbide Corporation, New York, 1977, pp. 32–34. 150. F. B. Pickering, in Towards Improved Toughness and Ductility, Climax Molybdenum Co., Greenwich, Conn., 1971, p. 9. 151. J. D. Baird and R. R. Preston, Processing and Properties of Low-Carbon Steels, AIME, New York, 1973, p. 1. 152. B. Mintz, Met. Technol., vol. 11, 1984, pp. 265–272. 153. B. Mintz, Met. Technol., vol. 11, 1984, pp. 52–60. 154. A. R. Marder, in Proceedings of the International Conference on Phase Transformations in Ferrous Alloys, TMS–AIME, Warrendale, Pa., 1984, pp. 11–41. 155. F. B. Pickering, in Constitution and Properties of Steels, vol. ed. F. B. Pickering, VCH, Weinheim, 1992, pp. 41–94. 156. M. G. M. F. Gomes, L. H. Almeida, L. C. F. C. Gomes, and I. L. May, Materials Characterization, vol. 39, 1997, pp. 1–14. 157. J. H. Martens, and D. P. Wirick, Proceedings of the International Symposium on Rail Steels for 21st Century, Iron and Steel Society, Warrendale, Pa., 1995, pp. 1–3. 158. T. Gladman, in Constitution and Properties of Steels, vol. ed. F. B. Pickering, VCH, Weinheim, 1992, pp. 401–432. 159. R. K. Steele, Rail Quality and Maintenance for Modern Railway Operation, International Conference, Delft 1992, Kluwer Academic Publishers, Dordrecht, Netherlands, 1993, pp. 77–97. 160. R. K. Steele, private communication, 1999. 161. X. Su and P. Clayton, Wear, vol. 197, 1996, pp. 137–144. 162. B. L. Bramfitt, Metals Handbook, vol. 20: Materials Selection and Design, ASM International, Materials Park, Ohio, 1997, pp. 357–382. 163. A. J. Perez-Unzueta and J. H. Beynon, Wear, vol. 162–164, 1993, pp. 173–182. 164. W. H. Hodgson, Rail Quality and Maintenance for Modern Railway Operation, International Conference, Delft 1992, Kluwer Academic Publishers, Dordrecht, Netherlands, 1993, pp. 29–39. 165. Railway Age, September 1996, p. 55. 166. B. L. Bramfitt, R. L. Cross, and D. P. Wirick, Proc. of the International Symp. on Rail Steels for 21st Century, Iron and Steel Society, Inc., Warrendale, Pa., 1995, pp. 23–29. 167. B. L. Bramfitt, D. P. Wirick, and R. L. Cross, Iron & Steel Engineer, vol. 75, no. 4, 1996, pp. 33–36. 168. D. Utrata, Rail Steels Symp. Proceedings, Iron and Steel Society, Inc., Warrendale, Pa., 1994, pp. 131–135. 169. B. L. Bramfitt, Mechanical Working and Steel Processing Proceedings Conference, Cincinnati, Ohio, 1990, pp. 485–495. 170. Carbon and Alloy Steels, ASM International, Materials Park, Ohio, 1996, pp. 169–200, 684–687.

PEARLITE AND PROEUTECTOID PHASES

7.69

171. AREA Manual for Railway Engineering, 1996, pp. 4-2-6 to 4-2-10. 172. C. Esvald, Modern Railway Track, Thyssen Stahl, Duisburg, Federal Republic of Germany, 1989. 173. D. M. Fegredo and J. Kalousek, Wear of Materials, ed. K. C. Ludema, American Society of Mechanical Engineers, New York, 1987, pp. 121–132. 174. R. K. Steele, 1990 Mechanical Working and Steel Processing Proceedings, 1990, pp. 131–142. 175. C. O. Frederick, Rail Quality and Maintenance for Modern Railway Operation, International Conference, Delft 1992, Kluwer Academic Publishers, Dordrecht, Netherlands, 1993, pp. 3–14. 176. O. Orringer, Rail Quality and Maintenance for Modern Railway Operation, International Conference, Delft 1992, Kluwer Academic Publishers, Dordrecht, Netherlands, 1993, pp. 253–271. 177. L. Reid, Rail Quality and Maintenance for Modern Railway Operation, International Conference, Delft 1992, Kluwer Academic Publishers, Dordrecht, Netherlands, 1993, pp. 337–347. 178. P. Clayton and X. Su, Wear, vol. 200, 1996, pp. 63–73. 179. R. J. Glodowski, Metals Handbook, vol. 1: Properties and Selection: Iron, Steel, and High Performance Alloys, 10th ed., ASM International, Materials Park, Ohio, 1990, pp. 272–276. 180. Steel Products Manual: Carbon Steel Wire and Rods, Iron and Steel Society, Inc., Warrendale, Pa., 1993. 180a. Z. Muskalski, J. W. Pilarczyk, H. Dijja, and B. Golis, Wire J. Int., December 1999, pp. 108–113. 181. Wire Rope User’s Manual, 3d ed., Wire Rope Technical Board, Baltimore, Md., 1993. 182. Ferrous Wire, vol. 2, The Wire Association International, Inc., Guilford, Conn., 1989. 183. F. L. Jamieson, Metals Handbook, vol. 11: Failure Analysis and Prevention, ASM, Metals Park, Ohio, 1986, pp. 514–528. 184. 1998 SAE Handbook, vol. 3: On-Highway Vehicles and Off-Highway Machinery, Society of Automotive Engineers, Warrendale, Pa., 1998, pp. 30.01–30.78. 184a. X. Zhang and S. V. Subramanian, Wire J. Int., December 1999, pp. 102–107. 184b. G. M. Faulring, I&SM, July 1999, pp. 29–36. 185. A. B. Dove, Metals Handbook, vol. 1: Properties and Selection: Iron, Steel, and High Performance Alloys, 10th ed., ASM International, Materials Park, Ohio, 1990, pp. 276–288. 185a. I. Ochiai, S. Nishida, and H. Tashiro, Wire J. Int., December 1990, pp. 50–61. 186. R. Lagneburg, O. Sandberg, and W. Roberts, in Fundamentals of Microalloyed Forging Steels, eds. G. Krauss and S. K. Banerji, TMS, Warrendale, Pa., 1987, pp. 39–54. 187. J. F. Held, in Fundamentals of Microalloyed Forging Steels, eds. G. Krauss and S. K. Banerji, TMS, Warrendale, Pa., 1987, pp. 39–54. 188. F. D. Gear, in Fundamentals of Microalloyed Forging Steels, eds. G. Krauss and S. K. Banerji, TMS, Warrendale, Pa., 1987, pp. 291–296.

CHAPTER 8

MARTENSITE

8.1 INTRODUCTION The name martensite was first used by Osmond in 1895, in honor of German metallurgist Adolf Martens, to identify the very hard, platelike or acicular constituent produced in many steels rapidly quenched from the austenite state. Martensite in steel is a metastable body-centered tetragonal (bct) phase. Later, this name was extended to include a number of other solid-state transformations in pure metals, nonferrous alloys (e.g., Cu-Al-, Cu-Sn-, and Ti- and Zr-base alloys), semiconductors, ceramics, minerals, superconducting compounds (such as V3Si and Nb3Sn just above the superconducting transition temperature), solidified gases (such as oxygen and helium), proteins, and polymeric materials which exhibit some common features, notably diffusionless displacive (shear) transformation.1,2 It is now common to describe all such transformation processes as martensitic and the product phase as martensite, irrespective of its crystal structures. Different types of martensitic transformation in both ferrous and nonferrous alloys (e.g., athermal or diffusionless; isothermal or diffusional; interstitial; substitutional; etc.) have now been recognized which may be distinguished from one another by kinetics, morphology, and crystallography or internal structure. Table 8.1 provides such a classification of diffusionless displacive transformation in metallic materials, where the alloy systems are grouped into three categories.3 The following definition has been given: A martensitic transformation involves the coherent formation of one phase from another without change in composition by a diffusionless, homogeneous lattice shear.4 In this case, diffusion means long-range diffusion. Christian coined the term military transformation for this type of reaction in which the most orderly and highly coordinated or disciplined atomic rearrangement occurs where every atom has the same neighbors as in the parent but where the individual atoms move by a fraction of an interatomic distance.5,6 It has been defined by Cohen as a subset of diffusionless, displacive phase transformation, involving sufficiently large lattice-distortive shear displacements where strain energy dominates the transformation kinetics and product morphology.7 Martensitic transformation is classified as a nonequilibrium phase transformation which always occurs far away from equilibrium and which results in nonequilibrium products.8 Martensitic transformation involves a cooperative motion of a set of atoms across an interface, causing a shape change and sound.9 Thus, martensitic transformation can be defined as diffusionless, lattice-distortive, shear-dominant transformation occurring by nucleation and growth.10 It is this class of solid-state transformation and its products that constitutes the focus of this chapter. 8.1

CHAPTER EIGHT

8.2

TABLE 8.1 Classification of Metallic Alloy Systems Showing Diffusionless Displacive Transformations3 1. Martensite based on allotropic transformation of solvent atom 1. Iron and iron-based alloys 2. Shear transformation, close packed-to-close packed 1. Cobalt and alloys fcc Æ hcp, 126 R 2. Rare earth and alloys fcc, hcp, dhcp, 9R 3. MnSi, TiCr2 NaCl Æ NiAs, Laves 3. Body centered cubic to close packed 1. Titanium, zirconium and alloys bcc Æ hcp, orth fcc 2. Alkali and alloys (Li) bcc Æ hcp 3. Thallium bcc Æ hcp 4. Others: plutonium, uranium, mercury, Complex structures etc., and alloys

SF†

tw, d†

2. b-bcc Hume-Rothery and Ni-based martensitic shape-memory alloys 1. Copper-, silver-, gold-, b-alloys (disord., ord.) bcc 2. Ni-Ti-X b-alloys Nickel b-alloys (Ni-Al) Ni3-xMxSn (M = Cu, Mn) (Cobalt b-alloys, Ni-Co-X)

AB, ABABCBCAC, ABAC bcc Æ 9R, AB tw, SF† bcc Æ ABC tw, SF† bcc Æ AB tw*

3. Cubic to tetragonal, stress-relaxation twinning or martensite 1. 2. 3. 4.

Indium-based alloys Manganese-based alloys A 15 compounds, LaAgxIn1+x Others: Ru-Ta, Ru-Nb, YCu, LaCd

fcc Æ fct, orth. fcc Æ fct, orth. b-W Æ tetr.

tw, tws† tws†

† SF: stacking faults; tw: twins; tws: (stress relaxation) twins; d: dislocated. Courtesy of L. Delaey.

In steels, a large proportion of hardness and strength may be developed in martensite in which distortion produced by forcibly retaining the carbon atoms within the ferrite lattice is intense and unidirectional. In substitutional martensite (e.g., carbon-free Fe-30% Ni and nonferrous alloys), less hardening is developed11 because of the small and omnidirectional distortion. It should be emphasized that the properties of some martensites are of vast technological importance. These include conventional hardening of steel, a type of age hardening (maraging); ductility increase in transformation-induced plasticity (TRIP) steels; partially stabilized zirconia (PSZ); rubberlike elastic ductility; shape memory effects; and high damping capacity.12 This chapter deals with the characteristics, types, and products of martensitic transformations, phenomenological theory of crystallography martensite, morphology, nucleation and growth of the transformation, shape memory alloys, and strengthening mechanisms. Finally, omega transformation, another type of diffusionless, displacive solid-state transformation, is briefly described.

MARTENSITE

8.3

8.2 GENERAL CHARACTERISTICS OF MARTENSITIC TRANSFORMATION The main characteristics of martensitic transformation and of the product phase are summarized below: 1. It is a diffusionless transformation, which means that the chemical compositions of the parent and product phases are identical. 2. The transformation interface between the martensite and the parent phase depends greatly on the transformation growth process. Such an interface remains highly glissile and does not tend to require thermal activation for its movement, as verified from low-temperature experiments. This interface may be completely coherent or semicoherent, depending upon the crystallography of the particular material undergoing transformation. In the case of most ferrous martensites, the interface remains semicoherent and the product and parent lattices are coherently accommodated over a small portion of the boundary, producing an accumulating misfit. In contrast, in fcc Æ hcp transformations in Co and its alloys, the austenite/martensite interface is fully coherent.1,2 3. Shape change and surface relief. When martensite is formed, a macroscopic deformation (or shape change) is observed which results in surface upheavals, called surface relief, on a polished flat surface of the parent phase (Fig. 8.1), indicating that the transformation occurs by a displacive shear parallel to the habit plane (i.e., the interface plane between the parent and product phase), which is an invariant plane (i.e., an undistorted and unrotated plane like the K1 plane in twinning). The habit plane is usually expressed as a plane in the parent phase and is of special importance in the crystallography of martensite. This observation of surface relief is one of the main experimental criteria of a martensitic transformation. The surface in the transformed region remains plane but is tilted about its line of intersection with the habit plane. Straight lines inscribed on the surface are

FIGURE 8.1 Schematic shape deformation produced during the formation of a martensite plate.13 (Courtesy of The Institute of Metals, England.)

8.4

CHAPTER EIGHT

FIGURE 8.2 (a) A simple shear strain occurring in twinning; (b) invariant plane strain of martensite (P1 = 1 + m1d1P¢1) comprising a simple shear component (mP1 dP1 ) and tensile component (mn1 P1).

transformed into straight lines, and planes are transformed into planes.13,14 Such a transformation is said to be homogeneous. Sometimes it is described mathematically as an affine transformation. The strain that produces a net macroscopic deformation associated with an invariant habit plane is termed an invariant plane strain (IPS) deformation. In this case the displacement of any point is proportional to its distance from the invariant plane. In martensitic transformations, the shear that produces the invariant plane strain is complex, consisting of (1) simple shear strain and (2) uniaxial tensile or compression strain normal to the habit plane (Fig. 8.2); the normal component of strain is attributed to the volume change produced during structural change in martensitic transformation. In contrast, merely the simple shear strain occurs in twinning. Table 8.2 illustrates two strain components for several martensites.13,14 4. Martensite occurs usually in the form of plates or laths which seem to be embedded in the matrix along certain well-defined planes (Fig. 8.3). The martensite (sometimes designated as a ¢) plate, forming closer to room temperature, is usually of lenticular shape which is partially internally twinned, but has a core, called the midrib, within it [as a result of constraints imposed by the untransformed matrix (Fig. 8.1)]. On two-dimensional metallographic sections, the lenticular platelets lead to the characteristic microstructure. The plates forming at very low temperatures are usually thin and flat. Thin plate is fully internally twinned of the {112} type, a common deformation twinning in bcc crystals. However, in low-carbon steels and dilute iron alloys, martensites form with lathlike morphology (i.e., with planar interface extending entirely across the parent grains). It has been suggested that a¢ crystals form initially at the midrib plane and grow laterally.15 5. The habit plane is usually irrational (i.e., the Miller indices are not very simple) in almost all cases. This fact is best illustrated by the experimentally determined habit planes in three steels on the stereographic triangle shown in Fig. 8.4. There is a considerable amount of scatter in the experimental data for the habit plane,14 but martensite plates in a particular alloy possess a unique or definite habit plane.

MARTENSITE

8.5

TABLE 8.2 Crystallographic Strain Component Data for Several Martensites13,14

System

Structure change

Habit plane

Direction displacement

Shear strain component g T

Normal strain component en

0.19

0.09

0.05

Fe-C (1.35% C)

fcc Æ bct

~{225}

Fe-C (1.8% C)

fcc Æ bct

~{259}

Fe-Ni (30% Ni)

fcc Æ bcc

~{9, 22, 33}

~

0.20

Fe-Ni-C (22% Ni, 0.8% C)

fcc Æ bct

~{3, 10, 15}

~

0.19

~{8, 9, 12}

~

0.22

0.70 -0.69 0.21

0.66 0.73 0.18 ~

Pure Ti

bcc Æ hcp

Au-Cd (47.5% Cd)

bcc Æ Orthorhombic

冦 冧

In-Tl (20% Tl)

fcc Æ fct

{011}

冬 冭

0.05 0.02

Reprinted by permission of John Wiley & Sons, Inc, New York.

FIGURE 8.3 Optical micrographs. (a) Lath martensite formed in 1018 steel by water-quenching from 925°C (1697°F). (b) Plate martensite in 1060 steel by water-quenching from 815°C (1500°F).

It has been found that the habit plane is a function of composition and temperature rather than strain. In nearly pure iron or in lower-carbon steels, the crystals of martensite appear to be needle- or lath-shaped in cross section with habit planes of {111}g or {112}g type, for carbon steels containing 0.5 to 1.4% carbon,16 and the usual plane observed is near {225}g and is formed at higher temperatures. In carbon

8.6

CHAPTER EIGHT

FIGURE 8.4 Experimentally measured habit planes of martensite plates on the stereographic triangle.14 (Reprinted by permission of John Wiley & Sons, New York; after Greninger and Troiano, and Dunne and Bowles.)

steels of still higher carbon content (1.5 to 1.8% C), the habit plane may be approximately {259}g. 6. Lattice orientation relationships. Associated with the habit planes, there is always a precise orientation relationship between parent phase and martensite.4 The Kurdjumov-Sachs (K-S) relation17 is {111}g // {011}a¢ with g // a¢ (a¢ = martensite) and is usually associated with {225}g habit plane. This relationship has been found to occur in Fe-C alloys with carbon content in the range of 0.5 to 1.4 wt% (Table 8.3).3 The Nishiyama (N) relation, similar to the Greninger-Troiano (G-T) relation,18,19 is again {111}g // {011}a¢ but with g // a¢ and is generally associated with an ª {259}g habit plane. The G-T orientation relation is an irrational relation between K-S and N and corresponds to the prediction of the IPS crystallographic theory. This relationship has been found to occur in Fe-C alloys with carbon content >1.4 wt%, which was observed by Greninger and Troiano in Fe-22 wt% Ni-0.8 wt% C alloy and by Nishiyama for Fe-Ni alloys containing 27 to 34% Ni18,19 (Table 8.3).3 As regards the transformation of fcc (austenite) Æ hcp martensite and that of hcp Æ bcc martensite, the following relations are observed:3

(111)g // (0001)e // (101)a ¢

and

[110]g // [1210]e // [111]a ¢

(8.1)

It is evident from the K-S relationship that the close-packed planes and closepacked directions, respectively, of the g lattice are parallel to those of the a ¢ lattice. Moreover, this direction is parallel to the Burgers vector. In the K-S relations, {111} represents any four kinds of austenite planes, namely, (111), (1¯11), (11¯1), or (111¯). In each plane, any one of six different directions can exist, as shown in Fig. 8.5a. These directions consist of three pairs in one direction and three pairs in the opposite direction. These pairs of crystals are twin-related.

MARTENSITE

8.7

TABLE 8.3 Crystallographic Observable Parameters of the Martensitic Transformations in Some Metals and Alloys3 Alloy system Fe-C

Structural change fcc Ø bc tetr.

Composition, wt% 0–0.4% C 0.55–1.4% C 1.4–1.8% C

Fe-Ni

Orientation relationship (111)p//(101)M [11¯0]p//[111¯]M K-S relationship K-S relationship Idem

(111)P (225)P

fcc Ø bcc

27–34% Ni

Fe-C-Ni

fcc Ø bc tetr.

0.8% C, 22% Ni

(111)p ª 1° of (101)M (12¯1)p ª 2° of [101¯]M G-T relationship

(3, 10, 15)P

Fe-Mn

fcc Ø hcp (e-phase)

13–25% Mn

(111)p//(0001)e [11¯0]p//[12¯10]e

(111)P

Fe-Cr-Ni

fcc Ø

18% Cr, 8% Ni

(111)p//(0001)e//(101)a ¢ [11¯0]p//[12¯10]e//[111¯]a ¢

e(111)P a¢(211)P

Austenite inox. iron

hcp (e), bcc (a¢)

Cu-Zn b

bcc Æ 9R

40% Zn

ª(2, 11, 12)P

Cu-Sn

idem

25.6% Sn

(011)p//?(1¯1¯4)M [11¯1]p//?[1¯10]M

Cu-Al

bcc Ø hcp distorted

11–13.1% Al

(101¯)P at 4° of (0001)M [111]P//[101¯0]M (101¯)P//[101¯1]M [111]P//[101¯0]M

2° of (133)P

12.9–14.7% Al

(111)p//[101]M [12¯1]p//[101¯]M N relationship

Habit plane

ª(259)P

ª(133)P

3° of (122)P

Pure Co

fcc Ø hcp

(111)P//(0001)M P//M

(111)P

Pure Zr

bcc Ø

(101)P//(0001)M [111]P//[112¯0]M

(596)p (8, 12, 9)P

Pure Ti

hcp

Pure Li

(334)P (441)P Burgers relations

Source: Courtesy of G. Guénin et al.; after P. F. Gobin, G. Guénin, M. Morin, and M. Robin, in Transformations de Phases à l’ État Solide-Transformations Martensitiques. Lyon: Dep. Génie Phys. Mat., INSA, 1979.

Therefore, K-S relations yield a¢ crystals with 4 ¥ 6 = 24 different orientations in a g crystal. These different oriented martensite crystals are termed variants. In the N relations there are four types of austenite planes, and three different directions exist in each plane, as shown in Fig. 8.5b. Thus N relations lead to 4 ¥ 3 = 12 variants, i.e., half of those in the K-S relations. It should be pointed out that all these relationships are not precise and that they represent a deviation of 1° (or

[1-0 1]g

[10 1]g

-1]g [01

- 1]g [01

[112]g

CHAPTER EIGHT

8.8

[121]g

-[211]g

-

[110]g (b)

[110]g (a)

FIGURE 8.5 Directions of shears in (111)g plane: (a) K-S relationship; (b) N relationship.15 (Reprinted by permission of Academic Press, Orlando, Florida.)

TABLE 8.4 The Ms Temperature and Approximate Hardness of Martensite Product for a Number of Materials2 Composition

Ms / K

Hardness HV

ZrO2 Fe-31Ni-0.23C wt% Fe-34Ni-0.22C wt% Fe-3Mn-2Si-0.4C wt% Cu-15Al Ar-40N2

1200 83 34 J/cm2

400 300 0.1

1

10

100

1000

10000

Aging time, hr (b)

FIGURE 10.32 Time-temperature-transformation diagrams (a) for S31803 (2205) showing both s and a¢ phases, and (b) for some duplex grades showing impact strength lower than 34 J/cm2 (278-J full-size specimen),196a and (c) showing the effect of alloying additions on precipitation reactions in duplex stainless steel.53 [(a): After Lacombe et al., eds., Stainless Steels, Chapter 18, p. 624, Les Editions de Physiques, Les Ulis Cede A, France, 1993.]

CHAPTER TEN

10.94

Mo, W, Si

825

M7C3 carbide, CrN nitride HAZ σ phase Cr2N nitride χ phase γ2 phase M23C6 carbide R phase

Cr, Mo, W, Si

650

π phase ε phase (Cu) α' phase G phase

Cr, Mo, Cu, W 475

300

1832

1514

1202

887

Temperature, °F

Temperature, °C

1000

572 Cr, Mo, Cu, W Time (c)

FIGURE 10.32 (Continued) Time-temperature-transformation diagrams (a) for S31803 (2205) showing both s and a¢ phases, and (b) for some duplex grades showing impact strength lower than 34 J/cm2 (278-J full-size specimen),196a and (c) showing the effect of alloying additions on precipitation reactions in duplex stainless steel.53 [(a): After Lacombe et al., eds., Stainless Steels, Chapter 18, p. 624, Les Editions de Physiques, Les Ulis Cede A, France, 1993.]

the TTT diagram for some duplex grades which have led to impact strength less than 34 J/cm2 (27-J full-size specimen). Because of this risk of embrittlement, duplex grades are not provided in equipment with design temperatures above about 300°C (575°F).196a In addition to a¢ phase, G phase has been implicated in the embrittlement of duplex alloys, as reported by Miller and Alexander. There is also a localized R-phase formation at an early stage of aging which causes extreme loss of Charpy impact toughness before the embrittlement attributed to the s-phase formation.197 Figure 10.32c shows the TTT diagram illustrating the phases formed, approximate temperature ranges of their formation, and the effect of alloying elements on transformation kinetics.53 10.7.2

Mechanical Properties

In general, the high yield strength of duplex stainless steels (about 2 to 3 times greater than that of the austenitic steels—400 to 500 versus 200 to 250 MPa, or 58 to 80 versus 29 to 36 ksi) provides designers with the use of thin-wall material with sufficient load-bearing and pressure-containing capacity. This high yield strength factor can cause marked reduction in weight and welding time.† The elongation of duplex steels, although adequate for most service conditions and for fabrication, is lower than that of the austenitic steels. † Note: But the ASME pressure vessel code is based on the tensile strength. This reduces the advantages of duplex alloys in North America. Many European design codes are yield strength-based. That is one reason why duplex alloys are more popular in Europe.51

AUSTENITE

10.95

TABLE 10.13 Room-Temperature Mechanical Properties for Some Duplex Stainless Steels per ASTM A 79053 Minimum yield strength

Minimum tensile strength

UNS no.

MPa

ksi

MPa

S31200 S31500 S31803 S32304 S32550 S32750 S32760† S32900 S32950

450 440 450 400 550 550 550 485 480

65 64 65 58 80 80 80 70 70

690 630 620 600 760 800 750 620 690

†

Hardness

ksi

Elongation (minimum), %

HB

HRC

100 92 90 87 110 116 109 90 100

25 30 25 25 15 15 25 20 20

280 290 290 290 297 310 200–270 271 290

... 30.5 30.5 30.5 31.5 32 ... 28 30.5

Not listed in ASTM A790.

Because of their high yield strengths, duplex steels experience severe difficulties in cold forming when compared to the austenitic stainless steels. Further, since many high-temperature embrittlement phenomena can occur in these alloys, forging and other hot working operations need more attention than those with the austenitic stainless steels using the same processing operations.53 Table 10.13 lists the roomtemperature tensile properties and hardnesses of selected duplex stainless grades per ASTM A790. The nonwelded base metal toughness, as expressed by the ductile-to-brittle transition temperature of the duplex steel, lies between the austenitic and ferritic grades; the extent of toughness is a function of ferrite content, the orientation of austenite-ferrite band, and the cooling rate from the annealing temperature. These alloys are not suitable for cryogenic applications. Most alloys contain about 50% ferrite to maintain a fairly good toughness because ferrite content exceeding 60 to 70% decreases the CVN energy sharply. Maximum toughness is achieved if impact tests are run with the crack growth transverse to the banded structure. Rapid cooling from the annealing temperature produces greater toughness, while slower cooling or holding at intermediate temperature, in the range of 400 to 500°C (750 to 930°F) and above 700°C (1290°F), gives rise to a variation in embrittlement due to precipitation of a¢ phase (475°C or 885°F embrittlement) and s-phase particles. 10.7.3 Applications The total tonnage of duplex stainless steels currently produced in the world is estimated to be 0.5%) and 0.5% Mo as well as stress-relieving at 650°C (1200°F) appear to prevent or reduce the tendency of graphitization. 8. Presence or formation of small voids, during rolling, at the cementite particles present in the ferrite matrix plays a vital role in nucleating graphitization. Inokuti has proposed the mechanism of graphite nucleation at voids, augmented by the formation of CO gas in the voids. This was supported by the experimental observations made by Berge et al.33

20 mm

50 mm

(a)

(b)

FIGURE 12.10 Optical photomicrograph of flat stock for 0.035-in. cold-rolled 1074 spring pin showing graphitization produced at subcritical temperature. The dark-inclusionlike features are almost pure carbon as confirmed by scanning electron microscopy/energy disperse x-ray. Picral etch. (a) Low magnification and (b) high magnification.

BASIC HEAT TREATMENT

12.23

9. Localized graphitization near a welded joint of steam piping or superheated tube appears to be much more damaging than general uniform graphitization because the former apparently produces notches that concentrate stress, provides a weak continuous path for fracture, and reduces load-bearing capability.36

12.6 NORMALIZING Normalizing consists of heating the steel to a temperature 40 to 50°C above the upper critical points (that is, the Ac3 or Accm temperatures), holding it there for a period depending on the dimensions of the part and type of steel being treated until it is completely austenitized, followed by cooling, in still air, free of drafts, to room temperature. (Because the microstructure is a function of cooling rate, accelerated cooling such as air blowing or mist cooling is sometimes used to achieve a finer ferrite-pearlite or bainite microstructure to improve the mechanical properties of thick sections of steel products.) This term originated from the wrought steel industry, where air cooling was the normal practice. The temperature range representing the normalizing of a plain carbon steel is shown in Fig. 12.1. The minimum period for holding (or soaking) at the austenitizing temperature is 15 min, necessitating longer periods with larger sections. Table 12.4 lists the typical normalizing temperatures for standard carbon and low-alloy steels.36a Figure 12.11 illustrates the microstructure of a normalized plain carbon steel. The main purpose of this operation is to (1) refine the grains of a steel bar, forging, welding, or casting that have become coarse grained because of being heated to a high-temperature treatment; (2) homogenize, that is, improve the uniformity of microstructure, by breaking up the coarse nonuniform structure (e.g., cast dendritic structure), although at a lower temperature and for shorter periods than those used for homogenizing; (3) allow smoother machining, thereby producing superior surface finish of the normalized product compared to that of the annealed one; and (4) produce harder and stronger steel than that of full annealing. The majority of ordinary engineering steels do not form martensite when normalized. However, martensite can be produced in the case of highly alloyed steels; such steels are called air-hardening steels, and, in the true sense, it is not proper to call the air-hardening operation normalizing because it does not put the steel in the “normal” pearlitic conditions. Since the cooling rate lies between that used for quenching and annealing, respectively, the hardness and strength resulting from this treatment will be somewhat less than if quenched and somewhat higher than if annealed. The hardness of the normalized product depends on the composition and dimensions of the steel. The difference in the cooling rate between the surface and center during this treatment is small for light sections and large for heavy sections. Since air cooling does not represent the equilibrium condition, the Fe-Fe3C phase diagram cannot be employed to predict the proportions of proeutectoid ferrite and pearlite or proeutectoid cementite and pearlite which will be present at room temperature. In fact, less time is available for the formation of proeutectoid constituents; consequently, a lesser proportion of the proeutectoid constituent will form in normalized steels than in annealed ones. Moreover, the faster cooling rate in normalizing will, in general, depress the temperature of austenite decomposition and will produce a finer pearlite.37 Figure 12.2 shows the schematic time-temperature cycle for normalizing superimposed on the TTT diagram.

TABLE 12.4 Typical Normalizing Temperatures for Standard Carbon and Low-Alloy Steels.36a Temperature† Grade

°C

°F

Plain carbon steels 1015 1020 1022 1025 1030 1035 1040 1045 1050 1060 1080 1090 1095 1117 1137 1141 1144

915 915 915 900 900 885 860 860 860 830 830 830 845 900 885 860 860

1675 1675 1675 1650 1650 1625 1575 1575 1575 1525 1525 1525 1550 1650 1625 1575 1575

Standard alloy steels 1330 1335 1340 3135 3140 3310 4027 4028 4032 4037 4042 4047 4063 4118 4130 4135 4137 4140 4142 4145 4147 4150 4320 4337 4340 4520 4620 4621 4718 4720

900 870 870 870 870 925 900 900 900 870 870 870 870 925 900 870 870 870 870 870 870 870 925 870 870 925 925 925 925 925

1650 1600 1600 1600 1600 1700 1650 1650 1650 1600 1600 1600 1600 1700 1650 1600 1600 1600 1600 1600 1600 1600 1700 1600 1600 1700 1700 1700 1700 1700

Temperature† Grade

°C

°F

4815 4817 4820 5046 5120 5130 5132 5135 5140 5145 5147 5150 5155 5160 6118 6120 6150 8617 8620 8622 8625 8627 8630 8637 8640 8642 8645 8650 8655 8660 8720 8740 8742 8822 9255 9260 9262 9310 9840 9850 50B40 50B44 50B46 50B50 60B60 81B45 86B45 94B15 94B17 94B30 94B40

925 925 925 870 925 900 900 870 870 870 870 870 870 870 925 925 900 925 925 925 900 900 900 870 870 870 870 870 870 870 925 925 870 925 900 900 900 925 870 870 870 870 870 870 870 870 870 925 925 900 900

1700 1700 1700 1600 1700 1650 1650 1600 1600 1600 1600 1600 1600 1600 1700 1700 1650 1700 1700 1700 1650 1650 1650 1600 1600 1600 1600 1600 1600 1600 1700 1700 1600 1700 1650 1650 1650 1700 1600 1600 1600 1600 1600 1600 1600 1600 1600 1700 1700 1650 1650

† Based on production experience, normalizing temperature may vary from as much as 27°C (50°F) below, to as much as 55°C (100°F) above, indicated temperature. The steel should be cooled in still air from indicated temperature. Reprinted by permission of ASM International, Materials Park, Ohio.

BASIC HEAT TREATMENT

a

12.25

b

FIGURE 12.11 Microstructure of a normalized 1025 steel. (a) Produced by austenitizing at 1095°C (2000°F) and then air-cooling. Coarse grain structure; pearlite (black areas) in ferrite matrix (white areas). (b) Produced by austenitizing at 925°C (1700°F) followed by air cooling. Fine-grain structure. Picral etchant. (Reprinted by permission of ASM International, Materials Park, Ohio.)

The best practice of cooling in air is to suspend the steel parts in air, placing them on a special cooling bed in order for them to be surrounded by the cooling action of the air. Careless practice, such as dumping the material into a pile on the shop floor, may cause very poor treatment and very nonuniform structures. Normalizing is frequently applied prior to spheroidize-annealing to disperse carbides homogeneously. It is also done as a prior heat treatment before quenching and tempering of thick sections. Normalizing often precedes hardening and annealing operations of cast steels of various cross-sectional sizes. Normalizing the part prior to hardening tends to minimize cracking upon quenching.38 Spheroidization annealing and normalizing may be used to improve machinability; the method to be chosen depends on the carbon content. Normalizing is not usually employed for hypereutectoid steels, except as a pretreatment prior to subsequent annealing. The treatment is generally confined to in-process treatments performed by steel suppliers and is seldom required by tool manufacturers. Effect of Hot Working. The normalized microstructure is a function of the prior hot-working conditions. The finer the microstructure of the as-rolled microstructure, the finer the austenite grain size after reheating for normalizing. Hot-rolling at lower temperature corresponding to the nonrecrystallized austenite leads to a finer normalized austenite grain size. The reduction ratio during hot rolling has also a significant bearing on the austenite grain size during normalizing. Figure 12.12 shows that the reduction ratio is more important in improving toughness than is the cooling rate during normalizing. A large hot-rolling reduction and slow cooling during normalizing lower the Charpy ductile-brittle transition temperature [fracture appearance transition temperature (FATT)] more than a small reduction ratio and fast cooling during normalizing do.39 Double Normalizing. In some cases a double normalizing treatment is used. This treatment consists of first heating to a temperature some 50 to 100°C above the usual temperature in order to produce complete dissolution of the constituents and homogeneous austenite grain size. The second normalizing treatment is accom-

12.26

CHAPTER TWELVE

FIGURE 12.12 Effect of cooling rate during normalizing and the hot-rolling reduction ratio of the (0.05C-3.5Ni-0.1Mo) steel slab on the ductile-brittle transition temperature (FATT) measured by the Charpy impact test. The cooling rate is represented by the corresponding plate thickness. The heat treatment involved normalizing from 840°C, tempering at 600°C followed by stress-relieving at 580°C and cooling to 300°C at a rate of 100°C/hr.8

FIGURE 12.13 Variation of the austenite grain size with double normalizing in forged carbon steel containing 0.55% C.8

plished near the lower limit of the temperature range to produce a fine-grained structure. It is usually applied to carbon and low-alloy steels of large dimensions where too-high forging temperatures have been used. Figure 12.13 illustrates the change in austenite grain size during double normalizing treatment in forged medium-carbon (0.55% C) steels where the first normalizing lies between 850 and 1100°C and the second one at 850°C.8 Repeated normalizing (e.g., double normalizing) treatment is necessary for very coarse-grained forgings because it may be difficult to obtain a fine-grained structure in a single normalizing treatment.

BASIC HEAT TREATMENT

12.27

12.7 ANNEALING AND NORMALIZING OF CAST IRONS 12.7.1 Gray Iron 12.7.1.1 Annealing. Next to stress relieving, annealing treatment is occasionally applied to gray iron. Annealing of gray iron involves heating it to a temperature that is high enough to soften it and/or to minimize or remove eutectic carbides, thereby improving its machinability. Annealing significantly decreases the mechanical properties. In fact, it reduces the grade level approximately to the next-lower grade. Figure 12.14 illustrates the effect of annealing on the tensile strength of class 30 gray iron arbitration bars. The extent of decrease in properties is a function of the annealing temperature, the time at temperature, and the chemical composition of the gray iron. Annealing treatments commonly applied to gray iron include high-temperature (graphitizing) annealing, medium-temperature (“full”) annealing, and lowtemperature (ferritizing) annealing.40,41 High-Temperature (Graphitizing) Annealing. Graphitizing annealing produces the ultimate decomposition of chilled iron, massive, primary or free cementite, and reduction in strength and hardness. This annealing treatment consists of (1) heating to a temperature of 900 to 954°C (1650 to 1750°F) [however, at 925°C (1700°F) and above, the phosphide eutectic present in irons containing 0.10% P or more may melt]; (2) holding at this temperature for a period ranging from a few minutes to several hours; and (3) cooling at a rate depending on the final use of the iron. If it is desired to break down primary carbides and retain a pearlitic structure, and therefore a maximum strength and wear resistance, the casting should be air-cooled from the furnace to about 538°C (1000°F) to produce pearlite-graphite microstructure. If maximum machinability is the main concern, the casting should be furnace-cooled through the critical range to 538°C (1000°F) to obtain a ferrite-graphite structure. In both cases, cooling from 538°C (1000°F) to about 290°C (550°F) at no more than 110°C/hr (200°F/hr) should be employed to avoid the formation of residual stresses. Medium-Temperature (Full) Annealing. Full annealing in the absence of massive carbides or in the presence of small amounts of well-dispersed carbides can be performed by heating to just above the critical range between 790 and 900°C (1450 and 1650°F), depending primarily on the silicon content arising from the increase of critical temperature with the silicon content.41 The soaking time in the furnace is the same as for ferritizing annealing but is shorter than for graphitizing

FIGURE 12.14 Effect of annealing on tensile strength of class 30 gray iron. Specimens were arbitration bars from 31 heats. Bars were annealed at 925°C (1700°F) for 2 hr plus 1 hr/in. (25 mm) of section over 1 in., then cooled at a maximum rate of 160°C (285°F)/hr from 925 to 565°C (1700 to 1050°F). Cooling then continued from that level at a maximum rate of 130°C (230°F)/hr to 200°C (390°F); subsequently the bars were air-cooled to room temperature.40

CHAPTER TWELVE

12.28

annealing. This is followed by slow cooling through the critical temperature range from about 790 to 675°C (1450 to 1250°F) to allow the combined carbon to precipitate as graphite. Full annealing is used when a ferritizing annealing would be ineffective due to high alloy content of a particular iron. This treatment can also be used for pearlitic irons containing moderate amounts of Cr, V, Mo, or a higher than standard Mn content, but without free cementite, to obtain an entirely ferritic matrix.41 Low-Temperature (Ferritizing) Annealing. For unalloyed or low-alloy gray irons of normal composition with no free cementite, when the only requirement is to convert the pearlitic carbide to ferrite and graphite for improved machinability, it is advisable to ferritize anneal these castings by heating near the lower transformation temperature [i.e., between 704 and 760°C (1300°F and 1400°F)] and holding for approximately 1 hr/in. of section thickness, followed by slow cooling [that is, 55°C/hr (100°F/hr)].41 There is no significant influence of temperature, up to 595°C (1100°F), on the structure and hardness of gray iron. However, the rate of decomposition of iron carbide into ferrite plus graphite increases markedly with the increase in temperature above 595°C (1100°F), reaching a maximum of 760°C (1400°F) for unalloyed or low-alloy gray iron.40,42 Table 12.5 lists the effect of ferritizing anneal on tensile strength and hardness due to varying alloying addition.40,42 Figure 12.15 shows the conversion of an as-cast pearlite structure of unalloyed gray iron (with 980 BHN hardness) to ferrite and graphite (with 120 BHN hardness) by ferritizing annealing at 760°C for 1 hr.40,42 For unalloyed irons, the rate of ferritization depends on the silicon content and the temperature employed. For example, when the unalloyed iron contains ⬃2% Si and is annealed at 760°C (1400°F), the rate of ferritization becomes maximum, reaching 90% of conversion within 20 to 30 min in light sections. However, this reaction is retarded by the presence of alloying elements such as Cr, Ni, Cu, and Mo; however, Mn in the range of 0.3 to 0.98% at a constant sulfur level is not a matrix strengthener in gray iron.43 As the temperature is lowered, the rate of ferritization decreases drastically; and below 650°C (1200°F), the rate of pearlite conversion to ferrite is so slow as to require an excessively long holding time for the completion of the process.41 12.7.1.2 Stress-Relieving. Invariably, as-cast gray irons contain residual stresses, the magnitude of which is a function of different parameters such as casting methods employed, composition, and properties of the cast material, as well as shape and

a

b

FIGURE 12.15 (a) As-cast unalloyed pearlitic gray iron structure (180 BHN). (b) Ferrite and graphite structures (120 BHN) produced by ferritizing annealing of (a) at 760°C (1400°F) for 1 hr.40 (Reprinted by permission of ASM International, Materials Park, Ohio.)

TABLE 12.5 Effect of Ferritizing Annealing at 760°C (1400°F) on Hardness and Tensile Strength of Alloyed Gray Irons40 As-cast Gray iron no.

12.29

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Alloy additions to base iron, % Cr 0.61 0.47 0.56 0.50 0.49 ... ... 0.49 ... ... ... ... ...

Mo

Cu

Ni

Base iron without alloy additions 0.56 ... ... 0.43 0.52 ... ... ... ... ... 0.52 ... 0.43 ... 1.45 0.54 0.65 ... 0.47 ... ... ... ... 1.45 0.54 ... 0.66 ... ... ... ... ... 1.72 0.47 ... ... ... 1.80 ...

V

Hardness, HB

... ... ... ... ... ... 0.13 ... ... 0.12 ... ... ...

217 262 248 241 241 285 269 255 255 269 229 235 241 235

†

After annealing

Tensile strength psi

MPa

Hardness, HB

37,400 46,200 50,600 41,700 43,000 55,500 52,400 48,400 45,400 50,000 41,000 41,700 44,000 43,500

258 319 349 288 297 383 361 334 313 345 283 288 303 300

143 217 207 207 201 197 187 179 156 156 156 149 146 143

† Base iron analysis: 3.26% total C, 1.92% Si, 0.94% Mn, 0.03% S, 0.11% P. Source: G. A. Timmons and V. A. Crosby, Foundry, vol. 69, October 1941, pp. 64–66, 145–147; November 1941, pp. 64–66, 142–147. Reprinted by permission of ASM International, Materials Park, Ohio.

Tensile strength psi

MPa

27,700 44,500 43,100 40,000 38,200 42,600 37,400 40,000 34,800 33,000 31,200 29,900 31,600 29,900

191 307 297 276 263 294 258 276 240 228 215 206 218 206

12.30

CHAPTER TWELVE

section size of the casting. Stress-relieving does not influence tensile strength, hardness, or ductility. The temperature used for stress relieving lies well below the transformation range of pearlite to austenite. When maximum stress-relief (up to 75 to 85%) together with minimum decomposition of carbide in unalloyed irons is desired, stress-relief treatment at a temperature range of 540 to 565°C (1000 to 1050°F) for 1 hr is recommended. To attain more than 85% stress relief in unalloyed irons, a minimum temperature of 595°C (1100°F) can be used, but some sacrifices in hardness, strength, and wear resistance are inevitable.40,42 Low-alloy gray irons usually require a higher stress-relieving temperature of 595 to 650°C (1100 to 1200°F), which depends on the alloy content. Similarly for highalloy gray irons, a temperature of 620 to 675°C (1150 to 1250°F) may be needed to eliminate most of the internal stresses. Figure 12.16 shows the effect of stressrelieving time and temperature on the extent of stress relief for seven low-alloy gray irons, and the accompanying table indicates that the stress relieving at 620°C (1150°F) for 8 hr does not have an adverse effect on hardness.40 For austenitic (Niresist) cast irons, stress-relieving is done at 620 to 675°C (1150 to 1250°F) to remove residual stresses due to casting and/or machining. Stress-relieving should follow rough machining, especially for castings that require close dimensional tolerances, that have been extensively welded, or that are exposed to high stresses in service. In these alloys, stress-relieving at 675°C (1250°F) will eliminate about 95% of the stress. It is the usual practice to cool castings in air at a rate of 1 to 2 hr/in. (25 mm) of section thickness. It is of particular importance in stress relieving to ensure that the rate of heating and rate of cooling are slow, to avoid the imposition of additional thermal stresses. It is normal practice to load the furnace at a temperature not exceeding 95°C (200°F), and the rate of heating should be such that it would take 3 hr to attain 620°C (1150°F), be held there for 1 hr, and be furnace-cooled to 315°C (600°F) or lower in about 4 hr before being allowed to cool in air. For castings of intricate shapes, it is recommended to continue furnace cooling until a temperature of about 95°C (200°F) has been attained.40 12.7.1.3 Normalizing. As in steels, normalizing in gray iron castings is accomplished by heating about 50°C (100°F) above the critical temperature range or by heating typically to 885 to 920°C (1625 to 1700°F). The holding time at the normalizing temperature in the heat-treating furnace is about 1 hr/in. of maximum thickness, followed by cooling in still air to room temperature. Iron castings may be normalized in three different ways, and the choice among them solely depends on the existing condition and the final hardness desired:41 1. Castings from the sand molds are usually given a separate normalizing treatment to obtain the increased hardness of castings. 2. Air cooling, just after the completion of solidification of castings, is done to remove excess (or free) cementite. 3. After the solidification of castings, and its subsequent fall of temperature to a region above the critical temperature range, they are stripped off from the mold, freed of sand, and cooled in still air to achieve an improvement in hardness. Normalizing may be used to improve mechanical properties such as tensile strength and hardness or to restore as-cast properties that have been modified by

FIGURE 12.16 Effect of stress-relieving time and temperature on extent of stress relief obtained in low-alloy gray irons. Table lists compositions and negligible effect of maximum stress-relieving conditions on hardness.40

12.32

CHAPTER TWELVE

another heat-treating process, such as graphitizing or the preheating and postheating associated with repair welding. The tensile strength and hardness of normalized gray iron castings are a function of combined carbon content, pearlite spacing, and graphite morphology. Normalizing produces fine pearlite matrix, the fineness depending on the maximum (normalizing) temperature and alloy content. This structure exhibits good wear resistance with reasonable machinability and an excellent response to flame hardening. For unalloyed gray irons, unless fans are used, normalizing produces structures softer than the as-cast material irrespective of the temperature used. For alloy irons, however, harder and stronger normalized structure is produced with higher normalizing temperature. It is thus apparent that normalizing retains as-cast properties to gray irons, and if the carbon equivalent is significantly small, normalizing even results in an improvement of these properties. Additionally, the alloying elements Cr, Ni, and Mo increase the strengthening effect of normalizing.40 Normalizing of alloy irons is often followed by tempering at 500 to 625°C (950 to 1150°F) to reduce hardness and to relieve some of the residual stresses developed when parts have variable section thicknesses.

12.7.2 Ductile (or Nodular or Spheroidal Graphite) Cast Iron The principal types of annealing and normalizing treatments are discussed here. 12.7.2.1 Annealing. Castings containing higher levels of C and Si are readily annealed due to accelerated decomposition of pearlite and carbides, thereby reducing the time at annealing temperature. This increases ductility and produces good machinability. Alternately, the alloying elements such as Mn, P, and Ni and the carbide stabilizers such as Cr and Mo retard the annealing process. This, in turn, reduces machinability on annealing. Annealing is practiced to remove pearlite and eliminate any eutectic carbide which may form during solidification, especially in light sections.44 Satisfactory annealing can be performed in four different ways:45–47 Single-Stage (Subcritical) Annealing. The casting is heated to a subcritical temperature of 705 to 730°C (1300 to 1350°F), held for 1 hr/in. of section thickness, and then furnace-cooled at 55°C (100°F)/hr to 345°C (650°F), followed by air cooling to obtain grades 65-45-12 and 60-40-18. This treatment does not eliminate carbides, and therefore it is used when maximum impact strengths are not required. Figure 12.17 shows the effect of subcritical annealing at 705°C (1300°F) for various times on hardness of four ductile irons. For alloyed ductile irons, controlled cooling below 55°C (100°F)/hr through the critical temperature range down to 400°C (750°F) is recommended.46,47 Full (Two-Stage) Annealing with Carbides Present. The iron casting is austenitized first by heating to temperatures in the range of 900 to 925°C (1650 to 1700°F), held there for 2 to 4 hr, then furnace-cooled at 95°C (200°F)/hr to 680 to 705°C (1256 to 1300°F); it is usually held for 2 to 6 hr to allow the graphitization process to proceed to completion with the production of a ferrite matrix. The material is finally furnace-cooled at 55°C (100°F)/hr to 345°C (650°F) prior to removal for air cooling to obtain grades 65-45-12 and 60-40-18.45–47 Figure 12.18 shows the distribution of graphite nodules within a ferrite matrix produced by this two-stage annealing process.46

BASIC HEAT TREATMENT

12.33

FIGURE 12.17 Effect of time at subcritical annealing temperature on hardness.46,47 (Reprinted by permission of ASM International, Materials Park, Ohio.)

a

b

FIGURE 12.18 Microstructure of a 4-mm (5/32-in.) section of ductile iron: (a) As-cast. (b) After two-stage annealing treatment consisting of holding at 900°C (1650°F) for 4 hr, furnace-cooling to 690°C (1275°F), holding for 5 hr, then furnace-cooling to room temperature.46 Picral etchant. (Reprinted by permission of ASM International, Materials Park, Ohio.)

Modified Two-Stage Annealing. The modified two-stage annealing cycles have been developed to produce the same or even better properties in remarkably shorter times. In this method, castings of 0.5-in. section thickness are first austenitized at 870 to 900°C (1600 to 1650°F) for 20 min to dissolve any free carbides and are then cooled rapidly (in 15 min) to 675°C (1250°F) to transform austenite to pearlite. In the second subcritical stage, the temperature is raised to 760°C (1400°F) and held there for 10 min, during which 90% ferritization of pearlitic carbide is achieved. The total time required after heating to 900°C (1650°F) for this two-stage process is 45 min.41 Full Annealing (for Unalloyed Casting in Absence of Carbides). The lower silicon iron casting is heated and held at 870 to 900°C (1600 to 1650°F) for 1 hr/in. of section thickness and then furnace-cooled at 55°C (100°F)/hr to 345°C (650°F), followed by air cooling to obtain grade 60-40-18 with maximum low-temperature impact strength.46,47

12.34

CHAPTER TWELVE

12.7.2.2 Stress Relieving. This treatment removes residual stresses in (1) unalloyed castings by heating at 510 to 565°C (950 to 1050°F); (2) alloy castings by heating at 565 to 595°C (1050 to 1100°F); (3) high-alloy castings by heating at 595 to 650°C (1100 to 1200°F); and (4) austenitic castings by heating at 620 to 675°C (1150 to 1250°F). The required time at temperature is normally 1 hr plus 1 hr/in. section thickness. Castings should be furnace-cooled to 290°C (550°F), followed by air cooling. In majority of cases, austenitic ductile iron can be uniformly air-cooled from the stress-relieving temperature.46,47 Cooling should be uniform to avoid reimposition of thermal stresses. 12.7.2.3 Normalizing. This treatment can be carried out by holding to a temperature between 870 and 940°C (1600 and 1725°F), typically 100°C (180°F) above the critical temperature range, for 1 hr/in. section thickness in the heat-treating furnace, followed by air cooling. This treatment can be used to break down carbides, increase strength and hardness, and produce more uniform properties, but with a large decrease in ductility. It produces a fine pearlitic structure together with spheroidal graphite, provided that Si is not too high and that it contains a moderate Mn content. Normalizing should be followed by tempering (or reheating) at 425 to 650°C (800 to 1200°F) and holding at that temperature for 1 hr/in. section thickness, which renders uniform hardness and mechanical properties (including high toughness and impact resistance) and relieves residual stresses arising from the uneven air cooling of different section thicknesses. The resulting microstructure depends on the composition of the castings and the cooling rate; the former, in turn, depends on the alloy content, while the latter depends on the mass and temperature of castings. Step normalizing, using a second lower-temperature stage prior to air cooling, can be employed to give the improved matrix control needed to produce the pearlitic/ferritic grades of ductile iron.48 The effect of tempering on tensile properties and hardness depends on both the composition of the iron and the hardness level attained in normalizing. Figure 12.19 illustrates the effect of tempering temperature on the hardness of normalized ductile iron.46 12.7.3 Compacted Graphite (CG) Iron Like gray and ductile irons, annealing and normalizing of CG iron can produce a variety of matrix structures such as ferrite, ferrite-pearlite mixture, and pearlite. However, CG irons are not heat-treated, due to their applications in the as-cast condition. 12.7.4 Malleable Iron 12.7.4.1 Annealing. Annealing, as well as other heat treatment, is done in a controlled atmosphere to avoid decarburization, scaling, and loss of silicon during the extended periods at high temperature (955°C). This atmosphere is generated by packing the castings in a mixture of sand (or gravel) and carbonaceous material when heated in an open-fired furnace. Alternatively, castings are carefully sealed from air or the usual furnace atmosphere, and annealing is continued in an atmosphere created by the castings themselves. However, in recent years, the most effective atmosphere to prevent decarburization has been (1) dry nitrogen mixed with 1.5% hydrogen and 1.5% CO or (2) vaporized liquid nitrogen with the addition of

BASIC HEAT TREATMENT

800

12.35

Tempering temperature, °F 1000 1200

450 Over 400 HB normalized

Hardness, HB

400

350

300 Under 400 HB normalized

250

200 400

500 600 Tempering temperature, °C

700

FIGURE 12.19 Hardness of normalized ductile iron tempered at various temperatures.46 (Reprinted by permission of ASM International, Materials Park, Ohio.)

some methane.49 The dew points of these mixtures should lie between -40 and -7°C (-40 and -20°F). Ferritic, pearlitic, and martensitic malleable irons are produced through variations of controlled annealing of white cast iron of suitable composition. Thus, annealing is an integral part of the manufacturing process for these irons. During the annealing cycle, carbon that is present in the combined form, either as a microconstituent in pearlite or as massive carbides, is converted into free graphite in the form of irregular clumplike nodules called temper carbon. It occurs early during the holding or soaking period. Ferritic Malleable Cast Iron. The annealing heat treatment for ferritic malleable cast iron consists of the following two stages. first-stage graphitization. The first stage converts primary carbides to temper-carbon nodules. In this stage, white iron castings are heated at a rate such that the temperature of 900 to 970°C (1650 to 1780°F) is reached in 4 hr. The soaking time varies from ⬃2 to 36 hr depending on the composition of the iron castings and the temperature of first-stage annealing. For example, longer soaking times are necessary if the silicon content of iron castings or the temperature employed is low. During holding at the first-stage temperature, iron carbide dissolves in the austenite matrix, and the excess carbon diffuses to nucleate temper-carbon nodules at preferred sites such as boron nitride or at the (primary) carbide/(saturated)

12.36

CHAPTER TWELVE

austenite interface. Thus the growth around these nuclei occurs by a reaction involving carbide decomposition and the diffusion rate of carbon. The rates of nucleation and graphitization are increased by high silicon and carbon contents. However, to solidify it as a white iron, these elements must be restricted to a certain maximum level. The first-stage annealing temperature also influences the rate of annealing and the number of temper-carbon nodules produced. second-stage graphitization. Castings are cooled as fast as practical to 740 to 760°C (1360 to 1400°F). This rapid cooling step requires 1 to 6 hr, depending on the equipment being used. Castings are then cooled slowly (through the transformation range of iron) at a rate of about 3 to 11°C (5 to 20°F)/hr to 649°C (1200°F); during this period, the remaining carbon dissolved in the austenite phase is converted to graphite and transferred to the existing temper carbon formed at the higher temperatures, and the austenite transforms to low-carbon ferrite. The final microstructure consists of uniformly dispersed temper-carbon nodules within a fully ferritic matrix50,51 (Fig. 12.20a). Pearlitic Malleable Iron. In the production of pearlitic malleable iron, the first-stage heat treatment includes annealing for about 13 hr at 970°C (1780°F). The second stage involves air cooling (Fig. 12.20b). Faster air cooling by an air blast avoids the formation of films of ferrite around the temper-carbon particles (bull’s-eye structure) and produces less ferrite and a finer pearlitic structure50,51 (Fig. 12.20c). The castings are then either tempered or reheated in a furnace to 870°C (1600°F), oil-quenched, and tempered. Martensitic Malleable Iron. Uniformly high quality irons are produced by agitated oil-quenching of the castings directly from the first-stage annealing after stabilizing the temperature at 845 to 870°C (1550 to 1600°F) for 15 to 30 min. The martensitic matrix produced has a hardness of 415 to 601 HB. Finally, the castings are tempered at 590 to 725°C (1100 to 1340°F) to develop the desired microstructure (of tempered martensite plus temper-carbon nodules) and mechanical properties. If the hardness decreases by prolonged tempering, the resulting microstructure may not possess a good response to selective hardening.50,51 Note that during the heat treatment process (such as annealing), expansion or growth of castings occurs due to the nucleation and growth of temper-carbon nodules. Any restriction of the free growth such as the weight of castings placed on top of each other leads to distortion. This may, however, be corrected by diepressing. The design of the molding pattern should reflect this dimensional change.49

12.8 ENGINEERING PROPERTIES AND APPLICATIONS OF CAST IRONS 12.8.1 Gray Iron Mechanical Properties. Different graphite flake types, the amount of prior austenite dendrites, and the matrix structure present in the iron structure affect the mechanical properties. Tensile strength is an important criterion in selecting a gray iron for parts that are subjected to static loads in direct tension or bending. Such parts include pressure vessels, housings, autoclaves, valves, fittings, and levers. Based on the uncertainty of loading, safety factors of 2 to 8 are used to estimate allowable design stresses. The tensile strength of the plain and lower-alloy gray irons varies from 69 to 414 MPa. The carbon equivalent and rate of solidification or section size play an

BASIC HEAT TREATMENT

a

12.37

b

c FIGURE 12.20 Microstructure. (a) ASTM A602, grade 3210, ferritic malleable iron with tempercarbon (type III graphite) nodules (black constituents) in a matrix of granular ferrite produced after two-stage annealing by holding at 954°C (1750°F) for 4 hr, cooling to 704°C (1300°F) in 6 hr, followed by air cooling. Small gray particles are MnS (2% nital etch). (b) Pearlitic malleable iron with the formation of free ferrite films (white constituent) around temper carbon (bull’s-eye structure) obtained after first-stage annealing by austenitizing at 971°C (1780°F) for 13.5 hr and then aircooling slowly (2% nital etch). (c) Malleable iron containing 2.5% C and 1.5% Si with temper-carbon (type III graphite) nodules in a matrix of fine pearlite (variegated gray), produced after first-stage annealing by holding at 954°C (1750°F) for 2 hr and air-blast-cooling (1% nital etch). (Reprinted by permission from Metals Handbook, vol. 7, 8th ed., American Society for Metals, Metals Park, Ohio, 1972.)

important role on this strength level. Higher strength is achieved by a combination of small proportions of graphite, decreasing carbon equivalent, finer flakes, more austenitic dendrites, and stronger matrices (obtained with decreasing section sizes or faster solidification and cooling times).49 Gray irons do not show straight stress-strain curves (Fig. 12.21) due to the development of localized plastic deformation, during loading, which results from the stress concentration (notch) effect at the edges of graphite flakes.52 The amounts of graphite, graphite flake types, and matrix structure also influence the modulus of elasticity. For example, increasing graphite content, longer graphite flake length, increasing carbon equivalent, increasing section size of the gray iron castings, and softer matrix structures (produced by annealing of unalloyed

12.38

CHAPTER TWELVE

FIGURE 12.21 Typical stress-strain curves for three classes of gray iron in tension.52 (Reprinted by permission of ASM International, Materials Park, Ohio.)

FIGURE 12.22 Two methods for determining the modulus of elasticity for gray iron. The secant modulus to 25% of the tensile strength is the more commonly used.41 (Courtesy of Iron Castings Society, Inc., Des Plaines, Illinois.)

iron) reduce the modulus.41 Figure 12.22 shows the two methods for determining the modulus of elasticity for gray iron castings.42 The more commonly used method of measuring the modulus of elasticity is the secant modulus, which corresponds to the slope of a straight line from the origin (i.e., at no load) to a point on the stressstrain curve at one-fourth of the ultimate tensile strength; and the deviation of the

BASIC HEAT TREATMENT

12.39

stress-strain curve from linearity is usually 24-in.-) diameter carburized automotive parts, automotive, truck and naval crankshaft production, alloy-steel forgings, tube stock for oil drilling applications, large alloy castings for the oil industry;47 and induction hardening and spray quenching.49 PAG solutions, now referred to as “Type-1” polymer quenchants, have been used in the aluminum heat-treating industry for about 35 years as an excellent alternative to hot-water quenching for residual stress and distortion reduction and crack prevention of both forged and cast aluminum parts.51,52 Polyvinyl Pyrrolidone (PVP). PVP solution (aqueous) is characterized by its complexing and colloidal properties, by its physiological inertness, and by not exhibiting a thermal separation temperature. This is available in four molecular weight grades and may be used up to the boiling point of water. A 10% aqueous

1000

0

10

Cooling rate, °C/s 30 40

20

50

60 1800

900

1600

800

Cooling rate curves

1400 700

25%

15%

5%

1000 500 800

400 300

600

200

400 25%

15%

5%

100

Temperature, °F

Temperature, °C

1200 600

200 Cooling curves

0 0

15

30

45

60

75

90

105

120

Time, s (a)

0

0 1000

15

10

Cooling rate, °F/s 45 60 75

30

20

90

105

Cooling rate, °C/s 30 40 50

60

120

1800 Cooling rate curves

900

1600 27°C

800 38°C

49°C

1400 1200

600

1000

500 400

800

300

600

200

Temperature, °F

Temperature, °C

700

400 Cooling curves

100

49

0 0

15

30

45

60 75 Time, s

90

38 27

105

200 120

(b)

FIGURE 13.18 Cooling curves and cooling rate curves for a 25-mm- (1-in.-) diameter stainless steel probe: (a) quenched in 5, 15, and 25% PAG at 45°C (110°F) that is flowing at 0.25 m/s (50 ft/min); (b) quenched in 10% PAG at 27, 38, and 49°C (81, 100, and 120°F) that is flowing at 0.25 m/s (50 ft/min); and (c) quenched in 20% PAG at 45°C (110°F) that is flowing at 0, 0.25, and 0.50 m/s (0, 50, and 100 ft/min).6

13.32

HARDENING AND HARDENABILITY

0

0

10

30

20

90

105

Cooling rate, °C/s 30 40 50

60

120

1800

900 1600 800 0.50 m/s

700

1400

0.25 m/s

600

1200

0 m/s

500

1000

400

800 600

300 0.25 m/s

200

400

0 m/s

100

200 0.50 m/s

0 0

15

30

45

60 75 Time, s (c)

90

105

0 120

Temperature, °F

Temperature, °C

Temperature, °F

1000

15

(b) Cooling rate, °F/s 45 60 75

13.33

FIGURE 13.18 (Continued) Cooling curves and cooling rate curves for a 25-mm- (1-in.-) diameter stainless steel probe: (a) quenched in 5, 15, and 25% PAG at 45°C (110°F) that is flowing at 0.25 m/s (50 ft/min); (b) quenched in 10% PAG at 27, 38, and 49°C (81, 100, and 120°F) that is flowing at 0.25 m/s (50 ft/min); and (c) quenched in 20% PAG at 45°C (110°F) that is flowing at 0, 0.25, and 0.50 m/s (0, 50, and 100 ft/min).6

PVP solution should be used with a preferred rust inhibitor and a bactericidal preservative. Like other polymer solutions, the quenching performance of PVP polymer solutions is relatively sensitive to small changes in concentration, bath temperature, and agitation. The quenching rates seem to be rapid with PVP quenchants in the upper temperature range (during the stable film and nucleate and boiling stages), but slower during the convection stage. Contour plot relationships showing the effect of quenchant concentration, agitation, and bath temperature on the maximum cooling rate (CRmax), cooling rate at 365°C (650°F), and Grossman H-factor under laboratory quenching conditions are illustrated in Fig. 13.19.50 Since PVP is soluble up to the boiling point of water, a wider working range of temperatures for quenching can be used. Optical refractrometer readings can provide initial control of concentration, but backup with viscosity measurements is strongly suggested.6 Poly(sodium) Acrylate (PSA). Characteristic features of polysodium acrylate solutions are as follows: 1. They are ionic quenchants which separate them from the class of other nonionic polymer quenchants. This peculiar feature imparts strong polarity, which, in turn, produces water solubility and probably causes a different operating mechanism of heat extraction, as given below. 2. They neither split on heating nor form plastic films on the surface of the hot part. 3. By changing the molecular weights of the polymer, a complete family of quenchants can be developed, covering a wide spectrum of applications from the fast quenching of water to the slow quenching of oils. 4. Figure 13.20 shows the effect of concentration and temperature on cooling rates of polyacrylate quenchant. The unique property of the polyacrylate

13.34

CHAPTER THIRTEEN

FIGURE 13.19 Comparison of the predicted quenching characteristics of PAG and PVP polymers under varying conditions: (a) maximum cooling rate CRmax, (b) cooling rate at 345°C (650°F), and (c) Grossman H factor.50

HARDENING AND HARDENABILITY

13.35

quenchants allows its cooling curve to be nearly straight (Fig. 13.20) due to prolonged vapor-blanket stage and reduced nucleate boiling stage period.6 This property facilitates their use in hardening crack-prone parts made of high-hardenability steels. 5. With increasing PSA concentration and bath temperature, the cooling rate can be decreased to such a level that many ferrous alloys do not transform to martensite at all, but rather form bainite or fine pearlite, thereby achieving lower hardness values.

1800

45°C(110°C)

1600

1600

1000 800

400

600 400

200

1400 1200

600

2.5% 5.0% 10.0% 20.0%

400

0

20

40

60

200

400 200

0

80

0

20

40

2.5% 5.0% 10.0% 20.0%

600

1200 1000

600 200

800

1400

800

400

400

Cooling curves of polyacryiate solutions concentration, wt%

1600

2.5% 5.0% 10.0% 20.0%

1200

600

400

20

40 Time, s

60

80

1400

1000 800 600

200

400

200 0 0

80

1800

75°C(170°F)

1600

Temperature, °C

Cooling curves of polyacryiate solutions concentration, wt%

1000

Temperature, °F

Temperature, °C

1800

60°C(140°F)

800

60

Time, s

Time, s

1000

800 600

200 0

1000

Temperature, °F

1400 1200

Cooling curves of polyacryiate solutions concentration, wt%

800

200 0 0

20

40

60

80

Time, s

FIGURE 13.20 Cooling rate of an agitated polyacrylate quenchant as a function of concentration and temperature. Test specimen: 0.4-in. diameter ¥ 2.4-in. (10-mm diameter ¥ 60-mm) austenitic steel.6 (Reprinted by permission of ASM International, Materials Park, Ohio.)

Temperature, °F

600

1000

Temperature, °C

800 Temperature, °C

1800

27°C(80°C) Cooling curves of polyacryiate solutions concentration, wt% 2.5% 5.0% 10.0% 20.0%

Temperature, °F

1000

13.36

CHAPTER THIRTEEN

6. In general, higher agitation is recommended for hardening treatments, whereas minimal agitation is recommended for nonmartensitic quenching to form bainite or pearlite. Figure 13.21 illustrates a comparison of cooling curves of a PSA quenchant with those of water, conventional oil, and a few typical polymer quenchants. This provides benefits in applications requiring slow quenching.6 applications. PSA solution is not widely used in the heat treatment industry. However, it is employed to harden high-hardenability steels that are particularly susceptible to quench cracking. Applications include 1. Use of polyacrylate solution as the first quenchant in the commonly practiced double-hardening treatment in oil for deep-carburized parts such as bearing races, balls, and rollers. 2. Direct quenching of high-carbon steels (in polyacrylate solutions) to produce similar mechanical properties as obtained by quench and temper or austempering (e.g., railroad or automotive forgings or sway bars, and rod and wire patenting made of SAE 1070 or 1090) as well as spring plates, torsion rolls, and gears of certain types.17a 3. The relatively long vapor-blanket stages encountered with PSA quenchants which have promoted its application in the quenching of railway rails, where pearlite and fine bainitic transformation products are required. 4. Quenching of hot-worked parts (in polyacrylate solutions) to obtain the same microstructure as by air cooling, without inducing excessive scaling and decarburization. 5. Replacement of the commonly used salt or lead bath process by polyacrylate solutions for patenting of wire at 510 to 565°C (950 to 1050°F).

FIGURE 13.21 Cooling rates of various quenching media (including different polymer quenchants at 20% concentration) at 25 and 60°C (80 and 140°F). Test specimen: 10-mm diameter ¥ 60 mm (0.4in. diameter ¥ 2.4 in.). Scaleproof austenitic steel, medium agitation.6 (Reprinted by permission of ASM International, Materials Park, Ohio.)

HARDENING AND HARDENABILITY

13.37

6. Direct quenching of hot-worked parts (in polyacrylate solutions) to obtain good machinability without undergoing the conventional hardening and tempering process, especially for low-hardenability, low-alloy carburizing steel parts. 7. Quenching of large continuously cast steel slabs made of low- and high-carbon steel and alloy steels (in polyacrylate solutions) to allow inspection shortly after casting. This causes faster cooling without any tendency for quench crack.6,17 Aqueous PSA quenchants have some drawbacks that preclude their broader use in the heat treatment industry. However, with the probable exception of sensitivity to mechanodegradation, possibly the most important drawback is the propensity to form “polyelectrolyte complexes” with polyvalent cations that may form precipitates with polyvalent metals. Sources of polyvalent metal ions are Ca2+ and Mg2+ from hard water or solubilized Fe3+ ions from iron oxide quenching scale or quench tank corrosion products.8 Polyacrylate solutions can also be employed for quenching solution-treated aluminum alloys to minimize distortion and warpage. Polyethyl Oxazoline (PEOX). Recently, PEOX products developed maintain the main properties of oil together with additional benefit of “nontacky” behavior of the residue for applications in induction heat treatment systems where the potential formation of residual tacky residues after quenching may interfere with the robotics operations.12,51 Although aqueous solutions of PEOX, like PAG-based quenchants, display inverse solubility temperatures, PEOX is not applied to purify quench bath. This may be due to a possibility of susceptibility to hydrolytic degradation of the amide linkages present in the PEOX polymer.8 The PEOX quenchant is also subjected to a film rupture process, but the rupture seems to be associated with a partial dissolution of PEOX. The haziness of the film is attributed to polymer instability, because the film temperature lies above the cloud point.8 Selection of Polymer Quenchant. Selection of polymer quenchant compared to other quenching media can be made based on a sensitivity band of steel hardenability and section complexity, as shown in Fig. 13.22a, which illustrates the more extended applications of PSA, PVP, and PEOX to steels of higher hardenability and thinner sections than PAGs, while still maintaining the environmental advantages of polymer quenchants.25a Polymer Quenchant Stability. Polymer quenchant stability can be influenced by several factors such as mechanodegradation, thermal/oxidation stability, dragout, and hydrolytic stability. The extent of these effects is a function of the polymer structure and bath and quenching conditions.8 mechanodegradation. The polymer chains coil around each other in solution to provide a relatively long-range order, which increases the viscosity (thickness effect) with concentration (Fig. 13.22b).8 The extent of polymer chain enlargement increases with an increase in molecular weight (chain length). Thus, for a given concentration, the viscosity of a polymer solution increases with the increase in molecular weight. When mechanical energy (shear) is applied to a polymer solution by pump or impeller agitation, a shear-induced thinning of the solution takes place. With adequate energy application, covalent C—C bonds in entangled polymer chains easily break apart. With the occurrence of C—C bond scission, the average molecular weight of the polymer and the corresponding viscosity of the polymer solution are reduced. This is called mechanodegradation or shear degradation, which can be expected to produce a proportional increase in cooling rate.53

13.38

CHAPTER THIRTEEN

FIGURE 13.22 (a) Sensitivity band for quenchant selection which is influenced by concentration, agitation, and temperature of the quenchant bath.25a (b) Variations of aqueous solution viscosity with polymer molecular weight.8 (Reprinted by permission of ASM International, Materials Park, Ohio.)

Variations in solution viscosity reflect both the polymer stability and potential quench severity. Polymer can display different shear stabilities when subjected to the same shear field, and a variety of results can be available with different polymers. The heat treater should require data on the relative shear stability of a polymer quenchant prior to using it.8 polymer dragout. Polymer dragout occurs when the polymer after phase separation does not redissolve prior to the withdrawal of the part from the quench tank. In reality, polymer dragout always occurs because of the solution-wetting, whether polymer is completely or partially redissolved. However, dragout is expected to increase with the concentration (Fig. 13.23a),14 solution viscosity (Fig. 13.23b),14 and molecular weight of the polymer. thermal/oxidative stability. Figure 13.24a shows the schematic illustration of the impact of polymer chain oxidation where polymer chain scission occurs in the middle. The impact of this chain scission causes higher viscosity reduction with higher-molecular-weight (longer) polymer (II) with respect to shorter polymer (I) (Fig. 13.24b). The ultimate effect of extensive oxidative/thermal degradation results in quench severity closer to water itself.43

HARDENING AND HARDENABILITY

13.39

FIGURE 13.22 (Continued) (a) Sensitivity band for quenchant selection which is influenced by concentration, agitation, and temperature of the quenchant bath.25a (b) Variations of aqueous solution viscosity with polymer molecular weight.8 (Reprinted by permission of ASM International, Materials Park, Ohio.)

chemical stability. It is essential that the polymer quenchant selected be hydrolytically stable under quench bath conditions (e.g., elevated temperature and alkaline pH). This factor has limited the widespread use of polyacrylamide, PVA, and PEOX quenchants, because they are prone to hydrolytic degradation under some situations. Possible hydrolytic degradation reactions are shown in Fig. 13.25 and are based on the well-established organic chemistry reactions. A major purpose of quenchant development is to minimize degradation reactions and to increase long-term quenchant stability. Polymer Quenchant Bath Maintenance. Various authors have used the cooling curve analysis to show that polymer quenchants are susceptible to oxidative/thermal degradation processes.14,48,54–56 To ensure maximum performance, all precautionary measures should be taken so that the quenchants remain free of contaminants such as scale. It is, therefore, essential that a filter, preferably 5-mm, be used in the quenchant recirculation system.57 The most common methods used for polymer quenchant analysis include concentration, viscosity, refractive index, conductance,

13.40

CHAPTER THIRTEEN

FIGURE 13.23 (a) Dragout specimen shape and results for polymer quenchants of various concentrations in both the static and agitated conditions. (b) Effect of polymer solution viscosity on the level of dragout in static tests.14 (Courtesy of N. A. Hilder.)

separation temperature, and gel permeation chromatography (GPC) (Table 13.5). Only the most common testing methods are outlined here. polymer concentration. The cooling rate of an aqueous polymer solution critically depends on the polymer concentration (Fig. 13.26a) and the condition (molecular weight and oxidation) (Fig. 13.26b). Polymer loss also occurs much more by dragout and less by polymer degradation.55,56,58 It is, therefore, important that the quenchant concentration be maintained within approximately ±1.0%. refractive index. Refractive index nD, being a linear function of polymer or quenchant concentration, is the most common method to measure polymer concentration (Fig. 13.27a). The subscript D implies that the refractive index is measured at the wavelength of the D line of sodium (5893 Å). This is determined using ASTM Standard D1747 and D1218 methods and (properly calibrated) handheld refractrometer (Fig. 13.27b). A droplet of the solution, when put on the prism of this refractometer and directed against light, shows a related value to reading with the naked eye. The check should be made every week. This method is primarily restricted to quenchants derived from PAG and not applicable to dilute solutions of high-molecular-weight polymer such as PVP and PSA quenchants because of its unacceptable sensitivity to refractometry.8 Since refractive index measurements are also influenced by polymer degradation and the presence of contaminants in the quenchant system, its sensitivity decreases with the decrease in concentration.55,59 Hence, it is essential to check the periodic validation of the quenchant concentration by complementary viscosity measurement. If carried out together, these measurements provide a good understanding of the condition of a polymer quenchant.

HARDENING AND HARDENABILITY

13.41

a

Polymer l undegraded X/2

X/2

Polymer l degraded at a

b Polymer ll undegraded Y/2

Y/2

Polymer ll degraded at b (a)

Polymer II

Viscosity

D

C B A

MI/2 MII/2

MI

MII

Molecular Weight (b)

FIGURE 13.24 (a) Schematic illustration of the polymer chain scission process for a polymer chain with a low (polymer I) and a high (polymer II) molecular weight. (b) Relative effects of the polymer chain degradation process shown in Fig. 13.23b.43

Possible contaminants include the following: (1) Solid contaminants such as scale or carbon/soot have little effect on the quench rate but can adversely affect the concentration control by refractive index. (2) Tramp fluids, e.g., cutting fluids and hydraulic oils, can act as nutrients for biological growth and can increase the

CHAPTER THIRTEEN

13.42

FIGURE 13.25 and amides.

Hydrolysis of esters

TABLE 13.5 Methods for Polymer Quenchant Analysis Test Concentration viscosity Refractive index Salt content Corrosion protection Biological activity

Procedure ASTM D445 ASTM D1749, D1218 Conductance pH additive analysis Dipstick test

vapor-phase stage of the quenching process. (3) Biological contamination such as fungi and bacteria results in foul smell and leads to corrosion-inhibitor depletion. Accumulation of fungus deposits can block nozzles and filters on spray systems. (4) Dissolved materials such as heat treatment salts, water hardness salts, and gases such as NH3 for carbonitriding operations can be contaminants.54 viscosity. Since viscosity h (cSt) of a polymer solution is dependent on concentration, molecular weight (or size), and temperature (Fig. 13.28), it is essential to compare the viscosities of fresh and used polymer quenchants at the same temperature by capillary viscometry. Low levels of salt contamination usually produce a minimal effect on viscosities of most quenchant solutions. As with quench oil, the viscosity of a polymer solution can be measured by using the ASTM Standard D445 procedure. Since viscosity is sensitive to low concentration of polymer in solution, this method can be readily used to measure low concentrations of relatively highmolecular-weight polymers such as PSA. salt content. The salt content of a polymer quenchant, which is introduced either from hard water or by contamination from salt baths, usually increases during use. Salt contamination of polymer quenchants can greatly increase the quenchant cooling rates and decrease the cloud point. Hence, the makeup water for the quench bath should be distilled or deionized. Perhaps the simplest and most inexpensive method to continuously check salt contamination is by variation in conductance, which has become an important quality control tool for some heat-treating products.59 Such increases in conductance reflect the occurrence of salt contamination (Fig. 13.28b), and

Cooling rate, °F/s 2000

0

30

60

90

120 150 180 210 240 270 300 330 360

1800

Probe temperature, °F

1600 1400 1200

5% 10% 15%

1000 800 600 400 200 0 0

5

10

2000

15

20

25 30 35 Cooling time, s (a)

40

45

50

55

60

Test probe, 10 mm dia. ¥ 60 mm stainless steel AISI 304 Medium Agitation (1 ft/s) Temperature 120°F Polymer quenchant PAG-11 Used 13.0% (wt) New

1800 1600

Used New

18.4% (wt)

Temperature, °F

1400

1200

1000

800

600 400 200 0 0

10

20 Time, s (b)

30

40

FIGURE 13.26 (a) Effect of polymer concentration on cooling rates. (b) Comparison of fresh and used PAG quenchant.8

13.43

13.44

CHAPTER THIRTEEN

FIGURE 13.27 (a) Refractive index versus quenchant concentration.8 (b) Use of a handheld refractrometer to measure polymer quenchant concentration. (Courtesy of Leica Microsystems Inc.)

decreases in conductance suggest the depletion of the corrosion inhibitor, which is mostly ionic. Membrane Separation. The principles of microporous membrane separation, such as reverse osmosis (RO) and ultrafiltration (UF), can be employed to change the concentration of a polymer quenchant, to recycle it (e.g., from a rinse tank), or to separate organic and additives (and salt contaminant) from an aqueous quenchant solution for subsequent disposal. The main differences between the two techniques are system pressure and molecular size that can be separated. Reverse osmosis is preferred because it offers the most complete separation.60 The beneficial effect of membrane separation is that it is a low-energy method which can be used on a continuous basis and can help to solve waste disposal problems. Similar systems can be used for quench oil separation and regeneration. Thermal (Cloud Point) Separation and pH Measurement. To obtain a homogeneous aqueous solution, it is necessary that the polymer undergoes a hydrogen bond

HARDENING AND HARDENABILITY

13.45

Viscosity, Saybolt Universal seconds

250 200 75°F 85°F

150 65°F

100°F 100 90 80 70 60 50 40 30

0

5

10

15 20 25 30 35 40 Quenchant concentration, %

45

(a)

(m mho/cm) ¥ 1000

17.0

13.0

9.0

5.0 0.0

0.4

0.8

1.2

NaCl, % (b)

FIGURE 13.28 (a) Viscosity/concentration relationship for PAG polymer quenchant.8 (b) Conductance versus salt (NaCl) concentration in 20% polymer quenchant solution.8

interaction with water.8 Some polymers experience a phase separation with an increase in solution temperature at the hot metal interface during quenching. As a result, a two-phase system forms, one phase being polymer-rich and the other waterrich. The temperature corresponding to the occurrence of this phase separation is

13.46

CHAPTER THIRTEEN

called the separation temperature or cloud point. It is this phenomenon that is employed to control the severity of quenching. Two currently used polymers, polyethyl oxazoline (PEOX) and polyalkylene glycol (PAG), display a separation temperature which depends on the ratio of monomers in these copolymers. For example, the separation temperature is reduced with an increase in the number of propylene oxide monomer units in the PAG copolymer.8 It has also been shown that the cloud point of PAG copolymers is a function of the salt concentration, particular polymer composition, and pH of the aqueous solution. The cloud point of a PAG polymer solution characteristically increases with the increase of polymer oxidation. Variations in the cloud point affect both the cooling rate and quench severity as well as the viscosity versus temperature properties of the solution. Hence, it is essential to avoid substantial pH variations in order to minimize variations in quench severity. The cloud point behavior of quenchant solutions is used advantageously to purify the salt-contaminated (PAG and PEOX) quenchant bath and should be part of any quality control program. The pH measurement should be made at least once a week (to prevent corrosion of quenched parts and installation) and, if required, corrected by adding inhibitors. Corrosion Protection. Polymer quenchants must contain (either inorganic or organic) corrosion inhibitor to ensure corrosion protection of the system; the most common inorganic corrosion inhibitor being NaNO2 (or KNO2). NaNO2 may be easily determined by readily available colorometric reagents. The analysis of other commonly used corrosion inhibitors, such as amine/carboxyl acid salt or the sodium salts of mixed organofunctional aromatic acids, is relatively more complex and should be provided as a service by the polymer quenchant supplier.8 Biological Degradation. In closed systems such as automatic inductionhardening equipment, the bath is frequently infested with bacteria. Hence, a new bath must be prepared as a remedial step, after a few months of its operation. In immersion-hardening systems, the service life of the baths is longer, up to several years. One way to counteract the bacterial infection of the bath is to agitate it periodically or blow through with the compressed air.17a Problems arising from microbial growth include foul odors, slime, staining, and depletion of nitrite corrosion inhibitors. In some cases, microorganisms (to be detected by simple dipstick test) develop that can degrade the polymer. When this occurs, the performance of the polymer quenchant is severely affected.8 Problems associated with biological degradation can generally be controlled by the addition of the proper biocide. Users faced with such problems should contact a biocide supplier or the polymer supplier. Oil Contamination. The performance of aqueous polymer quenchants is influenced by the presence of oil contamination such as hydraulic, metalworking, and quench oils. The effect of oil contamination on cooling curve performance is shown in Fig. 13.29. As oil and aqueous polymer solutions do not provide friendly mixtures, they will form a discontinuous film around the hot metal part, which will produce substantial thermal gradients and may, in turn, lead to increased thermal and transformational stresses, probably producing cracking or increasing distortion.8 Changes in Polymer Quenchant with Use. Some degradation of polymer quenchant is inevitable.61 Table 13.6 compares the physical properties of the “used” quenchant and a “fresh” quenchant comprising the same total polymer concentration. The cooling curves for these quenchant solutions are illustrated in Fig. 13.30. A large difference in the viscosities of used (1.93 cSt) and fresh (5.65 cSt) quenchants reveals

HARDENING AND HARDENABILITY

13.47

FIGURE 13.29 Effect of quench oil contamination on the cooling curve behavior of an aqueous polymer quenchant.8

TABLE 13.6 Characterization of a Severely Degraded and Contaminated Polymer Quenchant8

Physical properties Concentration, % Viscosity, cSt at 40°C (100°F) Conductance, mmho/mm Cloud point, °C (°F) GPC area shift, % Cooling curve results (bath temperature of 40°C, or 100°F) Time to cool from 732 to 260°C (1350 to 500°F), s Maximum cooling rate, °C/s (°F/s) Temperature at maximum cooling rate, °C (°F) Cooling rate, °C/s (°F/s), at: 704°C (1300°F) 343°C (650°F)

Fresh quenchant

Contaminated quenchant

Water

20.0 5.65 5000 88.0 (190.0) ...

20.0 1.93 >20,000 >100 (212) -40

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

15.4

10.8

10.4

39.2 (70.6) 534 (993)

69.3 (124.7) 650 (1200)

61.8 (111.2) 610 (1130)

20.6 (37.1) 29.6 (53.3)

65.0 (117.0) 29.6 (53.3)

56.8 (102.2) 33.0 (59.4)

that extensive polymer degradation has taken place. This observation is supported by the GPC peak area shift of 40%. The amount of polymer degradation shown in Fig. 13.30 is clearly in excess of a permissible level. Loss of the characteristic cloud point (>100°C or 212°F, versus 88°C or 190°F) for the used polymer quenchant further substantiates its large polymer degradation.

13.48

FIGURE 13.30 quenchant.8

CHAPTER THIRTEEN

Cooling curves for an excessively degraded aqueous PAG quenchant and a fresh

Generally, used polymer should not display greater than a 1 or 2°C (2 to 4°F) cloud point elevation during normal use. Table 13.6 also indicates that the used quenchant is contaminated with salt, as evidenced by higher conductance than that of the fresh quenchant. Although higher conductance can result from excess addition of corrosion inhibitor (in this case, NaNO2), perhaps the high conductance suggests hard water contamination. It is, therefore, recommended to use deionized or distilled water to obtain maximum lifetimes for a polymer quenchant. Cooling curve analysis shows the effects of both polymer degradation and salt contamination (Table 13.6). Here, the quench severity of the used quenchant was very close to water compared to that of the fresh quenchant. The degradation level of a polymer quenchant rarely approaches this point. In reality, if the physical properties and the quench severity of the bath had been regularly checked, proper corrective action presumably could have been taken.8 Gaseous Contamination. Gaseous contaminants, such as NH3, arising from nitriding or nitrocarburizing may produce a significant extension of vapor-blanket cooling behavior, shown in Fig. 13.31.62 Similarly, other gases such as CO or CO2 may extend A-stage cooling times, which results in inadequate hardening. Although a technical solution is possible, this explains why aqueous polymer quenchants are not widely used in nitriding and nitrocarburizing processes. 13.2.3.5 Molten Salt Baths. Molten salt baths of various compositions are available to cover the temperature range from 150 to 1320°C (302 to 2408°F). They are made by heating selected mixtures of chemical compounds until they melt. The melt is then raised to the temperature at which the parts are to be processed. Molten salt baths are prepared in pot-type furnaces which are heated externally with gas, oil,

HARDENING AND HARDENABILITY

13.49

800

700

Temperature, °C

600

500 0.5 1.5%

0

5% ammonia

400

300

200

100

0

0

1

2

3

4

5 6 Time, s

7

8

9

10

FIGURE 13.31 Effect of ammonia contamination on the cooling curve performance of an aqueous polymer quenchant.8

or electricity. Internal heating can also be done either with immersed electrodes or suitably protected electric resistors dipping into a salt bath contained in a metal (or ceramic) pot or with gas-fired immersion heaters. These are commonly employed together with a thermocouple which dips into the molten salt bath and is adjacent to the part being treated.63 Salt baths can be used for either heating or cooling. Agitation of the bath is important to improve uniform heating and/or quenching severity. This is effected by centrifugal pumps, air bubbles, or a belt-driven propeller; the last one is widely used.64 To ensure consistent results, salt bath compositions are analyzed at regular intervals to maintain strict quality control. Bailing of the bath, coupled with replenishment with a suitable additive, is regularly done for surface-hardening molten salt baths. Sludge removal at regular intervals is also necessary. Advantages and Disadvantages of Salt Bath Process. The salt bath process offers a number of well-established advantages over alternative processes as follow:65–76 1. There are a wide range of operating temperatures from 150 to 595°C (300 to 1100°F) for one composition salt compared to the inability of oil to be used above 230°C (450°F). Hence, the latter is restricted to low-temperature processes. 2. It is simple to operate, requiring only semiskilled labor. 3. Circulation of the molten salt bath by convection and/or stirring of the melt (a) causes rapid and uniform heating for all parts at the same time, (b) maintains a

13.50

CHAPTER THIRTEEN

constant temperature, which, in turn, provides (c) a precise temperature control within ±1°C and (d) extremely uniform and reliable results. 4. It is energy-efficient because an extremely rapid rate of heat transfer is obtained by the very excellent thermal conductivity of the bath. Salt bath provides greater productivity due to rapid attainment of equalization temperature of parts compared to oil. Hence, it is well suited for quenching large parts and heavier sections without cracks by varying the temperature, agitation, and water content. This allows high-production systems to occupy a relatively small space (i.e., more compact furnace). 5. A greater range of quench severities (or rates) than in oil are possible. 6. In addition to the rapid heat-up capability, the automatic salt bath has a shutoff feature for turning the bath completely off or to a very low idle and is equipped with covers, which can yield a considerable energy saving at all temperatures. As a result of these features, the salt bath process can be regarded as one of the most cost-effective methods for high production. 7. The vapor stage (if present) in salt bath quenching is very little, and hence one can quench very rapidly past the pearlitic nose. 8. It has quite a different quenching mechanism from that of oil. Mostly the heat extraction during salt bath quenching is through the third (i.e., convection) stage liquid cooling, resulting in very low distortion and more uniform and consistent hardness. 9. There is superior thermal and chemical stability to oil with relative insensitivity to contaminants. Hence it enables one to go to a temperature for austempering where oils cannot compete. Also, it offers adequate quenching performance for many years. In contrast, oil deteriorates with use, necessitating close control and even partial or complete replacement. 10. Washing of adhering salts by water is easy, and salt recovery for reuse is simple, if desired. In contrast, washing of oil requires special cleaners and equipment, and its recovery is not simple. The salt recovery option eliminates disposal and reduces operating cost. 11. Being inflammable, salt bath does not pose a fire hazard compared to oil, which poses a serious fire hazard at a comparable temperature. 12. The capital cost of salt bath equipment is less than that for competitive methods of handling the same volume of work. The production costs with salt bath are lower due to rapid heating/cooling, cleaner surfaces, reduced cracking and distortion leading to uniform heating, and less furnace equipment. 13. Another distinct characteristic of salt baths is the ability to suspend parts from the top of the pot rather than laying them flat in a basket, and there is a buoyancy effect on the work immersed in them. The parts in salt baths weigh only about two-thirds to three-fourths of their weight in air. Thus the salt bath provides some degree of support to the components it surrounds and therefore offers better dimensional control and less distortion. Usually a distortion within 0.003 to 0.004 in. on automotive parts may be attained. (A little distortion can be corrected by post-heattreatment straightening, the amount of which is far less than would be necessary with atmospheric or vacuum processing.)74 14. The salt bath process is extremely versatile and much more flexible than other processes. This process is carried out for a variety of treatments, such as liquid carburizing, liquid nitriding, liquid nitrocarburizing, quenching or quench-

HARDENING AND HARDENABILITY

13.51

hardening, tempering, martempering, austempering, annealing, normalizing, stressrelieving of ferrous alloys, and aluminum dip brazing. It provides the treatment of small and large quantities of volume production as well as small individual piece weights. 15. There is virtual elimination of scaling, oxidation, and decarburization of the treated parts because of their full immersion in the neutral salt baths, without air contact. After processing through a chemical washing plant, parts have clean, safe, and good-quality finishes. After continued use, some salts exhibit the tendency to decarburize, which can be overcome by rectification, and self-rectifying hardening salts are available. 16. It produces a cocoon-type effect when a cold steel is immersed into a molten salt by freezing a large gob of molten salt around the part. This minimizes distortion due to uniform melting of this gob. 17. The salt bath offers speedy, reproducible microstructure, and uniform production by proper installation and operation of a salt bath. 18. Quench severity can be controlled to a large extent by varying the agitation, temperature, and water content of the salt bath. The disadvantages of molten salt bath include the following:65 1. Since salt freezes at about 135°C (275°F), it cannot be used for lowertemperature quenching applications. Rather, it has to be used above its melting point of about 150°C (300°F).67 2. Freezing can lead to intermittent use, which is complicated and energyinefficient. 3. Since nitrates are oxidizers, greater precautions are needed to avoid the introduction of combustible/oxidizable materials such as soot, oil, or cyanides and thereby the possibility of any violent reactions. Excessive overheating must be avoided; otherwise, salt can even oxidize steel or cast iron. 4. Since salt is a costlier quenchant on a volume basis, greater care must be exercised to avoid excessive dragout. Efficient washing and recycling can overcome this problem. 5. Removal of certain contaminants can be more complicated than that for conventional quenchants, although methods for removal are well-proven. In some cases, salt has the advantage over conventional quenchants with respect to contaminants. 6. Salt may seem to pose safety and environmental problems, but the newer technology removes these problems. Advancement in Salt Bath Technology. The following advancement and undergoing developments make salt bath quenching more attractive in the future:65 1. Computer-controlled robot handling systems for moving parts are much faster and more convenient from austenitizing to quench. 2. More energy-efficient austenitizing and salt bath furnaces, respectively, use superior electrode designs and better insulation. 3. Integrated agitation and heater systems in the quench baths provide more uniform flow and temperature control together with safe water addition. 4. Water addition monitoring systems use fast quench rate determination. Continuous monitoring is also expected in the near future.

13.52

CHAPTER THIRTEEN

5. Development is underway of quench salt reclaim and treating systems that allow economic recycling and reduce disposal problems. 6. Data of quantitative salt quench rate-quench variables can be used to correlate with steel hardenability. 7. There is an established family of austempered ductile irons (ADI) that are superior to steel casting and forgings and other cast irons for both wear and structural applications. Types of Salt Bath. It has been established that one salt bath will never perform the entire range of heat-treating operations. Special types of salt baths have, therefore, been used to suit different heat treatment operations and metallurgical conditions.69 As a rule, manufacturers of heat-treating salts rarely specify the composition of the salts; instead, they provide detailed information regarding their properties and uses. The salt baths can be classified into four main groups, as discussed herein.70 low-temperature salt baths. Having an operational range of 150 to 620°C, these baths are used chiefly for isothermal treatments such as austempering, martempering, low-temperature tempering of steels, and the solution treatment of Al alloys. Baths operating in the temperature range of 150 to 500°C consist of equal parts of KNO3 and NaNO2; those operating in the range of 260 to 620°C contain mixtures of KNO3 and NaNO3. The phase diagram shown in Fig. 13.32 illustrates the dependence of the melting point on the composition of the mixture.77 Like all quenchants, the quench severity of molten KNO2/NaNO2/NaNO3 salt mixture depends on the agitation rate and temperature, as shown in Fig. 13.33.78 These baths are readily water-soluble and strongly oxidizing, especially in the higher temperature range. Salt bath agitation can be used in several ways. One

FIGURE 13.32 Freezing points of ternary alkali nitrate-nitrite mixtures with composition (%).77

HARDENING AND HARDENABILITY

13.53

FIGURE 13.33 Effect of agitation on the quench severity of molten salt.78

FIGURE 13.34 Salt circulation system in a molten salt bath for martempering.79 (Courtesy of the Institute of Materials, London.)

method of effective salt circulation with violent downward flow of liquid salt within the working space is described by Liscic and shown in Fig. 13.34.79 By using a twospeed electromotor-driven propeller pump, it is possible to achieve agitation rates of 0.3 and 0.6 m/s in the working space.79 Another method uses a dual-impeller bath agitation system (Fig. 13.35).80 Figure 13.36a illustrates a newly developed device for automatic continuous addition of small quantities (0.1 to 2.7%) of water into the hot salt bath, which improves its quenching speed.79,81 The device comprises three essential portions: the probe for temperature measurement, the electronic control, and the system for water addition. To evaluate the effect of water addition and the agitation rate on the quenching intensity of the hot salt bath (Degussa AS-140) at 200°C, tempera-

13.54

CHAPTER THIRTEEN

FIGURE 13.35 Dual-impeller salt bath agitation system.80

ture versus time curves have been determined in the center of a 25-mm-diameter steel specimen, and relevant curves have been calculated.79 For safe working conditions without explosion hazard, the recommended working temperature of 180 to 250°C (355 to 480°F) should not be exceeded.82 Figure 13.36b shows the effect of the agitation rate and of 1 vol% water addition on the maximum cooling rate of a hot salt bath. It is vital to note that all cyanides and all organic chemicals must be kept out of these baths. The baths, when satisfactorily maintained, have low viscosity and are excellent for the precise control needed in martempering and austempering. They can also be used for tempering, nonferrous annealing, steel blueing, general treatment of aluminum, and heat-treating copper and beryllium. The cooling rates of nitrate salt baths are similar to, or even faster than, those of quenching oils. In addition, nitrate quenching baths have longer operating life. Thermal decomposition of nitrate and subsequent reaction with atmospheric CO2 form carbonates which decrease the quenching severity. However, reestablishment of the bath to a neutral condition can be achieved by adding a rectifier to the nitrate salt.64 One advantage of the use of nitrate/nitrite-type salts is their 90% recovery or recycling from the spent salts in washing waters. It saves not only the original cost of salt but also the problems caused by putting the nitrate down the drain and the cost of shipping it out to a safe disposal site.72 medium-temperature neutral baths (650–1000°C). These baths are generally binary or ternary chloride baths comprising KCl, NaCl, BaCl2, or CaCl2. The typical compositions and working temperatures for these types of salt baths are 45% NaCl + 55% KCl (675 to 900°C) and 20% NaCl + 80% BaCl2 (675 to 1060°C), which show that the composition of the salt bath mixtures is based on the required operating temperature range. The melting point of the bath is around 550°C, and these baths can be operated up to a maximum of 1000°C. When freshly prepared, these baths are inert to the steel surface so that neither decarburization nor carburization of the surface occurs. However, with prolonged use, atmospheric oxygen becomes absorbed by the molten salt, resulting in diminution of passivity and subsequent

HARDENING AND HARDENABILITY

H2O—regulation by impulse Duration of the impulse gives the quantity of water to be added (t/min)

Distur bance Operation H2O—inflow Control cabinet 0.1

13.55

Display of the actual value Adjustment of the required value Main switch Water inflow

Solenoid

Water supply

Probe Valve for pressure reduction AS 140 Hot salt bath

Control electric wire (a) 50 Cooling rate calculated from temp-time curve measured in the center of a 25 mm dia steel specimen.

Maximum cooling rate [K/s]

45

1% (vol) water added

40

without water addition

35

30 0

0.3 Agitation rate [m/s] (b)

0.6

FIGURE 13.36 (a) Automatic system developed by Degussa for continuous addition of water to molten salt.82 (b) Maximum cooling rate of a hot salt bath (Degussa AS-140) of 200°C as a function of the agitation rate and percentage of water addition.82

13.56

CHAPTER THIRTEEN

decarburization. To avoid this difficulty, a rectifying agent such as borax or boric oxide is introduced into the bath at regular intervals, which converts the oxychlorides to a sludge; this sludge is then periodically removed. Note that these rectifiers are not suitable for baths using mixtures of chlorides and carbonates. In that case, methyl chloride gas should be used as a salt bath rectifier. This type of bath is employed for austenitizing carbon-, medium-, and high-alloy steels operating at temperatures up to a maximum of 1000°C. It can also be used for annealing stainless steel, ferrous and nonferrous annealing, cyclic annealing, and normalizing. It is usual practice to carry out tests at intervals for hardening characteristics and the presence of decarburization of the tool steel surface to retain a neutral condition of the bath. high-temperature salt baths (1000–1300°C). This type of bath is employed for austenitizing the high-speed steels (T and M series) and the hot-work steels (H series). These baths contain mixtures of barium chloride (BaCl2), borax or sodium tetraborate (NaB4O7), sodium fluoride (NaF), and silicates. These salt baths are heated by utilizing the electric resistance of the salt mixture itself. The proprietary baths usually require introduction of a rectifying agent such as graphite or ferrosilicon, to prevent decarburization of high-speed steels as the bath ages. The melting range of the salt baths is usually between 870 and 1040°C. The high-melting-point salts do not drain well from the part and leave a layer that is somewhat difficult to remove. Since the bath develops the decarburizing tendency after aging, tests are carried out periodically by quenching the test specimens in salt bath and by simply checking the surface by file. A “file soft” surface indicates the need for more rectification. This test can be combined with chemical analysis for barium oxide (BaO) content, which indicates the alkalinity of the bath. Another test for the decarburizing condition is performed with a double-edge carbon steel razor blade dipped in the bath. By changing the immersion time or temperature, a good prediction of bath conditions can be designed. If upon bending, the blade breaks after immersion, it means an absence of decarburization. On the other hand, if it bends, then the bath is decarburizing. Of course, stainless steel blades will not hold for this test.66 cyanide- and noncyanide-bearing salt baths for thermochemical surfacehardening treatments. These baths are used for liquid carburizing and carbonitriding, nitriding, and nitrocarburizing of steel surfaces, which are treated in detail in Chap. 16. Safety Precautions of Salt Baths. The following recommendations should be followed to protect the operating personnel from any hazards:63,67 1. Good exhaust around the salt bath is necessary, especially during charging of fresh salt and during the quenching operation. 2. Operators should always stand behind the furnace shields and wear face masks and gloves while salt or work is either introduced into or removed from the furnace. 3. To prevent salt from spurting or sudden expulsion from the molten bath, never introduce wet salt, wet work, or water into the high-temperature molten bath. That is, parts, fixtures, and conveyor entering a quench bath must be dry and free of any moisture, oil, or other liquid. 4. Before switching off furnaces using an externally heated pot, reduce the salt level so that the pot is no more than two-thirds full (depending upon the position of the burners), to avoid the risk of spurting upon remelting. Place the cover on the pot during remelting or cooling. Neutral salts, baled out to comply with the above,

HARDENING AND HARDENABILITY

13.57

are normally reused when the furnace is restarted, provided that it has not been allowed to become damp in the meanwhile. 5. Dry out electrically heated brick-lined furnaces with an electric heater prior to restarting after use, to avoid spurts owing to entrapped moisture in the brickwork. 6. Never heat salts containing nitrate above 550°C. 7. Never mix salts containing cyanide with salts containing nitrates; otherwise, an explosion will take place upon heating. 8. Salts containing cyanide or barium are very poisonous. Do not take food where these salts are being used, and always wash the hands thoroughly prior to handling food. 9. Keep salts containing cyanides away from acids. 10. Handling and disposal of cyanide products are done by using special clothing, special handling, and state-of-the-art methods. 11. No water sprinkler should be installed near a molten salt system. No water or liquid extinguisher should be used in case of fire. 12. The salt bath protection from accidental overheating should be provided by installing audio/video alarms which go on when temperature is in excess of a set limit. If temperature rises above 595°C (1100°F), thermal breakdown of salt and the resultant reactions with the salt container may lead to salt leakout. 13. Salt should be stored in well-marked, sealed containers in a dry place and away from incompatible materials such as cyanide salts. 14. When parts are quenched from an atmosphere furnace into a salt bath, splashing of salt into the furnace should be prevented via a salt curtain, an intermediate chamber, or a vestibule. Maintenance of Salt Baths. Molten salt baths are usually maintenance-free, primarily due to their excellent thermal stability and their tolerance for the contaminants. They offer consistently satisfactory performance for many years by the addition of new or recovered salt to replenish the dragout salt. The quantity of dragout salt is a function of the mass and the configuration of the parts and fixtures or conveying system. However, this is in the range of 50 to 100 g/m2 (⬃1 to 2 lb/100 ft2). Contaminants such as metallic debris from the parts and carbonate formation may build up over a prolonged time. Agitation provides suspension of fine metallics, and when such suspension exceeds 0.5%, the quench severity is influenced. Similarly, when carbonate exceeds its solubility limit, it starts to slow down the quenching speed. Both metallics and carbonates can be eliminated by desludging the bath. The easy way to do this is to lower its temperature as far as possible, shut off agitation and heating, and wait for some period. This permits contaminants to settle to the bottom, where they are then scooped out as sludge. If water addition is the usual practice, the water content needs to be checked periodically. Water addition can be decreased or increased based on the desired quench severity. The melting point of the bath should be checked only occasionally, because it rarely changes over time. However, if any significant increase is observed, it may suggest thermal breakdown of the salt, perhaps due to unintentional overheating. The reason for the overheating should be examined and rectified.67 13.2.3.6 Lead Bath. With its low melting point (327°C, or 620°F), lead can be employed at temperatures up to 1200°C; however, it is commonly used up to a

13.58

CHAPTER THIRTEEN

maximum temperature of 900°C. Since lead oxidizes readily, particularly with increasing temperature, sludge so produced forms a coating or scum on the surface of the steel being treated and is a problem, especially at higher temperatures. Therefore it is necessary to cover the surface of the bath with charcoal or a similar reducing agent above about 480°C (900°F), to minimize the oxidation. The advantages of a lead bath over a salt bath are as follows: 1. A lead bath neither contains nor picks up moisture. Hence scaling or oxidation of the steel surface does not occur. 2. There is no danger of an explosion resulting from the hidden formation of steam as new lead is added to replenish the bath. 3. There is an absence of chemical attack on steel. 4. There is rapid heating and cooling. 5. It possesses a high stability (excluding its oxidizing tendency). The disadvantages are as follows: 1. In view of the lower density of steel compared to that of lead, the steel parts must be held down by fixtures in the bath. 2. The rapid oxidation of the molten lead surface and the retention of lead oxide, mainly at the surface, are an adverse factor. 3. Being extremely toxic and health-hazardous when small traces are either inhaled or ingested, lead requires elaborate precautions and permission by the factory inspector for its use. 4. All parts and fixtures must be dry when immersed in the bath, to avoid the steam formation and resultant violent expulsion of the molten lead. The method is used for specialized applications, such as files, and rapid local hardening and selective tempering where only a portion of the tool needs to be fully hardened.16 13.2.3.7 Pressure Gas-Quenching. Currently high-pressure gas-quenching has become the fastest-growing technology in heat-treating of components made from hardenable stainless steels, tool and die steels, and oil-hardening tool steels.18,20,83 In the quench loop of a typical vacuum furnace, cold nitrogen, argon, helium, or hydrogen gas flows, under pressure (>5 bar), from the top to the bottom through the hot zone of the vacuum furnace, where it picks up heat from the hot workload. The heated gas is then passed out of the hot zone into a heat exchanger (water-cooled or through refrigerated coils) and is recirculated into the furnace chamber through the blower and directed at the work in a continuous fashion (Fig. 13.37).84 The cooling rate of the metal depends on the surface area and mass of the part and the type, velocity V, and pressure P of the cooling gas, according to the formula hg = C(VP)m, where hg is the heat-transfer coefficient for a given gas and m and C are constants depending on the furnace design, part size, and load configuration (m = 0.6 to 0.8).79 Thus, the cooling rate can be adjusted and controlled by varying the type, velocity, and pressure of the gas, thereby offering a significant flexibility. The most important factors required in the design of a gas-quenching system are (1) the design of the gas circulation system to minimize distortion by using turbu-

HARDENING AND HARDENABILITY

13.59

FIGURE 13.37 Vacuum furnace quench loop.84 (Courtesy of Wolfson Heat Treatment Center, England.)

lent flow normal to all component surfaces, (2) the type of quenching gas and its pressure, in order to achieve the appropriate cooling rate, and (3) the specification of any gas recycle system to reduce the cost.85 Gas-quenching units are designed for either batch or continuous processing. Usually hydrogen or helium gas is used, both to produce clean, bright surface and to increase the heat-transfer rate between the gas and the workload. However, hydrogen has the greatest heat-transfer properties, as shown in Fig. 13.38a.83 Compared to nitrogen (at 6 bar), hydrogen has 30% shorter cooling times and 40 to 50% higher heat-transfer coefficient. Hence, more heat treaters are using H2-N2, He-Ar, and He-N2 mixtures.84 The high cost of argon permits its use in situations where chemical inertness is required, such as aerospace parts, made of titanium, tantalum, or niobium.86 High-pressure gas (5-bar helium) of solution-treated alpha-beta Ti-6-Al-4V alloy castings in a vacuum furnace, together with improved circulation through the use of turboblower, has proved successful in achieving the very rapid cooling rates required for subsequent age hardening to obtain high strength. These titanium alloys with optimum mechanical properties (high strength and adequate ductility) are finding numerous and increasing applications in aircraft and aerospace parts.87 However, high-pressure gas-quenching technology (e.g., helium at 20 to 40 bar and nitrogen up to 10 bar)88 can provide quench severities comparable to those of conventional recirculated oil. Very high pressure (e.g., hydrogen at 50-bar pressure) can produce heat-transfer coefficients even greater than that for water (3000 to 3500 W/m2·K, or 530 to 620 Btu/ft2·hr·°F), as illustrated in Fig. 13.38b. In addition to single-chamber furnaces with high-pressure gas-quench systems for horizontal and vertical loading, double-, triple-, or multiple-chamber furnaces are now available with oil-quench or gas-quench chambers or both.89 Multichamber systems and directed gas flows also can be employed for selective, partial, or directional hardening of parts such as a soft-shank drill.90 Advantages of high-pressure gas-quenching in vacuum furnaces over liquid quenching include91–93

1200

Temperature, °C

1000

800

600

0

6

12 18

24

Argon

Nitrogen

200

Helium

Hydrogen

400

30

36 42

48

54

60

Time, minutes

(a)

Heat Transfer Coefficient ¥ 1000 (W/m2K)

6

Argon Nitrogen Helium Hydrogen

5 4 3 2

transverse gas-flow cylinder w = 0 m/s, T gas = 200°C

1 0

0

10

20

30

40 50 60 70 Pressure (bar)

80

90 100

(b)

Frequency

Gas quenching

Oil quenching

Dimensional change (c)

FIGURE 13.38 Comparisons of (a) cooling properties of common gases,83 (b) various high-pressure gas quenchants, and (c) distortion (or dimensional change) in quenched parts between gas-quenching and oil-quenching. [(a) and (b): Courtesy of Leybold Durferrit GmbH.]

13.60

HARDENING AND HARDENABILITY

13.61

1. Optimum hardening 2. Dramatic reduction of distortion (Fig. 13.38c) and crack-free hardening of the workload due to more uniform cooling rates 3. Absence of oxidation, decarburization, or carburization on the surfaces of the cooled parts 4. Pleasant working operations due to clean operation without heat radiation; fume, fire, and explosion hazards; ventilation requirement; and environmental pollution 5. Reduced labor and total production costs, and excellent reliability and reproducibility by accurate microprocessor control 6. Reduced or no posthardening and finishing costs 7. Use in quenching a wide variety of materials by changing the cooling rate during a single cycle 8. Ability to provide cleaner parts, without posttreatment cleaning and attainment of reproducible results from batch to batch The shortcomings of gas-pressure quenching over conventional liquid quenching are that94 1. Larger cross sections of some oil-hardening grades produce lower tensile properties, ductility or fracture toughness, and hardenability. 2. Certain carbon and alloy-steel grades must be liquid-quenched irrespective of their cross sections (such as 1045, 1075, 4130). Applications. The following parts made from various high-speed steels, coldand hot-worked steels, and air-hardening steels have been successfully hardened with high-pressure gas-quenching.92,95 High-speed steel: For drills, taps and reamers, hobs, shapers, shavers, milling cutters, flat and circular broaches, dovetail and circular form tools, and end mills having high-red hardness, good toughness, and excellent wear-resistance. Cold-work steels: For progressive die stamping and forming, motor lamination production, can manufacturing, powder compaction, recycling and slitting (knives), and fine blanking and rollforming requiring wear-resistant tool steels with good toughness and high hardness. These materials include 12% Cr steels; X155CrVMo12 steel; X210Cr12 steel; 90MnCrV8 steel; A2, D2, and M2; and high-vanadium P/M tool steels. Plastics-process steels: For mold/holder blocks, equipment for compounding extrusion injection molding, and pelletizing/granulating knives from P20 types, H13, S7, NAK55 (Ni + Al PH), 17-4(or 15-5)PH, maraging steels; A2, D2, 440C; highvanadium P/M steels (9V, 10V, 440V, M390, etc.). Hot-work steels: For aluminum die-casting molds, compression-molding dies, and hot-forging dies. Air-hardening steels: For dies and tools from X37CrMoV51; X45NiCrMo4; A2, D2, and M2 grades; and high-vanadium P/M tool steels. Other applications are for solution annealing/treating and hardening of aerospace/industrial turbine materials which include 300 series, 400 series, and PH stainless steels; maraging steels; iron-base superalloys; nickel-base superalloys; and high-strength (300M, 4340, etc.) alloys.94

CHAPTER THIRTEEN

13.62

Note that when endo gas atmosphere is already available, the cost of a vacuum furnace is very high. 13.2.3.8 Fluidized-Bed Quenching. A typical fluidized bed consists of a furnace system (shell, heaters, and insulation) and a quench system (gas diffusion assembly, fluidized-bed support, and gas) (Fig. 13.39). The fluidized bed is produced by blowing a gas such as nitrogen, argon, helium, hydrogen, or carbon dioxide through a solid support such as alumina, iron, copper, and molecular sieves.8,96,97 Heat transfer within a fluidized bed depends on the particle size of solid support (or medium), volumetric heat capacity, thermal conductivity of the gas, and gas flow rate through the bed. The effect of fluidized-bed variables on the heat-transfer coefficient is summarized in Table 13.7.71 The cooling curve behavior during fluidized-bed quenching can be described by the equation:78 T - Tf Ê Aht ˆ = expÁ ˜ Ë CpVr ¯ Ti - Tf

(13.12)

where Tf is the temperature of the fluidized bed, Ti is the initial temperature of the part being quenched, A is the cooling surface area, V is the volume of the metal

FIGURE 13.39 Cross section of an electric fluidized-bed furnace.8,46

TABLE 13.7 Effect of Fluidized-Bed Variables on Heat Transfer71 Parameter Particle size d Volumetric heat capacity Cv Gas conductivity k Gas flow rate Ug †

Effect†

Comment

d ≠, Ø Cv ≠, ≠

Valid for d > dcrit Al2O3

k ≠, ≠ amax for Ug ⬇ 5Umff

He, H2

Umff is minimum gas (fluid) flow rate.

HARDENING AND HARDENABILITY

13.63

(probe), r is the density of the metal, Cp is the heat-capacity of the metal, t is the cooling time, and h is the heat-transfer coefficient. The important variables in the selection of a fluidized bed are the medium, particle size, temperature, and gas. The two variables that strongly affect heat transfer are the thermal conductivity and the flow rate of the gas, as shown in Fig. 13.40.79 Clearly, helium and hydrogen provide much greater potential for cooling rate variation compared to the most commonly used nitrogen (and argon) gases utilized for fluidized beds. Reynoldson has shown that air-fluidized beds produce cooling rates intermediate between those displayed by a mineral oil and by a low-pressure vacuum quench.98 Other workers have shown that air-fluidized beds can produce the same hardness as oil and salt for several steels.99 A recent development in the fluidized-bed furnace includes computerized control of fluidization to optimize heat transfer/uniformity while minimizing gas usage as well as use of a specially-designed nitrogen-purged transfer hood to protect workload during transport from fluidized bed to quench facility, in order to achieve good surface finish.100 The advantages of fluidized-bed quenching over molten metal and molten salt bath quenching are the improved process control, improved process safety, flexibility, cleanliness, and reduced pollution and fire hazards. Cost is a significant shortcoming because the fluidizing gas is usually not recycled. 13.2.3.9 Spray Quenching. The term spray quenching refers to various quenching processes that allow a high rate of heat extraction from the hot metal surface by the impingement of a fast-moving stream of a quenchant medium, in order to produce a metastable structure with the required physical properties and development of the desired distribution and level of stress.101 Hence, the time-temperature history of the part must be precisely controlled during quenching. In spray quenching, the heat-transfer coefficient from the part to the quenching medium is a direct function of the flow rate, turbulence, and impingement pressures of the quenchant at the hot surface. By adjusting these parameters during the quench, a cooling profile (or a controlled cooling operation) can be obtained which is not possible by

300 He

2

h (600°C), Btu/hr·ft ·°F

H2 200

100

N2

1 2 3 4 5 6 7 8 9 10111213141516 Fluidizing flow Minimum fluidizing flow

FIGURE 13.40 Effect of fluidizing gas on heat-transfer properties.79 (Courtesy of the Institute of Materials, London.)

CHAPTER THIRTEEN

13.64

other methods (such as immersion quenching).8,102 The rate of heat extraction can be varied over a wide range by changing the amount of sprayed liquid (e.g., by varying water and air pressure). Examples of sprayed liquid include (1) a mixture of water (or other volatile liquid) droplets to a gas-quenching stream (fog quenching), (2) water, or water/air streams, (3) sprays of other volatile liquid quenchants other than water, and (4) high-pressure jets of oil, water, or aqueous polymer solution under the liquid level in a bath.8 By employing a CNC control system, it is possible to change continuously the heat flux extracted and to follow a predetermined cooling curve.79 Figure 13.41 shows the schematic heat transfer in the vapor-blanket stage for water-spray cooling of a heated metallic surface. The heat-transfer coefficient to water hw comprises an internal heat-transfer resistance inside the water layer and a capacitative transport resistance for the quantity of water: hw =

1 ˆ Ê 1 + Ë hwi m ˙ sCw ¯

-1

(13.13)

· s is the flux where hwi is the internal heat-transfer coefficient of water (W m-2 K-1), m of sprayed water (kg m-2 s-1), and Cw is the specific heat capacity (J kg-1 K-1).79 A well-established application of spray cooling, comprising 19 banks of water sprays covering a distance of 88 m, is the runout table of the accelerated cooling of hot strip mill in the integrated works of British Steel Strip Products, Port Talbot.103 13.2.3.10 Cold Die Quenching. Thin flat disks, long slender rods, thrust washers, and thin blades (with a large surface area and relatively small weight) are usually not quenched in conventional liquid quenchants owing to the occurrence of excessive distortion. Rather, they are often quenched by squeezing between a pair of water-cooled, cold copper or beryllium-copper die blocks to avoid any distortion, where die blocks provide necessary quenching action together with maintenance of

Metal surface

Accumulated spray-water Sprayed water vapor q·V ·q L q·

q·W

m· S

Str

d q0 q

w

· m F qSi q W

qW

FIGURE 13.41 Heat transfer in the vapor-blanket stage for spray cooling, according to Jeschar et al. q·v = boiling heat flux density, q·L = conduction heat flux density; q·str = radiation heat flux density; q·w = heat flux density transferred to water; m·s = flux of sprayed water; m·F = flux of water flowing off; d = thickness of vapor blanket; q0 = surface temperature; qsi = boiling temperature; qw = water temperature; and w = velocity of fluid. (After R. Jeschar, R. Maass, and C. Kohler, in Proceedings of the AWT–Tagung Induktives Randschichtharten, 23–25, March 1988, Darmstadt, pp. 69–81.)

HARDENING AND HARDENABILITY

13.65

flatness. In this method, various forms of cold, flat, or shaped dies are used in order to suit the shape of the part being quenched.6 13.2.3.11 Press Quenching. Quenching presses are used for controlled quenching of ring gears, bearing races, and other round, flat, or cylindrical parts, to accomplish heat treating with minimal distortion. In press quenching, the die contacts the hot and plastic part, and the pressure of the press aligns the part mechanically. The machine and dies then force the quenchant into contact with the part in a controlled manner. The amount of quench oil and its flow rate are controlled by the press. A uniform and high volume of oil flow around all parts is directed by the dies. Since quenching speed is a function of the quenching medium and the rate of oil flow, the cooling rate can be controlled by adjusting the rate of oil flow through the die.104 Although some basic types of dies are available for certain workload shapes, additional dies are usually needed to accommodate specific parts. Quenching to a tolerance of 0.025 to 0.050 mm (0.001 to 0.002 in.) and 0.025 to 0.125 mm (0.001 to 0.005 in.), respectively, for roundness and flatness is common practice for ring gears and bearing races when proper equipment and correctly designed dies are used.6,104 The use of press-quenching processes together with intense quenching using water has been developed. This significantly reduces cost and applied loading to reduce distortion compared to oil-quenching.105

13.3 QUENCH-HARDENING 13.3.1 Quench-Hardening of Steel Quench-hardening consists of heating to a temperature 30 to 50°C above Ac3 for hypoeutectoid (plain carbon and low-alloy) steels and above Ac1 for hypereutectoid (plain carbon, low-alloy, and numerous tool) steels, soaking at that temperature for about 1 hr/in. section thickness, followed by quenching in a liquid quenchant (such as water, brine, oil, or polymer solution) to produce a martensitic and/or bainitic structure throughout the entire component. The cooling rate necessary to enable austenite to transform fully into martensite must be equal to, or greater than, the minimum rate of cooling, usually designated as the critical cooling rate to ensure maximum hardening.106 Figure 13.42a shows the conventional quenching and tempering superimposed on the TTT diagram.17 In some of the very-high-alloy tool steels such as M1- and T1-type highspeed steels, austenitizing is carried out at a temperature greatly in excess of Ac3 for such a time (typically 2 to 5 min) that a maximum solution of the alloy carbide occurs, without undue coarsening of the matrix.107 But preheating of these parts for a longer time is required prior to high-temperature austenitization (see also Chap. 14). In some more highly alloyed steels, where the position of the pearlite nose of the TTT diagram is far to the right, full hardening can be obtained by a slow cooling rate, such as would be encountered in air cooling. These grades of steels are usually called air-hardening steels (see also Chap. 12). Advantages of conventional hardening of steel by rapid quenching in water or brine are that (1) the method is very simple, (2) minimum equipment and an easy control are needed, and (3) the quenchants are inexpensive.

FIGURE 13.42 Typical curves for (a) conventional quenching and tempering, (b) martempering, and (c) modified martempering superimposed on TTT diagrams.17 [(a) and (b): Courtesy of Association of Iron and Steel Engineers, Pittsburgh, Pennsylvania; The Making, Shaping and Treating of Steel, eds. W. D. Lankford and H. E. McGannon, United States Steel, 10th ed., 1985.]

13.66

HARDENING AND HARDENABILITY

13.67

FIGURE 13.42 (Continued) Typical curves for (a) conventional quenching and tempering, (b) martempering, and (c) modified martempering superimposed on TTT diagrams.17 [(a) and (b): Courtesy of Association of Iron and Steel Engineers, Pittsburgh, Pennsylvania; The Making, Shaping and Treating of Steel, eds. W. D. Lankford and H. E. McGannon, United States Steel, 10th ed., 1985.]

One serious disadvantage of this method, often observed in large-sized parts, is that after the conventional hardening of steel parts by rapid quenching, severe residual stresses occur, as a result of large variations in cooling rates across its entire cross section. This arises from a steep thermal gradient between the surface and center of the material as well as from nonuniformity in transformation. These stresses can produce an excessive distortion and can even cause cracking of the materials. (See also Chap. 17 for more elaborate discussion on residual stress, distortion, and cracking.) The factors responsible for successful hardening of a specific part are (1) the heat-transfer characteristics of the quenchant, (2) the quenchant use conditions (such as bath temperature and flow velocity), (3) bath loading, (4) section thickness of the part, and (5) transformation characteristics of the particular alloy being quenched. Successful hardening normally means attainment of the required hardness, strength, and toughness with the minimum residual stress, distortion, and likelihood of cracking.108

13.3.2 Quench-Hardening of Iron Castings Gray iron castings to be hardened are austenitized in a gas- or oil-fired furnace or in a salt bath at temperatures normally 30 to 55°C or sometimes as much as 95°C

13.68

CHAPTER THIRTEEN

(175°F) above the calculated A1 temperature for 2 min to 1 hr per inch of section thickness (depending on their size and shape), followed by quenching in an oil, molten salt, or polymer solution. However, for high-alloyed irons, forced-air quenching is often the choice. Immediately after quenching, both gray and ductile castings are normally tempered at temperatures well below the transformation range for about 1 hr/in. section thickness.109 The calculated A1 temperature of unalloyed gray iron is given by17 A1 (∞C) = 730 + 28.0(%Si ) - 25.0(%Mn )

(13.14a)

A1 (∞F ) = 1345 + 50.4(%Si ) - 45.0(%Mn )

(13.14b)

Ductile irons are usually austenitized at a temperature of 845 to 925°C (1550 to 1700°F) and quenched in oil to minimize stresses. Nevertheless, water or brine can be employed for simple shapes.110

13.4 INVERSE QUENCH-HARDENING OF STEEL The pattern of hardness distribution in round oil-quenched bearing-grade steel parts with relatively small mass and high quenchability has been found to exhibit lower hardness values at the surface than at the core. This phenomenon, called inverse quench hardening by Shimizu and Tamura, is attributed to the transformation of pearlite at the surface.110a–e Liscic and his coworkers110d,110e have also observed the inverse quenchhardening phenomenon in their controllable delayed quenching (CDQ) technology polymer quench tests on AISI 4140 cylindrical specimens. Their explanation for the inverse hardening phenomenon agreed well with that of Shimizu and Tamura.110a Such inverse hardness distribution obtained after quenching, when tempered at an appropriate tempering temperature, offers uniform microstructure of tempered martensite in the entire crosssection of the workpiece, providing the maximum ductility and toughness. The results of the bending fatigue tests and the impact loading behavior obtained in this investigation have shown an increase in the bending fatigue life by a factor of about 7, and an impact energy increase of about 7% in specimens with inverse hardness distribution, compared to specimens with normal hardness distribution.110d

13.5 DIRECT QUENCHING For some applications such as coil springs, roll-formed balls, axle beams, steering and suspension parts, and track shoes, direct quenching after hot forging (or from high austenitizing temperature) provides a reduction in heat treatment cost by removing reheating for hardening. Direct quenching can be applied to carbon, boron-treated, and low-alloy steels and is mostly followed by tempering.111 It is also used for rapid cooling of metals from the elevated solutionizing temperature.

HARDENING AND HARDENABILITY

13.69

13.6 INTENSE QUENCHING OF STEEL Intense quenching (IQ) may be defined as a very rapid, uniform quenching process with very high agitation rate obtained by high-pressure quenchant jets or rapid impeller agitation systems, to produce a maximum surface compressive stress and very hard as-quenched surface. This occurs when the Biot number ≥18. The IQ should be stopped at the point of formation of maximum surface compressive stresses, which corresponds to the attainment of maximum tensile stresses in the core (which, according to some researchers, is at 450 to 500°C core temperature).79 Successful implementation of IQ practices is associated with the emergence of a huge amount of dislocations (according to Ivanova) and the resulting physical property improvements.79,112 Pressurized jets of water, oil, 5% caustic soda, 3% sodium carbonate, and various concentrations of polymers can serve as quenchants. Here, H values of water can reach in the range of 2.5 to >5.0 and for oil >0.70.113 Intense steel quenching technologies have become an increasingly interesting alternative to heat treatment of steel. It will be accomplished for autoparts (semiaxles, engine crankshafts, gears, spherical journals, bearing rings, springs, moil points); tools (dies, punches, etc.); fasteners (nuts, bolts, washers, etc.); large steel parts for machine, building, and power industries; mining equipment parts; various types of forge quenching; and cable wire quenching. IQ has a great future owing to numerous advantages derived by using this technology. The advantages of IQ over conventional water-, water-polymer solution, and oil-quenching are112,113 1. Optimum quench uniformity achieved throughout the section thickness of the part. 2. Formation of nearly 100% martensite in the surface layers of a part and in critical areas of cheaper, shallow-hardenable steel parts (as in root fillets of a gear) (i.e., cost savings by allowing the replacement of alloy and high-alloy steels with cheaper 1040 steel, while attaining greatly enhanced ductility). 3. Development of maximum residual compressive stress value of about 1200 MPa (173 ksi) and the associated transformation of austenite to martensite, producing additional strengthening or superstrengthening of the material. 4. Increased hardening depth and high surface hardness for a given steel grade, which provide a reduced production cost of steel (Fig. 13.43).113 5. Minimum heat-treating distortion (e.g., 0.003 to 0.006 in./in. distortion is produced in an intensely oil-quenched AISI 1141 steel part against 0.015 to 0.022 in./in. in a normal oil-quenched part). This offers the elimination of many posthardening, straightening, or postquenching operations on steel parts.79 6. Increased safe life of machine steel parts by 3 to 4 times over oil-quenching.112 7. Effective and economical prevention of quench cracking in both batch processing of small parts and continuous conveyor lines.112 8. Maximum long-life bending and/or torsional fatigue properties of hardened parts. 9. Attainment of highest yield strength/tensile strength ratios with good toughness in the intensely quenched and tempered condition. 10. High abrasion resistance.

CHAPTER THIRTEEN

13.70 70

Intense Water Quench

Hardness, Rockwell C

60 50

Jominy

40 30

Free Water Quench

20 10 0

50 100 150 200 250 300 350 400 450 500 Distance below surface or from quenched end of jominy (in. ¥ 10–3) (a)

70

Hardness, Rockwell C

60 50 40

Jominy Intense oil quench

30 20

Free oil quench

10 0

50 100 150 200 250 300 350 400 450 500 Distance below surface or from quenched end of jominy (in. ¥ 10–3) (b)

FIGURE 13.43 Typical advantages of (a) intense water-quenching for 3-in.(76.2-mm-) diameter bars of AISI 1045 steel, H > 5.0, and (b) intense oilquenching for 2.5-in.- (63.5-mm-) diameter AISI 1141 steel, H > 0.70.113 The term free denotes mildly agitated. (Courtesy of R. Kern.)

11. 12. 13. 14. 15.

Use of water or water-polymer solution instead of expensive oil quenchants. Increased productivity during manufacturing. Protection from dangerous fires. Ease in automation of stable technological process. Possibility of combining turnery work with heat treating when using induction heating.

HARDENING AND HARDENABILITY

13.71

Table 13.8 lists the surface hardness and bending fatigue life of a 2.25-in.- (57.1mm-) diameter steel shaft [at applied stress of 50 ksi (345 MPa)] of varying compositions in the differently heat-treated condition.113 The disadvantages of intense quenching include the following:113 1. Intense quenching equipment is expensive. 2. Quenching of individual parts—or, at the most, 2 to 3 parts at a time—is possible compared to those of 6 to 12 parts (or more) quenched simultaneously by conventional method (Lowerator mechanism). For successful operations of intense water-quenching, the following guidelines should be adopted: 1. Parts to be intensely quenched should be finish-machined except for final grinding. 2. Proper austenitization should be carried out in a gaseous atmosphere furnace or by induction heating. In the former case, the carbon potential of the furnace atmosphere should match closely (±0.05%) the carbon content of the part. 3. Part and holding fixtures should be preheated before loading in the austenitizing furnace. 4. After austenitization, transfer of the part to the appropriate quenching fixture should be rapid. Figure 13.44 is a schematic section drawing of a scanning-type intense quenching fixture for a round bar. A typical intense oil-quenching fixture for a gear is shown in Chap. 17 (Fig. 17.11).113 5. The quenchant should be flowing prior to loading of the hot part into the quenching fixture. 6. Once an intense quenching is set up with precise control of the quench pressure, scanning rate, and so forth, a start button is pressed. (See Chap. 17 for more discussion on better quenching fixture design.) 7. Vapor pockets should not be allowed to form. 8. The maximum carbon contents of steel to be used successfully with IQ are as follows: Water-quenching: Complex shapes (e.g., splined shafts) Simple shapes (e.g., round pins)

0.38% No limit

TABLE 13.8 Results of Hardness Values and Full-Size Bending Fatigue Tests on a 2.25-in.(57.1-mm-) Diameter Shaft at an Applied Stress of 50 ksi (345 MPa)113 Material 8650H 81B40 1045 8620H 1045‡

Heat treatment

Hardness, Rc

Life, cycles

Oil quench and temper Oil quench and temper Intense quench and temper Carburize and harden (intense quench) Intense quench and temper

32–34 30 43–45† 63 42

3.5–4.0 ¥ 105 2.9–4.8 ¥ 105 2.3–4.0 ¥ 106 1.8–2.3 ¥ 106 3.8 ¥ 105

Note: The hardened depth to 40 Rc on the 1045 shafts was approximately 0.13 in. † Decarburized 0.0015 in. deep. ‡ Reprinted by permission of Fairchild Publications, New York.

CHAPTER THIRTEEN

13.72

FIGURE 13.44 Schematic section drawing of an intense quenching fixture of the scanning type showing the completion of quenching of part.113 (Courtesy of R. Kern.)

Oil-quenching:

Complex shapes (as above) Simple shapes (as above)

0.44% No limit

9. Correct design of the part is also necessary to achieve its superior engineering properties after heat treatment. (See Chap. 17 for more details.)

13.7 MARTEMPERING OF STEEL Martempering is a form of delayed or interrupted quenching. Martempering of steel consists of (1) quenching steels from the austenitizing temperature into a hot bath (such as hot quench oil, molten low-temperature nitrate/nitrite salt bath, molten lead bath, or a fluidized particle bed) maintained at a temperature slightly higher (⬃10°C) than the Ms temperature; (2) holding at this temperature in the quenching medium for a sufficient period of time (but within the incubation period for bainite at this temperature) to reach the equalization of the temperature of a section (i.e.,

HARDENING AND HARDENABILITY

13.73

within the part); and (3) subsequent cooling relatively slowly, usually in air, through the Ms–Mf range to room temperature. This enables the martensitic transformation to occur at one and the same time over the entire section, since the difference in temperature between the exterior and interior of the part during air cooling is small.74 The term martempering is misleading since the martensite resulting from such treatment is not a tempered martensite in any sense of the word tempering. Since there is less risk of differential martensitic transformations within the section, distortion, cracking, and residual stresses are minimized. However, an increase in volume and loss of ductility result, accompanying the transformation which set up a high degree of microstresses. The final structure and properties obtained by this method are similar to those obtained from conventional quenchhardening, but the proportion of retained austenite is usually larger after martempering. Straightening or forming can be easily performed after removal from the martempering bath while the part is still hot.114 After the martempering operation, tempering of the component must be carried out in the same manner as in the conventional quench-hardening to obtain the desired hardness level. The time versus temperature relationship of martempering and tempering is illustrated in Fig. 13.42b.17 Usually, the quenching temperature maintained in this process is 175 to 230°C (350 to 450°F) when hot-quench oils are used. Molten nitrate/nitrite salts (with water addition and agitation) are effective and used in the range of 160 to 400°C (320 to 750°F) which offers more metallurgical and operational advantages, due to their higher heat-transfer coefficients.6 When parts are immersed into the quenching bath, the oil or salt circulation around them must be uniform, and the bath temperature should be accurately stabilized.115 For rapid cooling, the quenchant must have good cooling capacity and must be properly agitated, and a sludge-free bath must be maintained. Selection between salts and oil depends on several factors such as operating temperature, composition, and cooling power. Most installations for martempering are equipped with batch-type furnaces utilizing manual operation for the transfer of steel parts from the austenitizing furnace to the molten salt quench. However, mechanized plants with nitrate/nitrite salt bath furnace are now employed where the parts are normally quenched in the temperature range of 250 to 300°C (482 to 572°F). At this temperature, austenite may remain for a maximum period prior to its transformation to bainite. Martempering may also be feasible in vacuum, which reduces the thermal gradient, like the modified martempering process. The part held under vacuum at the austenitizing temperature is gas-quenched in the usual manner, but cooling is interrupted above the Ms to achieve temperature equalization in the part, this being followed again by gas-quenching to room temperature. Advantages 1. Martempering is applicable to a greater range of steel grades than the normal hardening process. 2. Low-temperature quenchant (particularly oil) often allows the utilization of simpler and less expensive quenching equipment. 3. It greatly reduces the problems of pollution and fire hazard when fluidized beds or nitrate/nitrite salt bath (with recovery of the salts from wash water) is used.

13.74

CHAPTER THIRTEEN

4. It eliminates the need for quenching fixtures, thereby reducing the cost of tooling and handling. 5. Uniform and reproducible results are secured by both the normal and modified martempering process. However, more severe distortion of sensitive parts is likely with the modified process, which requires grinding or other finishing allowances.17 The advantage over the austempering process is that martempering can be used to harden those steels that cannot be hardened by austempering because of the very long time involved in complete transformation to bainite. A harder transformation product (martensite) is produced instead of bainite. The advantages over through-hardening steels for carburized parts (especially splined shafts, cams, and gears) are realized because these parts do not require grinding or costly fabrication, due to fewer dimensional changes.114 Applications. This treatment is applied to oil-hardening and air-hardening (i.e., higher-alloy deep-hardening) steels. It is, however, not applicable to waterhardening tool steels or carburized carbon case-hardening steels unless these are of thin sections (to make them capable of hardening in oil), because the cooling rate in liquid bath is inadequate to prevent pearlitic transformation at higher temperatures of 500 to 600°C (932 to 1112°F). It is especially suitable for the treatment of irregular-shaped parts made from medium- or high-alloy steels. Steel castings are martempered in molten salts or hot oil baths at temperatures between 150 and 300°C (300 and 572°F).116,117 Typical applications of martempering steel parts in salt and oil baths are listed in Tables 13.9 and 13.10, respectively, which describe steel parts normally treated and provide details of martempering methods and hardness requirements.114 Modified Martempering. Figure 13.42c17 shows the modified form of martempering in that the temperature of the quenching bath is held just below the Ms temperature, usually in the 150 to 175°C (300 to 350°F) range. The lower-temperature bath increases quench severity. Hence, a section thickness that is larger than that for normal martempering can be hardened to the required hardness level. This process can be applied to a wide range of steel grades. Carbomartempering. Carbomartempering involves both carburizing or carbonitriding and martempering processes. It exhibits minimum distortion. If parts are carburized in a gas-fired furnace, they can be quenched directly into a salt bath. If they are carburized in a salt bath, however, they must pass through a neutral salt bath to prevent any violent reactions. The carbomartempering temperature is a compromise between the Ms temperature of the case and that of the core, and is selected not to jeopardize case hardness. It is usually in the 175 to 260°C (350 to 500°F) temperature range, and the holding time is only a few minutes.67

13.8 MARTEMPERING OF GRAY CAST IRON Gray iron castings are quenched from above the A1 temperature in a salt, oil, lead bath at temperatures slightly above the Ms [200 to 260°C (400 to 500°F) for unalloyed irons] until the entire section has reached the bath temperature, and then

TABLE 13.9 Typical Applications of Martempering Steel Parts in Salt114 Martempering conditions

Maximum section thickness Part

13.75

Compliant tube Thrust washer Chain link Cotton-picker spindled Accessory driveshaft Clutch-adjustment nut Seal ring Spur pinion Internal gear Dual gear f Drive coupling Spline shaft Arbor sleeve Screw-machine spindle Driving barrel Bearing raceh Hog knife Landing-gear spring Internal gear Spur pinion geark Screw-machine sprocket

Weight

Temperature of salt

Minimum time in salt, min

Required hardness, HRC

Grade

mm

in.

kg

lb

°C

°F

4130 8740 1045 Type 410

0.8 5.1 5.6 6.4

0.03 0.20 0.22 0.25

0.11 0.05 0.11 0.05

0.25 0.1 0.25 0.12

160a 230 205c 315

320a 450 400c 600c

5 1 1 11/2

50b 52 mmb 45–50b 44–48b

9310e 8740 52100 3312e 4350 4815e 4340 8720e 1117Le 8620e

6.4 7.6 7.6 7.6 8.9 9.4 10.2 10.2 10.2 10.2

0.25 0.30 0.30 0.30 0.35 0.37 0.40 0.40 0.40 0.40

0.45 0.14 0.18 0.23 0.36 2.13 0.27 0.50 0.59 6.35

1.0 0.3 0.4 0.5 0.8 4.7 0.6 1.1 1.3 14.0

190 230 190 175 245 260 230 190 205 205

375 450 375 350 475 500 450 375 400 400

21/2 2 10 11/2 2 2 21/2 21/2 3 3

90 (15N scale) 52 min.b 65b 90 (15N scale) 54 min.b 62–63b 52 min.b 90 (15N scale) ... ...

4350 52100 9260 6150 1117Le 4047 8620e

12.7 12.7 15.2 19.1 25.4 25.4 38.1

0.50 0.50 0.60 0.75 1.00 1.00 1.50

0.45 13.2 8.16 14.7 1.36 16.4 9.07

1.0 29.2 18.0 32.5 3.0 36.2 20.0

245 220 175j 260 205 230c 205

475 425 350j 500 400 450c 400

3 21/2 15 23/4 3 3 3

48–52g 63–64b 62b 56–57b ... 50–52m ...

Notes: OD, outside diameter; ID, inside diameter. a Salt contained 11/2% water. b As-quenched. c Salt contained water. d 6.4-mm (1/4-in.) diameter by 203 mm (8 in.) long. e Carburized. f 124-mm (47/8-in.) OD by 32-mm (11/4-in.) ID by 102 mm (4 in.). g Final. h 224-mm (813/16-in.) ID by 251-mm (97/8-in.) OD. j Salt contained 1% water. k 19-mm (3/4-in.) OD by 92-mm (35/8-in). ID by 140 mm (51/2 in.). m Asquenched hardness of teeth.

TABLE 13.10 Typical Applications of Martempering Steel Parts in Oil114 Maximum section thickness Part

Grade

Sleeve Spacer plate Bushing Shifter rail

13.76

Spur gear Helical gear Herringbone gear Shifter rail Spiral bevel gear Helical pinion Spur gear Splined shaft Spur gear Splined shaft Spur gear † ‡

mm

52100 3.2 1065 3.2 1117 4.8 1117 6.4 1018 9.5 1018 9.5 8620 12.7 4620H 19.1 4820 19.1

in.

Outside diameter mm

0.125 0.125 0.1875 0.25 0.375 0.375 0.5 0.75 0.75

... ... 51.0 76.3 ... ... 320.6 331.5 283.2

in.

... 0.1 ... 0.1 2.009 0.2 3.0034 0.6 ... 1.0 ... 1.4 12.620 12.7 13.050 16.9 11.150 16.3

0.001 in.

Temperature of martempering oil,† °C

Surface hardness, HRC

... ... 1015–1220 1015–1220 255–455 355–610 1145–1525 760–1015 1145–1525

... ... 40–48 40–48 10–18 14–24 45–60 30–40 45–60

790 790 910 910 845 845 845 845 845

165 165 190 190 165 165 150 150 150

58–59 56–57 58–62 55–60 55–60 55–60 55–60 58–63 55–61

mm

/4 /4 1 /2 11/4 21/8 31/8 28 37.2 36

... ... 910 910 845‡ 845‡ 925 925 925

Weight kg

Quenching temperature, °C

Carburizing temperature, °C

lb 1 1

Depth of case

1141 4620

25.4 25.4

1.0 1.0

25.4 210.6

1.0 8.29

0.9 5.1

17/8 11.25

885‡ 925

455–660 1015–1270

18–26 40–50

885 845

165 150

45–50 55 min.

8617H 8625 4817H 8625 8625 8625 8625 8620 8625 8625

25.4 31.8 34.0 38.1 39.7 44.4 44.4 50.8 65.0 84.7

1.0 1.250 1.340 1.500 1.564 1.750 1.750 2.000 2.559 3.3343

35.8 83.8 186.7 165.1 39.7 108.0 44.4 50.8 65.0 245.5

1.409 3.300 7.350 6.500 1.564 4.250 1.750 2.000 2.559 9.667

0.4 4.3 8.6 2.5 2.7 3.5 2.0 5.1 6.8 11.9

0.9 93/8 19 51/2 57/8 73/4 41/2 111/4 15 261/4

925 925 925 925 925 925 925 925 925 925

510–710 1525–1725 1400–1780 1525–1725 1400–1980 1525–1725 1525–1725 1525–1725 1525–1725 1525–1725

20–28 60–68 55–70 60–68 70–78 60–68 60–68 60–68 60–68 60–68

845 925 845 925 925 925 925 925 925 925

150 190 150 190 190 165 190 165 165 190

58–63 58–62 58–63 58–62 58–62 58–62 58–62 58–62 58–62 58–62

Minimum time in oil, 5 min. Carbonitriding temperature.

HARDENING AND HARDENABILITY

13.77

they are air-cooled to room temperature. A typical application includes cylinder liners (sleeves) for diesel and heavy-duty gasoline engines.109

13.9 AUSTEMPERING OF STEEL Austempering, a hardening process, is the isothermal transformation of austenite into bainite. This process was first introduced by Bain and Davenport in the 1930s118 and consists essentially of 1. Heating the steel part to the hardening or austenitizing temperature, usually between 790 and 915°C (1450 and 1675°F) in a molten nitrate-nitrite salt bath or in a controlled-atmosphere furnace and holding there for a proper length of time 2. Quenching in a molten lead or salt bath, vigorously agitated and maintained at an appropriate isothermal temperature, usually in the range of 260 to 400°C (500 to 752°F), and holding at bath temperature for a sufficient time to transform to fully bainitic structure 3. Cooling down to room temperature, usually in still air The procedure is shown schematically in Fig. 13.45. The actual temperature of isothermal transformation is dependent upon the hardness and properties desired and the transformation characteristics of the steel being processed, as indicated in the TTT diagrams. One example of the influence of the low temperature of a salt bath is that it produces a high hardness of the resulting microstructure. The time in the quenching bath, usually varying from 20 min to several hours, depends entirely on the transformation temperature used, the steel grade, and the degree of hardness desired. Bath agitation, accomplished by mechanical stirring, pumping, and air agitation, can be a vital factor in austempering due to its increased quenching speed. The quenching severity of nitrate-nitrite salt can also be increased by careful and

FIGURE 13.45 TTT diagram for 1080 steel showing a difference between conventional and modified austempering.17 When applied to wire, the modification shown is called patenting. (Reprinted by permission of ASM International, Materials Park, Ohio.)

CHAPTER THIRTEEN

13.78 TABLE 13.11 Quenches119

Quench Severity Comparison for Salt

Numbers given are estimated Grossman H–values. At temperature: Agitation

180°C (360°F)

370°C (700°F)

0.15–0.20 0.25–0.35 0.40–0.50 0.50–0.60 0.90–1.30†

0.15 0.20–0.25 0.30–0.40 0.50–0.6† Not possible

Still and dry Agitated and dry Agitated with 0.5% water Agitated with 2% water Agitated with 10% water †

Requires special enclosed quenching apparatus

TABLE 13.12 Effect of Transformation Temperature on Hardness and Strength of Austempered 0.8% C Steel120

Rockwell C hardness 40 45 50 55 58

Approximate tensile strength, kpsi

Approximate transformation temperature, °C (°F)

180 210 240 280 300+

400 (750) 343 (650) 316 (600) 271 (520) 260 (400)

Reprinted by permission of Fairchild Publications, New York.

periodic water addition, which involves agitation of salt to disperse water uniformly (Table 13.11).119 The rate of the last cooling stage (item 3) is not significant. Tempering after austempering may or may not be necessary because much depends on the properties desired in service conditions. However, if it is carried out, further improvement in ductility occurs by removal of carbon from any retained austenite. Table 13.12 shows the transformation temperature of 0.8%C steel austempered to various hardness or strength levels.120 Note that where high hardness and ductility are the prime considerations, it is best to austenitize at the highest temperature possible to dissolve all the carbides in the solution. Where both wear resistance and minimum distortion are required, such as in 1095 steel cutting edges of electric razors, a just-sufficient amount of carbides should be in solution during austenitizing, to meet the hardness by a bainitic reaction but to leave some undissolved carbides in it for wear resistance. Where minimum distortion is desired, austenitization is carried out at the lowest temperature possible and quenching into a bath at the highest temperature possible to meet the required hardness level.120 The mechanical properties of sway bars made of 1090 steel and hardened by austempering and quenching and tempering process are listed in Table 13.13. It is important to note that the austempered parts have desired mechanical properties in that they have a 100% bainitic structure. Table 13.14 lists section sizes of austempered parts made of various steels in which the formation of some pearlite or martensite is allowed in the microstructure.

HARDENING AND HARDENABILITY

13.79

TABLE 13.13 Comparison of Typical Mechanical Properties of Austempered and of OilQuenched and Tempered Sway Bars of AISI 1090 Steel119 Propertya

Austempered at 400°C (750°F)b

Quenched and temperedc

1415 (205) 1020 (148) 11.5 30 415 105,000e

1380 (200) 895 (130) 6.0 10.2 388 58,600f

Tensile strength, MPa (ksi) Yield strength, MPa (ksi) Elongation, % Reduction of area, % Hardness, HB Fatigue cyclesd a b c d e f

Average values. Six tests. Two tests. Fatigue specimens 21 mm (0.812 in.) in diameter. Seven tests; range, 69,050 to 137,000. Eight tests; range, 43,120 to 95,220.

TABLE 13.14 Hardness of Various Steels and Section Thicknesses of Austempered Parts119

Section size Steel 1050 1065 1066 1084 1086 1090 1090e 1095 1350 4063 4150 4365 5140 5160e 8750 50100 a b c d e f g

mm b

3 5c 7c 6c 13c 5c 20c 4c 16c 16c 13c 25c 3b 26c 3b 8c

in. b

0.125 0.187c 0.281c 0.218c 0.516c 0.187c 0.820c 0.148c 0.625c 0.625c 0.500c 1.000c 0.125b 1.035c 0.125b 0.312c

Salt temperature

Ms temperaturea

°C

°F

°C

°F

Hardness, HRC

345

655

d

d

d

d

d

d

d

d

320 275 260 200 215 ... ... 210g 235 245 285 210 330 255 285 ...

610 525 500 395 420 ... ... 410g 450 475 545 410 630 490 545 ...

41–47 53–56 53–56 55–58 55–58 57–60 44.5 (avg) 57–60 53–56 53–56 52 max. 54 max. 43–48 46.7 (avg) 47–48 57–60

d

d

315f

600f

d

d

d

d

d

d

d

d

d

d

345 315f 315

655 600f 600

d

d

Calculated. Sheet thickness. Diameter of section. Salt temperature adjusted to give maximum hardness and 100% bainite. Modified austempering; microstructure contained pearlite as well as bainite. Salt with water additions. Experimental value.

Advantages.

The advantages include the following:

1. There is reduced distortion and cracking. The danger of retained austenite is also reduced. Hence this is a process of considerable importance where it is desired to harden intricate-shaped components successfully and safely.

13.80

CHAPTER THIRTEEN

FIGURE 13.46 Comparison of impact toughness of a carbon steel after conventional hardening and tempering and after austempering, as a function of hardness.124 (Courtesy of Marcel Dekker, Inc., New York.)

2. Ductility, toughness, and strength are increased at a given hardness value in the hardness range of Rc 47 to Rc 55. Alternatively, the material can be given an austempering treatment to achieve the highest hardness and tensile strength but a toughness similar to that of tempered martensite.121 Figure 13.46 shows a comparison of the impact toughness produced by conventional quenching and tempering and austempering for 0.74% carbon steel and SAE 6150 steel. 3. Uniformity and consistency of properties throughout the section component are obtained. 4. There is a less heat treatment breakage. 5. These parts have high fatigue resistance; they often have a life 3 to 4 times longer than those of comparable quenched and tempered parts. Figure 13.47 shows the fatigue diagram of DIN 30SiMnCr4 steel after conventional hardening and tempering and after austempering. Note that the increase in fatigue strength is significant for notched specimens. 6. An austempered carbon steel wire has superior bendability to that hardened and tempered to Rc 50. 7. There are claims of less hydrogen embrittlement. 8. Much cleaner quench and less surface contamination are found. 9. Austempering in a protective atmosphere furnace with integral salt quenches has been designed to prevent air contact throughout the processing, which produces a clean and shiny attractive surface.122 In this situation, austempered parts can, therefore, be immediately painted, phosphated, or zinc-plated. However, oil-quenched and tempered parts need pickling treatment prior to these finishing operations.123 10. It allows the use of inexpensive, lower-grade materials without sacrificing mechanical properties, thereby reducing production costs. For example, high-

HARDENING AND HARDENABILITY

13.81

FIGURE 13.47 Bending fatigue strength of DIN 30SiMnCr4 steel after conventional hardening and tempering and after austempering. [Source: F. W. Eysell, Die Zwischenstufenvergutung und ihre betriebliche Anwendung, Z. TZ Prakt, Metallbearb., 66:94–99, 1972 (in German).]

tensile-strength bolts made of quenched and tempered 8640 steel can be replaced with austempered lower-grade 10B39 steel, resulting in a 40% saving on material costs alone while maintaining the 1724-MPa (250-ksi) minimum tensile strength and 17% elongation.122 11. It is a single-operation process and requires no tempering. Limitations 1. Since tensile strength and impact values are lower than those of the hardened and tempered product, this process is not used in the case of low- and mediumcarbon steels, except when risk of cracking and distortion is to be avoided. It appears to be more beneficial in steels with carbon content varying between 0.5 and 1.2%. 2. It is restricted to thin sections for plain carbon and low-alloy steels (Table 13.14), i.e., limited to those cross sections of steel which can be cooled from austenitizing temperature to austempering bath temperature with sufficient rapidity to prevent fine pearlite formation. It is common practice to austemper plain carbon steels up to a maximum diameter of 5 mm (0.2 in.) and certain low-alloy structural steels [e.g., spring steels up to 10-mm (0.4-in.) section thickness]; this is evidently dependent on the dissolved alloy content and grain size. Lower-carbon, boronbearing steels, however, can be successfully austempered in heavier sections. In some alloy steels, section thickness up to about 25 mm (1 in.) can be satisfactorily austempered to fully bainitic structures.119 The range of suitable sizes is wider for medium-alloy steels. With high-alloy steels, heavier sections (up to 25-mm, or 1-in., cross section) can be austempered, but an inordinately long time (sometimes several hours, days, or weeks) would be necessary to produce completely the bainitic structure. The austempering of such steel would be wholly uneconomical and

CHAPTER THIRTEEN

13.82

impractical. Hence this process is limited to small tools where exceptional toughness together with reasonable hardness is desired. 3. This is a slower and somewhat more expensive process than conventional quenching and tempering and requires closer supervision to maintain a salt bath at a suitable temperature in the range of 260 to 400°C. However, when fully automated, it is capable of being cost-competitive with “quench and temper” systems as a result of labor savings of 50%. Since most austempering is accomplished in molten nitrate-nitrite salt, the salt bath used for austenitizing must be compatible with the austempering salt. Hence, a chloride salt bath for austenitizing should have the following composition and characteristics: 40 to 55% NaCl + 45 to 55% KCl, melting temperature 650 to 675°C (1200 to 1250°F), working temperature range of 705 to 900°C (1300 to 1650°F).119 Recently a “supersaturation” method has been developed for salt bath quench on a high-production batch furnace. In this process, agitation—when combined with water addition up to 12% of the quench at 182°C (360°F)—raises the quench severity of the molten salt to a level equal to that of a brine solution. This allows austempering of parts up to 6 in. thick.122 Plain carbon steels containing (1) 0.5 to 1% carbon and at least 0.6% Mn, (2) more than 0.9% C and a little less than 0.6% Mn, and (3) less than 0.5% C and 1 to 1.65% Mn as well as low-alloy steels, such as 5140, 4140, 6145, and 9440, are well suited to austempering due to their high hardenability.123 It is not feasible to austemper AISI 1034 steel due to the extremely fast pearlite reaction at 540 to 595°C (1004 to 1103°F). The AISI 9261 steel is unsuitable to austempering treatment due to very slow bainite formation at 260 to 400°C (550 to 752°F).124 Comparison of Processes and Products. Table 13.15 shows the scheme of operation of austempering and of the oil-hardening and tempering process.125 Although no tempering operation is involved in the austempering process, some gas input is necessary to maintain the quench bath at a temperature in the region of 260 to 400°C. However, it is considered good practice to temper austempered parts at TABLE 13.15 Austempering versus Oil-Hardening and Tempering: Cost Compared125 Harden and temper Process

Austempering

Cost input

Process

Cost input

1. Heat components to hardening temperature

Fuel to raise temperature

Heat components to hardening temperature

Fuel to raise temperature

2. Quench (to room temperature)

Cost of oil + dragout + circulation + cooling

Quench (to 320°C)

Cost of salt + dragout + circulation + fuel to maintain temperature

3. Wash

Water heating

Wash

Water heating

4. Tempering

Fuel to raise temperature

5. Cleaning (option)

Shot blast/oil dip

HARDENING AND HARDENABILITY

13.83

TABLE 13.16 Comparison of Properties of Retaining Rings for 1080 Steel126 Property Tensile strength 0.2% yield stress Percent elongation Impact strength Fatigue life, bend test

Oil-quenching and tempering

Austempering

254 ksi 237 ksi 3.0 0.4 ft◊lb 60,000 cycles

260 ksi 246 ksi 4.6 1.9 ft◊lb 98,000 cycles

the quench temperature to transform any retained austenite present. Thus some saving in fuel can be made here. It should be pointed out that workpieces are going into an austempering bath at a higher temperature than the bath itself, whereas in tempering the work is heated from cold. In a tempering treatment, the heat remains on full for 1 hr per 1.5-hr cycle; in austempering, however, the heat remains on full for 0.25 hr per 1.5-hr cycle. In the case of gaseous fuel, a savings of 25% on gas consumed is expected, and savings of as much as 75% can be made on the cost of the quenching medium if austempering operations are practiced instead of oil-quenching and tempering. In a quenching operation, the costs involved are the cost of the quenching medium and dragout since the circulation cost is common in both. Another real savings in cost can be made in the salt bath austempering operation in the area of dragout.125 Table 13.16 shows a comparison of mechanical properties of retaining rings for 1080 steel with the composition range of 0.75 to 0.85% C, 0.65 to 0.85% Mn, 0.03% S and P maximum, and 0.35% Si maximum; with ASTM grain size Nos. 5 to 8.126 Application. Table 13.17 provides processing data for numerous parts made of various plain carbon, alloy, and carburized steels. These data are for representative austempering practices used by more than a dozen manufacturing plants.119 Modified Austempering. Modified austempering, also called patenting, is the well-accepted practice in the wire industry to produce spring wire, rope wire, piano wire, and so forth. The main purpose of this process is to convert the coarse, nonuniform ferrite/pearlite or carbide/pearlite structure of hot-rolled wire rod into a uniform structure of varying amounts of fine pearlite and bainite, to facilitate drawing of high tensile strengths accompanying large reduction of area without fracture. The production process consists in heating the wire or rod to a temperature above the Ac3 point (850 to 1100°C) to enable it to be completely austenitized, followed by continuous quenching of the wire by passing it in air or lead or salt bath maintained at a temperature between 500 and 550°C (932 and 1022°F) according to the composition; holding in this bath for periods varying from 10 s (for small wire) to 90 s (for rod); and then cooling in air to ambient temperature. The process is described by the curve superimposed on the typical TTT diagram (Fig. 13.45). The modified process can be applied to plain carbon steel and low-alloy steels having section thickness larger than normally considered practicable for austempering.127 During drawing, C atoms must abandon cementite to find favorable sites at dislocation. This dissolved the C pins dislocations, which act as a barrier to further

13.84

TABLE 13.17 Typical Production Applications of Austempering119 Parts listed in order of increasing section thickness.

Part

Steel

mm

in.

Plain carbon steel parts Clevis Follower arm Spring Plate Cam lever Plate Type bar Tabulator stop Lever Chain link Shoe-last link Shoe-toe cap Lawn mower blade Lever Fastener Stabilizer bar Boron steel bolt

1050 1050 1080 1060 1065 1050 1065 1065 1050 1050 1065 1070 1065 1075 1060 1090 10B20

0.75 0.75 0.79 0.81 1.0 1.0 1.0 1.22 1.25 1.5 1.5 1.5 3.18 3.18 6.35 19 6.35

0.030 0.030 0.031 0.032 0.040 0.040 0.040 0.048 0.050 0.060 0.060 0.060 0.125 0.125 0.250 0.750 0.250

Weight or parts per unit weight

Salt temperature

kg or kg-1

lb or lb-1

°C

°F

Immersion time, min

Hardness HRC

770/kg 412/kg 220/kg 88/kg 62/kg 0.5 kg 141/kg 440/kg ... 573/kg 86/kg 18/kg 1.5 kg 24/kg 110/kg 22 kg 100/kg

350/lb 187/lb 100/lb 40/lb 28/lb 1 /4 lb 64/lb 200/lb ... 260/lb 39/lb 8/lb 2 /3 lb 11/lb 50/lb 10 lb 45/lb

360 355 330 330 370 360 370 360 345 345 290 315 315 385 310 370 420

680 675 625 630 700 675 700 680 650 650 550 600 600 725 590 700 790

15 15 15 6 15 15 15 15 15 15 30 60 15 5 25 6–9 5

42 42 48 45–50 42 42 42 45 45–50 45 52 50 50 30–35 50 40–45 38–43

CHAPTER THIRTEEN

Maximum section thickness

Alloy steel parts Socket wrench Chain link Pin Cylinder liner Anvil Shovel blade Pin Shaft Gear

... 0.063 0.063 0.100 0.125 0.125 0.250 0.375 0.500

0.3 kg 110/kg 5500/kg 15 kg 1.65 kg ... 100/kg 0.5 kg 4.4 kg

/8 lb 50/lb 2500/lb 7 lb 3 /4 lb ... 45/lb 1 /4 lb 2 lb

1010 1117 8620

3.96 6.35 11.13

0.156 0.250 0.438

33 kg 66/kg 132/kg

15 lb 30/lb 60/lb

‡ §

Contains 0.65 to 0.75% C. Leaded grade. Case hardness.

365 290 325 260 370 370 370 385 305

690 550 620 500 700 700 700 725 580

15 25 45 14 30 15 45 15 30

45 53 48 40 37 45 40 35–40 45

385 385 290–315

725 725 550–600

5 5 30

30–35§ 30–35§ 50§

HARDENING AND HARDENABILITY

... 1.60 1.60 2.54 3.18 3.18 6.35 9.53 12.7

Carburized steel parts Lever Shaft Block †

1

6150 Cr-Ni-V† 3140 4140 8640 4068 3140 4140‡ 6150

13.85

CHAPTER THIRTEEN

13.86

dislocation motion. A high rate of dislocation pinning causes a very high rate of work-hardening.128 Carboaustempering. Carboaustempering involves both carburizing or carbonitriding and austempering processes. Austempering of low-carbon-content steel after carburizing produces a high-carbon bainitic case and either a bainitic or martensitic core, depending on quench severity and steel composition. Carboaustempered parts possess increased fatigue strength and wear resistance, and they are dimensionally and functionally superior to carburized and conventionally quenched components.67

13.10 AUSTEMPERING OF DUCTILE IRON Austempered ductile iron (ADI) is currently the subject of extensive study worldwide because of the development of the best combinations of low cost, design flexibility, good machinability, high strength-to-weight ratio, and good toughness, fatigue resistance, and wear resistance.129 Table 13.18 lists the five standard ADI grades with minimum property requirements, according to the ASTM-A897M standard. Austempering of ductile iron consists of austenitizing components at a temperature usually between 815 and 955°C (1500 and 1750°F) for nearly 1.5 hr in salt bath, controlled endothermic atmosphere, or fluidized bed, followed by fast-quenching into a lowtemperature quenchant such as molten salt bath, hot oil, or fluidized bed maintained isothermally in the bainitic transformation (or austempering temperature) range [205 to 450°C (400 to 842°F)], holding them isothermally for a specified time between 0.5 and 4 hr to complete the transformation; subsequently they are aircooled.130–134 Figure 13.48a and b shows a schematic diagram of the austempering heat treatment cycle (A through H) and the typical austempering cycles for different grades of ADI, respectively.129,134a Note that where limited facilities (with respect to the above-mentioned furnaces) are available, proprietary stop-off compounds can be used when heat-treating in air because these compounds effectively prevent surface degradation and can be easily removed after isothermal treatment. Maximum thicknesses for complete austempering of unalloyed ductile irons are 15 mm at a temperature of 450°C and 30 mm at a temperature of 250°C. As section thickness increases beyond these values, the amount of pearlitic transformation increases.131 When quench severity is improved by the addition of 0.2 to 2 wt% water in a salt bath (called a saturated salt bath), the section thickness can be increased to ⬃1 in. in unalloyed irons and up to 2.5 in. (64 mm) thick in alloyed ductile irons. Subsequently it was found that the water content of the salt bath could be increased TABLE 13.18 The Five Standard ADI Grades (ASTM A897M-1990)

Grade 1 2 1 1 1

Tensile strength, MPa 850 1050 1200 1400 1600

Yield strength, Elongation, MPa % 550 700 850 1100 1300

10 7 4 1 N/A

Impact energy, Typical hardness, J BHN 100 80 60 35 N/A

269–321 302–363 341–444 388–477 444–555

HARDENING AND HARDENABILITY

FIGURE 13.48 (a) Schematic diagram of the austempering process for a cast iron.134a (b) Typical austempering cycles for different grades of ADI.129 [(a) Courtesy of Wolfson Heat Treatment Center, England. (b) Courtesy of QIT, Fer et Titane, Inc., Chicago, Illinois.]

13.87

13.88

CHAPTER THIRTEEN

FIGURE 13.49 Percentage transformation and microstructural changes during stages I and II of austempering and processing window in unalloyed iron for higher austempering temperatures. The amount of high-C, retained austenite during the stage I reaction reaches a plateau representing the end of time t1. The amount of retained g starts to fall at time t2 when the high-C austenite begins to undergo the stage II reaction. This behavior exposes a welldefined processing window t2 - t1, during which the metastable structure of bainitic ferrite and high-C austenite is relatively stable and the austempered iron exhibits optimum properties.136

further from 2 to 12% with the application of high rates of recirculation of the bath. The use of this bath, called supersaturated salt bath, increases the quench severity to the extent that 2-in. section thickness can be completely austempered in unalloyed irons. Note that water addition is never made in an open quench tank.134 The austempering reaction in ductile iron, over the austempering range of 205 to 450°C, occurs in two stages, as shown in Figs. 13.48a and 13.49. The first-stage transformation starts by the nucleation of bainitic ferrite (at high temperature) at interphase and grain boundaries, and its growth into austenite. This is associated with the carbon rejection from growing ferrite platelets into the surrounding austenite.135,136 [The high Si content of ductile iron retards the iron carbide precipitation, which results in the carbon enrichment (up to 2%) of austenite, particularly between the growing ferrite plates, and the Ms temperature below -120°C (-180°F).] The carbon enrichment of the reacted austenite occurs (in the time period of EF) up to 1.2 to 1.6%. If the austempering period is extended further (FG), acicular ferrite continues together with additional enrichment of austenite up to 1.8 to 2.2%. This austenite is thermally and mechanically stable and is called reacted stable austenite, i.e., the stabilization of (retained) austenite after cooling to ambient temperature. This form of austenite is desirable in the ausferrite structures of grades 1 and 2.134a (Unreacted austenite is thermally unstable† and transforms to martensite on quenching to room temperature.) The structure of the transformation product corresponding to the first stage, nominally called upper bainite, consists of relatively coarse bainitic ferrite plates plus retained austenite (called carbon-enriched stable austenite) in the matrix (Fig. 13.50a and b). The reaction can be described as g Æ abainite (acicular) + ghigh carbon. This product is also known as upper ausferrite. The ADI transformed at a high austempering temperature (say, 400°C, or 750°F) exhibits high ductility, good fatigue and impact strength, and a tensile strength of 125 to 150 ksi (860 to 1030 MPa).129 In stage II with extended austempering period beyond about 2 hr (JKL), highcarbon (reacted stable) austenite further decomposes into the thermodynamically more stable fine acicular structure of bainitic ferrite and carbide and nominally † Thermally unstable austenite can be distinguished in a microstructure using a double etchant of nital and sodium metabisulfite. However, it is more difficult to recognize between reacted metastable and reacted stable austenite. It can be differentiated only by a special technique such as heat tinting.134a

HARDENING AND HARDENABILITY

13.89

FIGURE 13.50 Light photomicrographs of austempered ductile iron at 370°C (700°F) (a, b) and at 315°C (600°F) (c, d). Both are austenitized at 900°C (1650°F) but for different times, namely, 125 and 90 min, respectively, corresponding to the higher and lower austempering temperatures. Microconstituents observed are austenite (g), bainitic ferrite (abainite), and martensite (M). (Courtesy of R. Gundlach and J. Janowak.132)

called lower bainite or lower ausferrite. The reaction can be represented by ghigh carbon Æ abainite (acicular) + carbide. (See Fig. 13.50.136) (A transitional e-carbide mostly occurs in the initial stages followed by an incoherent Fe3C carbide. Usually this transformation is deleterious to ADI properties; however, the optimum properties can be achieved if the iron is cooled from the austempering temperature at a time between the two stages.134a) The ADI structure, transformed at around 260°C (500°F), has therefore a much finer structure that exhibits high hardness and excellent wear resistance equivalent to the case-hardened steel, a tensile strength >230 ksi (1600 MPa), and reduced ductility and impact strength.129,137 The structure of this transformation product, nominally called lower bainite, contains a smaller amount (up to 15%) of stabilized austenite (Fig. 13.50c and d). In some cases (as seen in the above microstructures), complete stabilization of austenite is not found to occur, and the austempered structures invariably contain up to 12% martensite. Untransformed low-carbon austenite from the stage I reaction, which is thermally unstable and readily transforms into martensite in the austempered structure, and the brittle carbide from the stage II reaction have deleterious effects on mechanical properties, especially ductility and toughness and fatigue strength.136,137a The important mechanical properties of ADI are realized when the quantity of these phases in the microstructure is small or inadequate to be deleterious to the mechanical properties.138 Increasing the carbon content of austenite favors 13.89

13.90

CHAPTER THIRTEEN

the stability of retained austenite, thereby eventually diminishing the martensite formation. A marked reduction in ductility and toughness results when incomplete transformation occurs, i.e., when parts are quenched too early from the austempering temperature or when the parts are held for prolonged periods because the former produces varying amounts of martensite and retained (stabilized) austenite after cooling to ambient temperature and the latter induces undesirable transformation/decomposition of the stabilized austenite to form additional ferrite and carbide.139 It is clear that the two structures obtained from austempering ductile cast iron and steel and the accompanying properties are quite different; therefore it is appropriate to term them ausferrite and bainite, respectively, because ADI consists of two phase mixtures comprising acicular ferrite and high-carbon austenite rather than ferrite and carbide, in steel. Moreover, the ausferritic reaction in ductile iron is remarkably slower than that of bainitic reaction in steel (Fig. 13.51).132 The presence of Si in ductile iron suppresses the formation of carbides. Austempering reaction in ductile iron is more sensitive to variations in the metal chemistries, times, and temperatures. It is the morphology of this austenite-ferrite structure that offers ADI its remarkable properties (Fig. 13.52).140 Based on the microstructure, transformation characteristics, and properties of the transformation product, ADI can be classified into the following grades, presently used commercially. Grades 4 and 5. Ductile irons transformed from high austenitizing temperature to low austempering temperatures (that is, 205 to 235°C) represent lower bainite (ausferrite). These irons, containing lower amounts of retained austenite and higher volume fraction of finer acicular ferrite, have high hardness (400 to 500 BHN or 45 to 50 Rc) and high tensile and fatigue strengths and wear resistance, but limited ductility and fracture toughness.129,134,137a Grades 1 and 2. Ductile irons transformed at higher austempering temperatures in the range of 330 to 450°C exhibit upper bainite (ausferrite) and contain large amounts of retained austenite (⬃20 to 50%) and ferrite. These irons possess hardness in the range of 260 to 300 BHN (29 to 35 Rc). They have excellent toughness, ductility, and fatigue strength, and good machinability. Typical grade 2 irons are used for applications requiring torsion stress, high impact loading, and high cycle fatigue life. Process Control. Process control includes the consistent production of highquality ductile irons and heat treatment process control. Austempering cannot be expected to improve the properties of poor-quality castings consistently. ADI castings should therefore be produced free from surface defects and segregation and should possess the microstructural features such as freedom from carbides and porosity, high nodule count (100 nodules/mm2 or higher), good nodularity (80% type I and II nodules), uniform distribution of nodules, minimum inclusion content, and controlled pearlite/ferrite ratio, if specified.129 Ductile irons require strict control of not only the austempering treatment parameters such as austenitizing temperature and time, austempering temperature and time, and transformation cooling rate, but also the materials-related parameters such as composition, hardenability, and dimensional control.132,139 When selecting the composition, the first consideration includes the limiting elements which adversely affect the casting quality through the production of nonspheroidal graphite, the improvement of shrinkage, and the formation of carbides

1000 (1832)

SAE 1080

Ductile lron

800 (1472)

13.91

Temperature, C (F)

A( + G)

T

A

Ac1 Ac1 F + C( + G) A + F + C( + G))

A

600 (1112)

A+F +C

A( + G)

F+C

400 (752)

200 (392)

Ms

Ms 0.1

50%

50% 1

10 (a)

102

103

104/0.1 Time, s

1

102

10

FIGURE 13.51 TTT diagrams for (a) SAE 1080 steel (0.79C-0.76Mn) and (b) ductile iron (3.3C-2.5Si-0.29Mn) showing the difference in bainitic reaction between the two. Both were austenitized at 900°C (1650°F).132 (Reprinted by permission of ASM International, Materials Park, Ohio.)

(b)

103

104

13.92

CHAPTER THIRTEEN

FIGURE 13.52 Yield and tensile strengths in ADI castings.140

and inclusions. The second consideration should be given to the control of carbon, silicon, and major alloying elements that control the hardenability of the iron and the properties of the transformed microstructure. When determining alloy requirements, both the section thickness and the severity of austempering quench must be given importance. For heavy section castings, selective alloying is needed to prevent pearlite/ferrite formation. For light section castings (up to ⬃3/8 in., or 10 mm), a very rapid quench is normally adequate to avoid the pearlite/ferrite formation in even an unalloyed iron.129 Advantages of ADI 1. They have excellent mechanical properties, such as high strength, toughness, and wear resistance and through-hardenability, which are superior to those of quenched and tempered structures.139 The new tougher material allows iron foundries to compete with heat-treated or high-alloy steels where greater reliability is in demand. In this case, the ADI is used as an upgrade for standard ductile iron parts and can also replace Ni hard in some applications.141 In the dry-sliding wear test mode, ADI wear resistance is about 4 times greater than that of pearlitic ductile iron (PDI) grade 100-70-03, more than 12 times that of leaded-tin bronze, and nearly 14 times that of aluminum bronze.141a In abrasive wear tests, ADI exhibits equivalent wear resistance to that of AISI 4340 steel, nearly twice less than that of hardened and tempered AISI 1050 carbon steel, and remarkably greater than that of white and alloyed cast irons.141a The higher hardness grades of ADI are used in wear applications while lower hardnesses are used in structural applications. When used as a substitute for forged/cast steel parts, the ADI parts exhibit fatigue strengths greater than those of forged steels and require an increased stiffness and larger fillet radii than do steel parts, to prevent stress concentration.141

HARDENING AND HARDENABILITY

13.93

2. No other single engineering materials available today show the range of properties illustrated by ADI (Fig. 13.52), because of its energy effectiveness, material conservation, better castability to near net-shape, reliability, and increasingly competitive world market. 3. ADI can be subjected to (a) work-hardening treatment of retained austenite under stress in the surface layers and (b) strain-induced transformation hardening of retained (enriched) austenite into martensite, producing a localized increase in volume and compressive stresses in the transformed material by surface layer deformation142 by using surface treatments such as shot-peening and rolling,129 which have been shown to produce inhibition of crack formation and an increase in surface hardness and, therefore, very high fatigue (bending) strength, endurance ratio, and rolling and sliding wear resistance. As a result of this hardening feature, care must be exercised in the sequence of operations on machined parts. The higher-hardness grades should be machined prior to heat treating whereas lower-hardness grades are best machined after heat treatment.141 4. ADI gears provide better noise damping, higher thermal conductivity, and far superior machinability compared to austempered forged steel gears.143 5. The excellent wear resistance of ADI (i.e., twofold improvement over steel at the same hardness level) is due to the presence of retained g in the microstructure, transformation of high-carbon g to martensite occurring in the surface layers during the wear tests, and the lubrication provided by the graphite content of the material.141a,144 However, this property certainly causes problems with final machining operation of the heat-treated product. 6. ADI parts specified by ASTM A897 outperform steel castings, forgings, and fabrications (weldments) in both structural and wear applications.141 They can replace wear-resistant steels and cast irons. 7. There have been significant cost and weight savings (10% lighter than steel), ease of machining, ready availability, and inexpensive raw materials.144 8. Austempered chilled ductile iron containing increased Ni and Mo contents is used as gears and pinions, crankshafts, driveshaft yokes, and related components to replace forged and case-hardened steels.145 9. ADI is 3 times stronger and weighs 2.6 times more than the best cast or forged aluminum. Since it has twice the stiffness, a properly designed ADI is a potential candidate to replace an aluminum part as a weight savings.137 10. Typically, an ADI part consumes 50% less energy than a steel casting and about 80% less energy than a steel forging.137 Applications. Automotive applications include camshafts, crankshafts (for highspeed diesel engines), transmission shafts, connecting rods and pistons (for diesel engines), timing gears, steering knuckles, CV joints, differential gears, differential housings, mounting brackets, chassis components, suspension components, spring hangers, spring stops, U-bolt stop plates, pintle hooks, pump rings, rocker arms, wheel hubs, axle structures, snowplow shoes, ring and pinion gears, bevel gear (with hypoid gearing) in rear-axle drives of light- and medium-duty trucks, and forklift truck parts.144 Railroad applications include car wheels, bogie and rail wheels, top caps, wear shoes, nipper and gauging hooks, shock absorbers, track plates and hardware, latches, covers, tie bars, engine parts, railway line components, suspension parts, and so forth.

13.94

CHAPTER THIRTEEN

Agricultural applications include plow and till points, eyebolts, wheel hubs, wear plates, chisels, steering shafts, tow hooks, plow shears and tips, slip clutch parts, fertilizer knives, soil turners, and so forth. Military applications include projectiles, armor, track shoes and plates, track guides, engine rotors, suspension and engine components, end connectors, struts, projectiles, and so forth. Construction and mining applications include digger teeth, grader blades, pavement breakers, carrier ring, guides and rollers, gears, housings, structural parts, slides, yokes, connectors, structural members, collars, drag conveyors, highway and pole line hardware, snowplow shoes, pitman arms and differential cases, mining drilling heads, track wheels, draw rolls, cam tracks, shift forks, and crusher components (in excavation).144 Industrial applications include gears (face, spur, bevel, and bull), brackets, grates, tool holders, jackhammer housings, paper mill components, parts for paper cutting industry, parts for printing machinery, water pump components, textile mill components, textile machinery-cam guides, steel mill rollers, steel rolling mill guides, rolls for pressing dies, bogie pins, cams, cover plates, shredder knives, high-volume repetitive punch for canning industry, parts for bottle-mold industry, mixer blades and worm wheels for food stuffs, crane support wheels, conveyor chain links, links for chain manufacturers, power handling parts, air compressor pump parts, shot-blast parts, gear wheels and gearbox parts for building equipment and stationary transmission units, differential spiders and spring seats, high-speed punching dies, gear wheels of stationary power units made of low-alloy (Cu-Mo) ductile iron weighing 30 kg to 2 tonnes and 300 to 2000 mm in diameter, planetary gear units as a substitute for case-hardened steel, joint sleeves for locomotives, cranes, and other hoisting equipment, on-site drill/milling components, segmented girth rings up to a diameter of 12 m, dolly wheels, driveshaft yokes, stub shafts, starter clutches, and others.122,137,144,146

13.11 QUENCH CRACKING Usually, quench cracking follows austenite grain boundaries; i.e., it is intergranular, but it does not appear to be related to the austenite grain size. Although a majority of the quench cracking is associated with the quenchants, notably their excessive quench severity, many of the cases can be produced by material or mechanical flaws. To avoid quench cracking of steel, the proper quenching media should be selected for the range of alloys and type of heat treatment to be accomplished. It is also essential to control the bath temperature and flow characteristics of the quenching media. It is thus recommended to accomplish metallurgical analysis to establish the true cause of the problems.147 (See also Sec. 17.4.) The potential sources contributing to quench cracking are given below.147 1. Nonuniform quenching is due to poor system design, inappropriate racking of parts to be quenched (which restricts the quenchant flow and uniform heat extraction), or incompatible quenchant contamination (water-in-oil or oil-in-water), air entrainment and excessive foaming, etc. All these factors have the ability to produce increased thermal gradients, leading to quench cracking. 2. Prior steel structure, e.g., cast, cold-formed, forged, extruded, etc., may increase the likelihood of cracking during the quench. Each as-formed structure needs a certain time and temperature cycle to condition the material for proper hardening.

HARDENING AND HARDENABILITY

13.95

For example, homogeneous cast structures, normalized and annealed cold-formed structures, and grain-refined forged structures (by normalizing) reduce the potential for cracking during quench. 3. Nonuniform heating, localized overheating, exceptionally high austenitizing temperature, and grain coarsening have the potential of cracking due to the presence of quench stresses. Overheating for hardening provides coarse grain size that leads to cracking due mostly to lower toughness. 4. Excessive heating rate to the austenitizing temperature causes surface oxidation and/or decarburization which, in turn, can result in quench cracking. 5. Rapid cooling through the Ms-Mf range causes internal stresses.147a 6. Steel transformation temperature range (Ms-Mf) may show effects on cracking propensity. The higher the carbon and alloy contents, the lower the Ms and Mf temperatures. The cracking tendency typically decreases with the increase of Ms temperature. Steels containing coarse austenite grains and with low Ms temperature are especially susceptible to quench cracking. Usually, the depth of transformation to martensite producing transformation stresses increases with the increase of steel hardenability. Excessively high austenitizing temperature increases the surface-tocore temperature differentials, which, in turn, causes a corresponding increase in residual stress and cracking potential. Cracking can thus result from surface carburization or decarburization, which can affect the transformation characteristics of the surface layers and particularly change the Ms temperature. 7. Stress risers created by improper design of keyways and holes; sharp changes in sections (such as fillets, threads, and gear roots from machining operations); notches and machining or grinding marks; and materials-related defects such as nonmetallic inclusions, lap or seams, surface discontinuities, alloy nonuniformity such as chemical segregation-banding and alloy depletion, and porosity (due to trapped gases in the casting) promote quench cracking. Examples of alloys prone to alloy depletion are AISI 4100, 4300, and 8600 series. 8. Improper selection of steel. Hardenable alloy steel with more than 0.25% C may suffer from quench cracking due to martensite transformation. If the steel chemistry is in excess of the required hardenability limits for the products being processed, it results in cracking. Usually, quench cracks prevail in steels when the carbon equivalent (CE) [given in Eq. (13.15)]148 is in excess of 0.52 to 0.55, as shown in Fig. 13.53 Mn Mo Cr Ni + + + (13.15) 5 5 10 50 where elemental concentrations are in wt%. Additions of small amounts of Nb and B reduce the tendency to quench cracking. Grain refinement of the austenite and high-purity steels, made by vacuum melting, are less sensitive to quench cracking.149 High-carbon and tool steels that are inherently hard and brittle are invariably prone to quench cracking. In carbon tool steels, the fracture grain size should not be less than ASTM 8 (Fig. 13.54) to avoid quench cracking. It appears from Fig. 13.54 that an increasing austenitizing temperature causes less quench cracking for a particular grain size, presumably due to the formation of more retained austenite at higher austenitizing temperature. It is likely that retained austenite inhibits quench cracking. In high-speed steels, fine undissolved V, Nb, and Ti carbides are most effective in resisting grain coarsening.149 9. Time delays between quenching and tempering may also be a contributing factor. CE = C +

13.96

CHAPTER THIRTEEN

FIGURE 13.53 The relationship between the carbon equivalent and the percentage of specimens which quench-crack.148

FIGURE 13.54 The effect of fracture grain size (ASTM) on the tendency to quench-crack in carbon tool steels, using different austenitizing temperature.150

13.12 EFFECT OF CARBON ON HARDNESS IN HARDENED STEEL Hardness is one of the important indicators of the suitability of a steel for a specific application. Figures 13.55 and 13.56 are the plots of hardness of Fe-C martensites as a function of carbon content.151,152 In the former case, the hardness is plotted against the Vicker’s pyramid hardness numbers, while in the latter case it is plotted against the Rockwell C-scale. The essential feature of the Rockwell hardness plot

HARDENING AND HARDENABILITY

13.97

FIGURE 13.55 Relationship between Vickers hardness and carbon content for martensitic, ferrite-pearlite, and spheroidized microstructures in steels. Hatched area shows the influence of retained austenite.151 (Reprinted by permission of ASM International, Materials Park, Ohio.)

FIGURE 13.56 Rockwell hardness of martensite as a function of carbon content.152 (Reprinted by permission of PWS-Kent Publishing Co., Boston, Mass.; after J. L. Burns, T. L. Moore, and R. S. Archer, Trans. ASM, vol. 26, 1938, p. 1.)

is that the hardness curve reaches about 60 Rc at 0.40% C, but it levels off at about 0.6% C, where the hardness attains about 65 Rc. Further increase in the hardness level, therefore, does not occur on exceeding the carbon content above 0.6% because the Rockwell hardness tester is insensitive in the hardness readings found

CHAPTER THIRTEEN

13.98

FIGURE 13.57 Average relationship between carbon content and hardness for steels containing different amounts of martensite in their microstructure.153 (Reprinted by permission of Butterworths, London; after Hodge and Orehoski.)

in the hardened high-carbon steel. The reason for this is that a cone-shaped indentor with a round point in the Rockwell C-scale becomes so blunted with its use that it does not appreciably move (i.e., depth of penetration remains the same) under a fixed load to account for any observable increase in hardness. In the Vicker’s hardness tester, on the other hand, since this indentor does not get blunted at its tip and the hardness measurement is based on load/surface area of an indentation formed, under a given load, by pressing into the metal a square-base diamond pyramid indentor, the DPH values (higher hardness branch of curve) provide a large effect of (increasing) carbon content on the hardness of martensite as compared to that of ferrite-pearlite or spheroidized microstructure in steel. However, a decrease in hardness at higher carbon contents in Fe-C alloys usually results (as indicated by the lower branch of the hardness curve) due to the Mf temperature lying below the room temperature, which causes the formation of a significant proportion of retained austenite in the microstructure. Alloy martensites are usually harder at a given carbon level, due mainly to the fact that the Ms temperature is depressed and less auto-tempering occurs. Figure 13.57 shows the usual relationship between carbon content and hardness for quenched microstructures containing various proportions of untempered martensite. Some of the variations observed in the maximum hardness at various carbon contents may be due to differences in the austenite grain size.

13.13 HARDENABILITY 13.13.1 Definition The hardenability of steel can be defined in several ways. It is defined as the ability of a steel to be hardened by quenching (under given conditions) or as the property

HARDENING AND HARDENABILITY

13.99

TABLE 13.19 Normalizing and Austenitizing Temperatures†155

Steel series

Orderd carbon content, max., %

Normalizing temperature, °F (°C)

Austenitizing temperature, °F (°C)

1000, 1300, 1500, 3100, 4000, 4100

0.25 and under

1700 (925)

1700 (925)

4300, 4400, 4500, 4600, 4700, 5000, 5100, 6100,‡ 8100, 8600, 8700, 8800, 9400, 9700, 9800

0.26 to 0.36, incl. 0.37 and over

1650 (900) 1600 (870)

1600 (870) 1500 (845)

2300, 2500, 3300, 4800, 9300

0.25 and under 0.26 to 0.36, incl. 0.37 and over

1700 (925) 1650 (900) 1600 (870)

1550 (845) 1500 (815) 1475 (800)

9200

0.50 and over

1650 (900)

1600 (870)

† ‡

A variation of ±10°F (6°C) from the temperatures in this table is permissible. Normalizing and austenitizing temperatures are 50°F (30°C) higher for the 6100 series.

of a steel that determines the depth and distribution of hardness produced by quenching from the austenitizing temperature (Table 13.19).153–155 Essentially depth and distribution of hardness are measured by the depth and distribution of martensite structure. The ability or ease to form martensite depends on the ability to avoid or suppress (partially or completely) the formation of diffusion-controlled transformation products of austenite, such as proeutectoid ferrite and cementite, pearlite, and bainite. Hence the microstructural definition of hardenability is the depth to which steel can be transformed from austenite to structure with 50%, 90%, or full martensite at a given temperature.156 The 50% hardenability (martensite criterion) is widely used because of the ease with which it can be metallographically determined, and it is this position across which the hardness variations are most rapid and prominent. In some cases, the 90% martensite criterion is also employed with success (e.g., in shallow-hardenable steel). Hardness and hardenability refer to distinct factors, and a steel capable of developing a high martensite hardness (with a high carbon content) can have a low hardenability with the resulting development of potential maximum hardness in very small sections. Conversely, a steel with a high hardenability will have the capability of hardening larger sections. However, if the carbon content is low, the maximum hardness developed will be also low. This effect is shown in Fig. 13.56. It is clear that steels exhibiting deep-hardness penetration are known as highhardenable steels, while those with shallow hardness penetration are called shallowhardenable steels.

13.13.2 Relation between Nucleation Theory and Hardenability It has been agreed that hardenability of a steel is governed mainly by the nucleation and growth rates of the microstructural components ferrite, pearlite, and/or bainite. Thus a theory of hardenability must include the chemical and structural factors which inhibit or suppress the nucleation and/or growth of these microconstituents.

13.100

CHAPTER THIRTEEN

Phosphorus is next to boron as a promoter of hardenability in hypoeutectoid steel; and like boron, phosphorus is highly surface-active and tends to retard the nucleation of ferrite at austenite grain boundaries.157 According to a current theory of hardenability, nucleation of ferrite takes place preferentially at specific locations of grain boundaries where these are poisoned by the segregation of boron, phosphorus, or other alloying elements which retard their growth. However, the use of phosphorus should be limited to 0.07 to 0.08% in lowalloy, and that of boron to 0.0015 to 0.002% in low- and medium-carbon steels and microalloyed steels because of the problem of temper embrittlement.158

13.13.3 Importance of Hardenability Hardenability is an important property for any steel used in the heat-treated (i.e., quenched or quenched and tempered) condition. It is this concept that determines the formation of particular microconstituents during quenching of a steel component of a given size in a specific quenchant and, therefore, determines the final mechanical properties of the heat-treated material.159 The greater the hardenability of the steel, the slower the rate of cooling required to form a fully martensitic structure. Hardenability is primarily increased by the addition of alloying elements which help to stabilize austenite, resulting in sluggishness of transformation of austenite upon cooling, thereby promoting the response to heat treatment.160 The hardenability imparted by specific alloying elements is a very important consideration in the selection of steels for high-strength complex parts. The reason is that a certain depth of hardening is primarily controlled by the proportion of martensite formed in continuously cooled steels. If the hardenability is low, fast cooling is needed to achieve a large amount of martensite. However, fast cooling causes the development of residual stress, distortion, and even cracking. That is why parts must be cooled as slowly as practical from the austenite phase, and thus the addition of certain alloying elements (such as Mn, Cr, B, or Ni) is necessary either in single addition or more often in combinations, to impart increased hardenability. Low Al and N contents enable austenite grain growth to occur, thereby leading to high hardenability in sheet steel.161 The hardenability of cast and wrought steels is controlled mainly by the chemical composition, austenitizing temperature, grain size, and cooling rate from the austenitizing temperature. Excess hardenability usually represents excess cost. Consequently, it is expedient to optimize the least expensive and most efficient alloy system. In this way, less expensive alternatives are sought to substitute highly alloyed steels, while retaining somewhat the same hardenability.162 In the context of hardenable steels, boron acts as a unique alloying element. Less expensive boron addition (with optimum boron content of 0.0015 to 0.002%) allows an increase of hardenability in certain grades of microalloyed steels with low carbon content, which, together with rapid quenching, produce greater hardened depth and notch toughness without the risk of quench-cracking. Another high-hardenable boron steel is quenched and tempered grade (or martensitic steel) containing 0.29%) grades. For case-hardening steels, we have the following equation: J 4 – 25 (Rc ) = 87 C + 14 Cr + 5.3 Ni + 29 Mo + 16 Mn - 21.2 E + 2.21E + 22 Rc

(13.21)

where E is another factor, defined as the Jominy depth (in 1/16 in.) from the quenched end. For hardenable steels, we have J 4 – 25 (Rc ) = 78 C + 22 Cr + 21Mn + 6.9 Ni + 33Mo - 2.03 E + 1.86E + 18 Rc

(13.22)

As expected, the equation for the hardenable steel possesses higher coefficients. When both the composition and grain size of the 37 steels listed in USS Atlas were considered, Just194 derived two different formulae: J 4 – 40 (Rc ) = 88 C - 0.0135E 2 C + 16 Mn + 5Si + 19Cr + 6.3 Ni + 35Mo - 0.82GASTM - 20 E + 2.11E - 2 Rc

(13.23)

and J 4 –32 (Rc ) = 98 C - 0.025E 2 C + 19 Mn + 5Si + 20Cr + 6.4 Ni + 34 Mo + 28V - 0.82GASTM - 24 E + 2.86E - 1 Rc

(13.24)

The compositional ranges applicable to these formulae are 0.08 to 0.56 C, 0.20 to 1.88 Mn, 0 to 3.8 Si, 0 to 8.94 Ni, 0 to 1.97 Cr, 0 to 0.53 Mo, and 1.5 to 11 ASTM grain size G. † JD is the Jominy distance; when expressed with a numerical subscript instead of the subscript D (Jominy position), it means that number multiplied by 1/16 in. from the quenched end of the bar. For example, J2 = 1 2 ¥ /16 in. (3.175 mm).

13.130

CHAPTER THIRTEEN

These formulae do not predict the hardenability of steel precisely. However, these formulae are intended primarily to (1) aid the designer in determining the selection of steel and (2) assist the metallurgist in correcting the melt. Kirkaldy195,196 has proposed a method for calculating a Jominy curve by means of computers. He assumed initially that the nucleation of ferrite, pearlite, and bainite occurs immediately; however, the growth rate G(T) needs calculation. He has formulated analytically an expression for the cooling rates at various Jominy distances. He further assumed that the cooling rate which produces 0.1% of transformed product at the temperature representing the maximum G(T) also represents the Jominy inflection point where the martensite content is 50% (Fig. 13.77). He was able to plot the whole Jominy curve by determining, first, the carbon content corresponding to the maximum hardness; second, the DI value; and third, the position of the inflection point. Figure 13.78 shows the results due to Kirkaldy and his coworkers. 13.13.5.6 Use of Jominy Hardenability Curves. Jominy hardenability curves are the required method to characterize the steel. They are employed to compare the hardenability of different heats of the same steel grade as a quality control method in the steel manufacture and to compare the hardenability of different steel grades when selecting steel for a specific application. In the latter situation, Jominy curves are used to predict the hardening depth. Such predictions are usually dependent on the assumptions that the rates of cooling prevailing at different distances from the quenched end of the Jominy specimen may be compared with the cooling rates predominating at different locations on the cross sections of bars of different diameters. If the cooling rates are equal, it is assumed that equivalent microstructure and hardness can be expected after quenching. Figure 13.79 shows the correlation between equivalent cooling rates along the Jominy specimen and at four locations in actual round bars up to 4 in. (102 mm) in diameter when quenched in mildly agitated water and oil at 60 m/min (200 ft/min). The cooling rates at both the surface and interior points decrease with the increase in bar diameter. The Jominy data found from Fig. 13.79 can be used directly in steel selection.

FIGURE 13.77 Construction for locating the inflection point of a Jominy curve.195,196 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

HARDENING AND HARDENABILITY

13.131

FIGURE 13.78 Comparison of predicted and observed Jominy curves for five steels.195,196 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

13.13.5.7 Computer Calculations of Jominy Hardenability. More reliable formulae were developed in the late 1970s to be employed in calculations facilitating alloy development, quality control, and control of steelmaking and heat-treating processes.197 Several computerized on-line systems for calculating hardenability curves and numerous properties have been proposed and used. The International Harvester Company has developed the Computer Harmonized Application Tailored (CHAT) Process. This system consists of two basic concepts: the first procedure, termed CH, is designed to quantitatively determine the hardenability requirement in terms of DI by using a modified linear programming technique known as separable programming. The second procedure, termed AT, is employed to determine minimum and maximum hardenability levels required to design or select the most economical steel with composition sufficient to develop the desired engineering properties in a part for a given application.198 In the Creusot-Loire method, the general features of the kinetic properties of low-alloy steels are described by mathematical models fitted to experimental data by regression analysis. A set of 341 carefully selected experimental CCT diagrams have been employed to obtain regression equations for the critical cooling rate at 700°C for onset, specific percent transformation, or completion of different decomposition reactions of austenite in the quenched and quenched and tempered conditions, and to calculate the mechanical properties such as hardness, yield and tensile strengths, making use of a sum rule over volume fractions. The equations are valid in the following composition range: 0.2 to 0.5 C, 0 to 2 Mn, 0 to 1 Si, 0 to 4 Ni, 0 to 3 Cr, 0 to 1 Mo, and 0 to 0.2 V. This process is increasingly used for rapid calculation of hardness, Jominy curves, mechanical properties of welded steels, as-rolled steel products (plates and bars), and forgings and steels for nuclear application. The use of the personal computer software package HAZ CALCULATOR actually facilitates the application of this method with satisfactory accuracy.199,200 There are also several versions on the market which are sold as a compact disk. For best results, companies that use these should calibrate to their own

13.132

CHAPTER THIRTEEN

FIGURE 13.79 Equivalent cooling rates for round bars quenched in (a) water and (b) oil.153 Data for surface hardness are for “mild agitation”; other data are for 60 m/min (200 ft/min). (Reprinted by permission of ASM International, Materials Park, Ohio; after Jatczak.)

HARDENING AND HARDENABILITY

13.133

FIGURE 13.80 Outputs from Minitech Predictor data processing program for best fit to measured Jominy data. (a) Initial trial; (b) final trial.202

historical data. Current clean steels have a larger grain size and, therefore, higher hardenability. Another system, the Minitech Computerized Alloy Steel Information System, involves 12 computer programs based on fundamental theory, alloy composition, and other input parameters; and it provides entire Jominy curves (based on the theoretical developments of the inflection point principle of Kirkaldy and co-workers), mechanical properties of hot-rolled products, quenched and tempered products, carburized products, hardness distribution, and weldability parameters with good accuracy for boron and non-boron steels containing 0.07 to 1.3% C and alloy ranges of AISI-SAE hardenability steels.201 This system is strongly calibrated to plant data. Figure 13.80 shows a typical output of the Minitech Predictor operating in the data processing mode for best fit to measured Jominy data at the initial and final trials, where the input values include chemical composition, Jominy hardness values, and predicted grain size.202

13.14 HARDENABILITY (OR H-) STEELS Hardenability steels exhibiting hardenability limits have chemical composition ranges slightly broader than the equivalent normal AISI-SAE grades, to allow the manufacturers greater flexibility in achieving hardenability curves within the spec-

CHAPTER THIRTEEN

13.134

ified limits. Consequently, these hardenability steels offer a wide range of mechanical properties that are based on the maximum as-quenched hardness and the degree of tempering of martensite after quenching.203 These hardenability limits are also called hardenability bands. These steel grades are designated by the letter H after the composition code or preceding UNS designation. The width of the H-band for a particular steel is agreed upon by a compromise between the interests of the steel producer,who would prefer a larger width to increase the volume of heats that meet the specification, and the user, who would prefer a narrow band to obtain uniform mechanical properties of the finished products.

13.14.1 Shallow-Hardenable Steels (H-Band Carbon Steels) Tables 13.26 and 13.27 list composition ranges of H-band AISI-SAE carbon steels and alloy steels, respectively. Carbon steels with very low Mn (say, 0.3%) and practically free of residual Ni, Cr, and Mo have the lowest hardenability. Most of the H-band 1xxx series carbon steels have 0.6 to 0.9% Mn content, although there are many exceptions. Those steels containing Mn in the range of 1.00 to 1.65% constitute the 15xx series, where higher Mn level contributes a significant increase in hardenability. High-carbon steels (between 0.55 and 1.00% carbon) are more restricted in their applications because they are more costly and more difficult to fabricate due to decreased formability, machinability, and weldability. At the same time, they are less ductile in the quenched and tempered condition. Higher-carbon steels such as 1070 and 1095 are especially suitable for springs where resistance to fatigue and permanent set is a necessary requirement. Most of the parts made from steels in this

TABLE 13.26 Composition of Carbon and Carbon-Boron H Steels204 Ladle chemical composition, wt%

UNS No.

SAE or AISI No.

C

Mn

Si

P, maximum‡

S, maximum‡

H10380 H10450 H15220 H15240 H15260 H15410 H15211 H15281 H15301 H15351 H15371 H15411 H15481 H15621

1038H 1045H 1522H 1524H 1526H 1541H 15B21H† 15B28H† 15B30H† 15B35H† 15B37H† 15B41H† 15B48H† 15B62H†

0.34/0.43 0.42/0.51 0.17/0.25 0.18/0.26 0.21/0.30 0.35/0.45 0.17/0.24 0.25/0.34 0.27/0.35 0.31/0.39 0.30/0.39 0.35/0.45 0.43/0.53 0.54/0.67

0.50/1.00 0.50/1.00 1.00/1.50 1.25/1.75 1.00/1.50 1.25/1.75 0.70/1.20 1.00/1.50 0.70/1.20 0.70/1.20 1.00/1.50 1.25/1.75 1.00/1.50 1.00/1.50

0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.40/0.60

0.040 0.040 0.040 0.040 0.040 0.040 0.040 0.040 0.040 0.040 0.040 0.040 0.040 0.040

0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050

†

These steels contain 0.0005 to 0.003% B. If electric furnace practice is specified or required, the limit for both phosphorus and sulfur is 0.025%, and the prefix E is added to the SAE or AISI number. Source: 2001 SAE Handbook, vol. 1, Materials, Fuels, Emissions, and Noise, Society of Automotive Engineers, Warrendale, Pa., 2001. ‡

TABLE 13.27 Composition of Standard Alloy H Steel204 The ranges and limits on this table apply only to material not exceeding 1.3 ¥ 105 mm2 (200 in.2) in crosssectional area, 460 mm (18 in.) in width, or 4.5 Mg (10,000 lb) in weight per piece. Ranges and limits are subject to the permissible variations for product analysis shown in Table 4 of SAE J409. Ladle chemical composition, wt%

13.135

UNS No.

SAE or AISI No.

C

Mn

Si

Ni

Cr

Mo

V

H13300 H13350 H13400 H13450 H40270 H40280c H40320 H40370 H40420 H40470 H41180 H41300 H41350 H41370 H41400 H41420 H41450 H41470 H41500 H41610 H43200 H43400 H43406d H46200

1330H 1335H 1340H 1345H 4027H 4028Hc 4032H 4037H 4042H 4047H 4118H 4130H 4135H 4137H 4140H 4142H 4145H 4147H 4150H 4161H 4320H 4340H E4340Hd 4620H

0.27/0.33 0.32/0.38 0.37/0.44 0.42/0.49 0.24/0.30 0.24/0.30 0.29/0.35 0.34/0.41 0.39/0.46 0.44/0.51 0.17/0.23 0.27/0.33 0.32/0.38 0.34/0.41 0.37/0.44 0.39/0.46 0.42/0.49 0.44/0.51 0.47/0.54 0.55/0.65 0.17/0.23 0.37/0.44 0.37/0.44 0.17/0.23

1.45/2.05 1.45/2.05 1.45/2.05 1.45/2.05 0.60/1.00 0.60/1.00 0.60/1.00 0.60/1.00 0.60/1.00 0.60/1.00 0.60/1.00 0.30/0.70 0.60/1.00 0.60/1.00 0.65/1.10 0.65/1.10 0.65/1.10 0.65/1.10 0.65/1.10 0.65/1.10 0.40/0.70 0.55/0.90 0.60/0.95 0.35/0.75

0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 1.55/2.00 1.55/2.00 1.55/2.00 1.55/2.00

... ... ... ... ... ... ... ... ... ... 0.30/0.70 0.75/1.20 0.75/1.20 0.75/1.20 0.75/1.20 0.75/1.20 0.75/1.20 0.75/1.20 0.75/1.20 0.65/0.95 0.35/0.65 0.65/0.95 0.65/0.95 ...

... ... ... ... 0.20/0.30 0.20/0.30 0.20/0.30 0.20/0.30 0.20/0.30 0.20/0.30 0.08/0.15 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.25/0.35 0.20/0.30 0.20/0.30 0.20/0.30 0.20/0.30

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

ABLE 13.27 Composition of Standard Alloy H Steel204 (Continued) The ranges and limits on this table apply only to material not exceeding 1.3 ¥ 105 mm2 (200 in.2) in crosssectional area, 460 mm (18 in.) in width, or 4.5 Mg (10,000 lb) in weight per piece. Ranges and limits are subject to the permissible variations for product analysis shown in Table 4 of SAE J409. Ladle chemical composition, wt%

13.136

UNS No.

SAE or AISI No.

C

Mn

Si

Ni

Cr

Mo

V

H47180 H47200 H48150 H48170 H48200 H50401e H50441e H50460 H50461e H50501e H50601e H51200 H51300 H51320 H51350 H51400 H51470 H51500 H51550 H51600 H51601e H61180 H61500

4718H 4720H 4815H 4817H 4820H 50B40He 50B44He 5046H 50B46He 50B50He 50B60He 5120H 5130H 5132H 5135H 5140H 5147H 5150H 5155H 5160H 51B60He 6118H 6150H

0.15/0.21 0.17/0.23 0.12/0.18 0.14/0.20 0.17/0.23 0.37/0.44 0.42/0.49 0.43/0.50 0.43/0.50 0.47/0.54 0.55/0.65 0.17/0.23 0.27/0.33 0.29/0.35 0.32/0.38 0.37/0.44 0.45/0.52 0.47/0.54 0.50/0.60 0.55/0.65 0.55/0.65 0.15/0.21 0.47/0.54

0.60/0.95 0.45/0.75 0.30/0.70 0.30/0.70 0.40/0.80 0.65/1.10 0.65/1.10 0.65/1.10 0.65/1.10 0.65/1.10 0.65/1.10 0.60/1.00 0.60/1.00 0.50/0.90 0.50/0.90 0.60/1.00 0.60/1.05 0.60/1.00 0.60/1.00 0.65/1.10 0.65/1.10 0.40/0.80 0.60/1.00

0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35

0.85/1.25 0.85/1.25 3.20/3.80 3.20/3.80 3.20/3.80 ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

0.30/0.60 0.30/0.60 ... ... ... 0.30/0.70 0.30/0.70 0.13/0.43 0.13/0.43 0.30/0.70 0.30/0.70 0.60/1.00 0.75/1.20 0.65/1.10 0.70/1.15 0.60/1.00 0.80/1.25 0.60/1.00 0.60/1.00 0.60/1.00 0.60/1.00 0.40/0.80 0.75/1.20

0.30/0.40 0.15/0.25 0.20/0.30 0.20/0.30 0.20/0.30 ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 0.10/0.15 0.15

13.137

H81451e H86170 H86200 H86220 H86250 H86270 H86300 H86301e H86370 H86400 H86420 H86450 H86451e H86500 H86550 H86600 H87200 H87400 H88220 H92600 H93100d H94151e H94171e H94301e

81B4S5e 8617H 8620H 8622H 8625H 8627H 8630H 86B30He 8637H 8640H 8642H 8645H 86B45He 8650H 8655H 8660H 8720H 8740H 8822H 9260H 9310Hd 94B15He 94B17He 94B30He

0.42/0.49 0.14/0.20 0.17/0.23 0.19/0.25 0.22/0.28 0.24/0.30 0.27/0.33 0.27/0.33 0.34/0.41 0.37/0.44 0.39/0.46 0.42/0.49 0.42/0.49 0.47/0.54 0.50/0.60 0.55/0.65 0.17/0.23 0.37/0.44 0.19/0.25 0.55/0.65 0.07/0.13 0.12/0.18 0.14/0.20 0.27/0.33

0.70/1.05 0.60/0.95 0.60/0.95 0.60/0.95 0.60/0.95 0.60/0.95 0.60/0.95 0.60/0.95 0.70/1.05 0.70/1.05 0.70/1.05 0.70/1.05 0.70/1.05 0.70/1.05 0.70/1.05 0.70/1.05 0.60/0.95 0.70/1.05 0.70/1.05 0.65/1.10 0.40/0.70 0.70/1.05 0.70/1.05 0.70/1.05

0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35 1.70/2.20 0.15/0.35 0.15/0.35 0.15/0.35 0.15/0.35

0.15/0.45 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 0.35/0.75 ... 2.95/3.55 0.25/0.65 0.25/0.65 0.25/0.65

a Small quantities of certain elements may be found in alloy steel that are not specified or required. These elements are to be considered incidental and acceptable to the following maximum amounts: copper to 0.35%, nickel to 0.25%, chromium to 0.20%, and molybdenum to 0.06%. b For open hearth and basic oxygen steels, maximum sulfur content is to be 0.040%, and maximum phosphorus content is to be 0.035%. Maximum phosphorus and sulfur in basic electric furnace steels are to be 0.025% each. c Sulfur content range is 0.035/0.050%. d Electric furnace steel. e These steels contain 0.0005 to 0.003% B. Source: 2001 SAE Handbook, vol. 1, Materials, Fuels, Emissions, and Noise, Society of Automotive Engineers, Warrendale, Pa., 2001.

0.30/0.60 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 0.35/0.65 ... 1.00/1.45 0.25/0.55 0.25/0.55 0.25/0.55

0.08/0.15 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.15/0.25 0.20/0.30 0.20/0.30 0.30/0.40 ... 0.08/0.15 0.08/0.15 0.08/0.15 0.08/0.15

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

13.138

CHAPTER THIRTEEN

category are hardened by conventional quenching. Water-quenching is employed for heavy sections of the lower-carbon steels and for cutting edges, whereas oilquenching is employed for general purposes. Austempering and martempering often successfully take advantage of the considerably reduced distortion, negligible breakage, and greater toughness at high hardness. In spite of all favorable hardenability factors and the use of a drastic quench, these steels are much shallower hardening than are alloy steels because carbon singly, or in combination with Mn, in amounts even up to 1.65% does not promote deep hardening. The dangers of quench-cracking in these steel parts, especially in the nonuniform section, are very high, which requires closer supervision and control. Screwdrivers, pliers, and wrenches (except the Stillson type) are generally oilhardened, followed by tempering to the required hardness range. Even when reduction in as-quenched hardness is not desired, stress-relieving at 150 to 190°C (300 to 375°F) should be accomplished to help prevent sudden service failures. In Stillsontype wrenches, the jaw teeth are actually edges and therefore are always quenched in water or brine to produce a hardness of 50 to 60 Rc. To obtain considerable structural strength, either the jaws should be locally heated and quenched, or the parts should be heated all over and the jaws locally time-quenched in water or brine. The entire part is then oil-quenched for partial hardening of the remainder. Since hammers must possess high hardness on the striking surface and lower hardness on the claws, they are usually locally hardened and tempered on each end. The striking face is always quenched in water or brine. Stress-relieving is done at about 175°C (350°F). The final hardness on the striking face may range from 50 to 58 Rc, and that on claws may range from 40 to 47 Rc. Hand cutting tools (such as axes and hatchets) and mower blades must have high hardness and high relative toughness in their cutting edges, together with the ability to hold a sharp edge. For hardening of hand cutting tools, their cutting edges are usually heated fast in liquid (salt) baths to the lowest temperature at which the pieces can be hardened and are then quenched in brine. Here, all the spheroidal carbides don’t go into solution during austenitization. As a result, the cutting edge of the tool consists of martensite with less carbon than the actual composition of the steel and cementite particles. In this situation, the tool attains its maximum toughness with respect to its hardness, and particles of cementite provide a long service life of the cutting edge. The final hardness at the cutting edge is usually 55 to 60 Rc.204 It is more economical to use carbon steels whenever possible. The higher-Mn grades are more expensive than the lower-Mn grades, but are less expensive than the least expensive alloy grades.204 Agricultural machinery parts such as plowshares, moldboards, coulters, cultivator shovels, mower and binder knives, disks for harrows and plows, ledger plates, and band knives used for cutting or turning soil are made of high-carbon steels in the heat-treated condition. Grass-cutting and grain-cutting tools are usually made of 1090 or 1095 steel for the required edge to give a prolonged service life. Local hardening of cutting edges with the provision of fixture and subsequent oil quenching and tempering at low temperature produces a final hardness of 55 to 60 Rc on the cutting edges.204

13.14.2 Deep-Hardening Steels (H-Band Alloy Steels) The common alloying elements dissolved in austenite prior to quenching increase hardenability in the following ascending order: Ni, Si, Mn, Cr, Mo, V, and B. The

HARDENING AND HARDENABILITY

13.139

addition of several alloying elements in small amounts is more potent in increasing hardenability than the addition of much larger amounts of one or two alloying elements. To increase hardenability, alloying elements should go into the solution in the austenite during austenitization. Steels having carbide formers such as Cr, Mo, and V need higher temperatures to dissolve their (alloy) carbides. This dissolution proceeds more slowly than for the usual iron carbide. Hence, the heating schedule should be selected in such a way that a sufficient amount of these elements should dissolve in austenite. Otherwise, carbide-forming elements will remain undissolved, causing nonattainment of requisite hardenability. Advantages of Alloy H-Steels 1. The presence of adequate alloying additions allows the use of a lower carbon content for a given application. The decrease in hardenability due to decrease in carbon content may be readily compensated by the increasing hardenability of the alloying addition. 2. The lower carbon content will induce a much lower susceptibility to quenchcracking which arises from the increased MS temperature and from greater plasticity of low-carbon martensite. 3. It is possible to allow slower rates of cooling during quenching for a given section thickness due to increased hardenability, which, in turn, produces a decrease in thermal gradient and cooling stress. In fact, less drastic quenching used for deephardenable steels causes lower distortion and negligible cracking. 4. Increased hardenability of these alloy steels may allow austempering and martempering to be successfully performed, which decreases the residual stress to a very low level prior to tempering, thereby minimizing distortion and danger of cracking. 5. For the same hardness levels: (a) reduction in area is greater for alloy steels than for plain carbon steels, and (b) fully quenched alloy steels need higher tempering temperatures than carbon steels. This higher tempering temperature reduces the stress level in the finished components without sacrificing mechanical properties. Specifying Hardenability Band. Figure 13.81 summarizes the alternative method used to specify a steel based on the hardenability band.205 There are many methods to designate the hardness limits in the Jominy curve: (1) minimum and maximum hardness values at designated distance, such as A-A in the Jominy curve representing the section size used by the purchaser; (2) particular hardness value at the minimum and maximum Jominy distances, as at points B-B; (3) two maximum hardness values at two designated Jominy distances, as at points C-C; and (4) two minimum hardness values at two desired Jominy distances, as at points D-D; or (5) any minimum hardness plus any maximum hardness, as at E-E. When the full hardenability band is specified for alloy steel, it should be described by hardness values at the following distances from the quenched end of the Jominy bar: 1, 2, 3, 8, 12, 16, 20, 24, 28, and 32, in sixteenths of an inch.205 For carbon H steels, hardness values at 1/32 in. should be reported through the distance 8 (¥1/16 in.) from the quenched end as well as the distances listed for alloy H steel. For each H steel, the limits of hardenability band at each Jominy position are presented in tabular form and graphically (Fig. 13.82), and specifications are written for these tabulations and graphical plots.205

13.140

CHAPTER THIRTEEN

FIGURE 13.81 Examples showing alternative methods of specifying hardenability requirements.205 (Reprinted by permission of ASTM, Philadelphia, Pa.)

It is necessary to determine the minimum hardenability in order to attain the proper functioning of the heat-treated steel part; this also serves as a conservative basis for hardenability calculations. More than requisite hardenability should not be specified because it increases costs and gives rise to tensile residual stresses on the quenched surface. There are some approximate guidelines that can be adopted in the selection of an appropriate alloy steel.205 For highly stressed parts, a minimum of 80% martensite at the center of the largest round section is usually required. For moderately stressed parts, a 50% martensite structure at the center is frequently adequate. For most automotive parts, 80% martensite at the 3/4 radius position is sufficient. The state of stress in each component should be known. If stress applied in bending is such that the outer layers are the most highly stressed in tension, the steel should have lower hardenability.

13.15 ALLOY STEEL SELECTION GUIDE BASED ON HARDENABILITY This information is used as a guide in the selection of steel bars based on section thickness and mechanical properties desired in the final production part.

C

HARDNESS LIMITS FOR SPECIFICATION PURPOSES

13.141

MAX

MIN

60 60 60 60

53 53 53 53

5 6 7 8

60 60 60 60

53 53 53 52

9 10 11 12

60 60 59 59

52 52 51 51

13 14 15 16

59 58 58 58

50 49 49 48

18 20 22 24

58 57 57 57

47 46 45 44

26 28 30 32

57 56 56 56

43 42 41 40

0.65

1.55

0.20 0.30

0.95

2.00

0.35

0.90

Mo

Cr

Ni

Si 0.15

0.55 0.44

0300 N

DIAMETERS OF ROUNDS WITH SAME AS QUENCHED HARDNESS

QUENCH

LOCATION IN ROUND SURFACE

3.8 1.1 0.7

2.0 1.2

2.9 1.6

3.8 2.0

4.8 2.4

5.8 2.8

6.7 3.2

3.6

3.9

0.8 0.5 0.2

1.8 1.0 0.5

2.5 1.6 1.0

3.0 2.0 1.4

3.4 2.4 1.7

3.8 2.8 2.0

3.2 2.4

3.6 2.8

4.0 3.1

2

4

6

8

10

12

14

16

18

3/4 RADIUS FROM CENTER CENTER SURFACE 3/4 RADIUS FROM CENTER CENTER

MILD WATER QUENCH MILD OIL QUENCH

65 60 ROCKWELL HARDNESS C SCALE

“J” DISTANCE SIXTEENTHS OF AN INCH 1 2 3 4

Mn

0.37

55 50 45 40 35 30 25

HEAT TREATING TEMPERATURES RECOMMENDED BY SAE * NORMALIZE 1600 °F * AUSTENITIZE 1550 °F * For forged or rolled specimens only.

20

20

22

24

26

DISTANCE FROM QUENCHED END—SIXTEENTHS OF AN INCH NOTE – 1 in. = 25.4 mm.

FIGURE 13.82 Tabulation and graphical representation of the hardenability band of AISI-SAE 4340 H steel.205 (Reprinted by permission of ASTM, Philadelphia, Pa.)

28

30

32

TABLE 13.28 Alloy Steel Selection Guide for Highly Stressed Parts204,206 Unless otherwise indicated in the footnotes, any steel in this table may be considered for a lower strength level or a smaller section, or both. Steels to give 80% martensite, minimum, for indicated location in a round section of indicated diameter Required yield strength MPa

As-tempered hardness ksi

HRC

HB

13.142

Oil-quenched and tempered 620–860a 90–125b 23–30

241–285

860–1030c

125–150

30–36d

285–341

1030–1170e

150–170

36–41f

331–375

1170–1275g

170–185

41–46h

375–429

>185

46 min.j

429 min.

>1275i

At center £13 mm (1/2 in.) 1330H 5132H 4130H 8630H 1335H 5135H 1340H 5140H 4135H 8637H 94B30H 3140H 50B46H 5145H 50B40H 4140H 8640H 8642H 8645H 8740H 8742H 5150H 5155H 50B44H 5147H 9260H

13–25 mm (1/2–1 in.)

At 3/4 radius

At midradius 25–38 mm (1–11/2 in.)

38–50 mm (11/2–2 in.)

50–63 mm (2–21/2 in.)

63–75 mm (21/2–3 in.)

4137H

4142H

9840H

4140H 94B40H

4145H 9840H

86B45H 4337H

4147H 4340H

4150H

75–89 mm (3–31/2 in.)

89–102 mm (31/2–4 in.)

94B30H

4135H 8640H 94B30H 8740H 50B40H 4137H 8642H 8645H 8742H 5155H 50B44H 5147H 94B40H 6150H

81B45H 4142H 4145H 8650H 8655H 4337H

86B45H 9840H

5160H 50B50H 9262H 4147H 8655H

50B60H 51B60H 8660H

4150H

4337H 4340H 4340H

9805H E4340H

9850H

81B45H 8650H 86B45H 6150H

13.143

Water-quenched and temperedk 620–860a 90–125 23–30b

241–285

860–1030c

125–150

30–36d

285–341

1030–1170e

150–170

36–41f

331–375

1170–1275g

170–185

41–46h

375–429

>185

46 min.j

429 min.

>1275i

1330H 1335H 5130H 5132H 5135H 4130H 8630H 5140H 4037H 4042H 4137H 8637H 5046H 50B46H 5145H 4047H 4142H 8642H

5130H 5132H 4130H 8630H 1330H 5135H

5135H

4042H 4047H

1340H 50B46H 5140H 4135H 8637H 94B30H 3140H 5145H 50B40H 8640H 8642H 8740H

1340H 50B46H 3140H

5147H 4145H 8645H 86B45H

1335H

50B44H

4135H

94B30H

4135Hl 8640Hl 8740Hl 3140Hl 50B40Hl 4137Hl 8642Hl 8745Hl

1340Hm 8637Hm

50B40H 8642H 94B30H

4137H 4140H

94B40H

8640Hm 8740Hm

50B44H 5147H 4140H 8645H 8742H

94B40H

81B45H 4142H 4337H

50B44Hl 5147Hl 81B45Hl 94B40Hl

4140Hm 8645Hm 8742Hm

4142H

81B45H 4337H

81B45Hm

4147H

4145H 4147H 86B45H 9840H 4340H E4340H

a Tensile strength, 790 to 940 MPa (115 to 138 ksi). b As-quenched hardness, 42 HRC, or 388 HB. c Tensile strength, 940 to 1100 MPa (136 to 160 ksi). d As-quenched hardness, 44 HRC, or 415 HB. e Tensile strength, 1100 to 1300 MPa (160 to 188 ksi). f As-quenched hardness, 48 HRC, or 461 HB. g Tensile strength, 1300 to 1530 MPa (188 to 222 ksi). h As-quenched hardness, 51 HRC, or 495 HB. i Tensile strength, over 1530 MPa (222 ksi). j As-quenched hardness, 55 HRC, or 555 HB. k Through steels with 0.47% C nominal. l May be substituted for steels listed under the 50 to 63 mm (2 to 21/2 in.) column at same strength level or less. m Not recommended for applications requiring 80% martensite at midradius in sections 38 to 50 mm (11/2 to 2 in.) in diameter because of insufficient hardenability. Source: Republic Alloy Steels, Republic Steel Corporation, 1961.

TABLE 13.29 Alloy Steel Selection Guide for Moderately Stressed Parts204,206 Unless otherwise indicated in the footnotes, any steel in this table may be considered for a lower strength level or a smaller section, or both. Steels to give 50% martensite, minimum, for indicated location in a round section of indicated diameter Required yield strength MPa

As-tempered hardness ksi

HRC

HB

At center £13 mm (1/2 in.)

13–25 mm (1/2–1 in.)

13.144

Oil-quenched and tempered 620–860a 90–125 23–30b

241–285

1330H 5132H 4130H 8630H

8737H

860–1030c

125–150

30–36d

285–341

1335H 4042H 4047H 5135H

1030–1170e

150–170

36–41f

331–375

1170–1275g

170–185

41–46h

375–429

1340H 5140H 4135H 8637H 94B30H 3140H 5145H 50B40H 50B46H 4063H 4140H 8640H

4135H 8640H 94B30H 8740H 3140H 5150H 50B40H 4137H 8642H 8645H 8742H 5155H 50B44H 5147H 94B40H 6150H

At 3/4 radius

At midradius 25–38 mm (1–11/2 in.)

38–50 mm (11/2–2 in.)

50B40H 8642H 94B30H 8740H 3140H 50B44H 5147H 4137H 8645H 8742H 5160H 50B50H 4140H 94B40H 6150H

4140H 94B40H

81B45H 4142H 4145H 8650H 8655H 4337H

50–63 mm (2–21/2 in.)

63–75 mm (21/2–3 in.)

75–89 mm (3–31/2 in.)

89–102 mm (31/2–4 in.)

4142H

4142H

4145H

4147H 86B45H 9840H

51B60H 8655H

4145H 9840H

4147H 86B45H 4337H

4150H 4340H

86B45H 9840H

4147H 8660H 4340H

4150H

4337H 4340H

9850H E4340H

>1275i

46 min.j

429 min.

Water-quenched and temperedk 620–860a 90–125 23–30b

241–285

860–1030c

125–150

30–36d

285–341

1030–1170e

150–170

36–41f

331–375

>185

8642H 8745H 8740H 8742H 5150H 5155H 50B44H 5147H 9260H 81B45H 8650H 86B45H 6150H

13.145 1330H 1335H 5130H 5132H 5135H 4130H 8620H

5160H 50B50H 9262H 4147H 8655H

50B60H 51B60H 8660H

4150H

4037H 5130H 5132H 4130H 8630H 1330H 5135H

5135H

8637Hl

5140Hm

4135H

50B40H 8642H 94B30H 3140H

4137H

1335H

4135Hl

1340Hm 8637Hm

50B44H 5147H 4137H 8645H 8742H

4140H 94B40H

4042H 4047H

1340H 50B46H 5140H 4135H 8637H 94B30H 3140H

50B40Hl 4137Hl 8642Hl

5145Hm 8640Hm 8740Hm

50B40H 8640H 8642H 94B30H 8740H 3140H 50B44H 5147H 4140H 8645H 8742H

94B40H

81B45H 4142H 4337H

9850H

TABLE 13.29 Alloy Steel Selection Guide for Moderately Stressed Parts204,206 (Continued) Unless otherwise indicated in the footnotes, any steel in this table may be considered for a lower strength level or a smaller section, or both. Steels to give 50% martensite, minimum, for indicated location in a round section of indicated diameter Required yield strength MPa 1170–1275g

13.146

>1275i

As-tempered hardness ksi

HRC

HB

170–185

41–46h

375–429

>185

46 min.j

429 min.

At center £13 mm (1/2 in.)

At 3/4 radius

At midradius

13–25 mm (1/2–1 in.)

25–38 mm (1–11/2 in.)

38–50 mm (11/2–2 in.)

50–63 mm (2–21/2 in.)

5140H 4037H 4042H 4137H 8637H

1340H 50B46H 3140H

5145H 50B40H 8640H 8642H 8740H

50B44Hl 5147Hl 94B40Hl

5046H 50B46H 5145H 4047H 4142H 8742H

5147H 4145H 8645H 86B45H

50B44H

63–75 mm (21/2–3 in.)

75–89 mm (3–31/2 in.)

89–102 mm (31/2–4 in.)

4140Hm 8645Hm 8742Hm

4142H

81B45H 4337H

4145H 4147H 86B45H 9840H 4340H E4340H

81B45Hm

4147H

a Tensile strength, 790 to 940 MPa (115 to 138 ksi). b As-quenched hardness, 42 HRC, or 388 HB. c Tensile strength, 940 to 1100 MPa (136 to 160 ksi). d As-quenched hardness, 44 HRC, or 415 HB. e Tensile strength, 1100 to 1300 MPa (160 to 188 ksi). f As-quenched hardness, 48 HRC, or 461 HB. g Tensile strength, 1300 to 1530 MPa (188 to 222 ksi). h As-quenched hardness, 51 HRC, or 495 HB. i Tensile strength, over 1530 MPa (222 ksi). j As-quenched hardness, 55 HRC, or 555 HB. k Through steels with 0.47% C nominal. l May be substituted for steels listed under the 50 to 63 mm (2 to 21/2 in.) column at same strength level or less. m Not recommended for applications requiring 50% martensite at midradius in sections 38 to 50 mm (11/2 to 2 in.) in diameter because of insufficient hardenability. Source: Republic Alloy Steels, Republic Steel Corporation, 1961.

HARDENING AND HARDENABILITY

13.147

Table 13.28 shows the H-band alloy steel selection guide for highly stressed parts, oil- or water-quenched to produce a minimum 80% martensite for round sections up to 4 in. (102 mm) in diameter. Table 13.29 provides the H-band alloy steel selection guide for moderately stressed parts, oil- or water- quenched to produce a minimum 50% martensite at the indicated locations for round sections up to 4 in. (102 mm) in diameter. An H-steel with the same grade specification as a standard SAE-AISI steel is capable of meeting the same section and strength requirements as the standard steel and is the preferred method of specification. These are ranked approximately in order of increasing cost. When one steel is substituted for another, it should be shifted to the left in the table, to a smaller diameter, or upward to a lower strength.204,206

REFERENCES 1. R. W. Monroe and C. E. Bates, J. Heat Treat., vol. 3, no. 2, 1983, pp. 83–94. 2. K. E. Thelning, J. Heat Treat., vol. 3, no. 2, 1983, pp. 100–107. 3. Heat Treatment of Steels, B. P., England. 4. M. Tagaya and I. Tamura, KHG, vol. 6, no. 1, 1955, pp. 7–12. 5. M. Narazaki, A. Asada, and K. Fukahara, in Proceedings: Second International Conference on Quenching and Control of Distortion, 4–7 Nov. 1996, eds. G. E. Totten et al., ASM International, Cleveland, Ohio, 1996, pp. 37–46. 6. C. E. Bates, G. E. Totten, and R. L. Brennan, in ASM Handbook, vol. 4: Heat Treating, 10th ed., ASM International, Materials Park, Ohio, 1991, pp. 67–120. 7. W. Leslie, Physical Metallurgy of Steel, McGraw-Hill, New York, 1981. 7a. A. V. Reddy, D. A. Akers, L. Chuzhoy, M. A. Pershing, and R. A. Wildow, Heat Treating Progress, June/July 2001, pp. 40–42. 8. G. E. Totten, C. E. Bates, and M. A. Clinton, Handbook of Quenchants and Quenching Technology, ASM International, Materials Park, Ohio, 1993. 9. B. Liscic, Hart-Tech. Mitt., vol., 33, 1978, pp. 179–191. 10. V. Paschkis and G. Slotz, Iron Age, vol. 22, 1956, pp. 95–97. 11. M. Tagaya and I. Tamura, Technol. Rep. Osaka Univ., vol. 7, 1957, pp. 403–424. 12. B. Liscic and T. Filetin, J. Heat Treat., vol. 5, no. 2, 1988, pp. 115–124. 13. G. E. Totten, G. M. Webster, H. M. Tensi, and B. Liscic, Advanced Mats. & Process., June 1997, pp. 68LL–68OO. 14. N. A. Hilder, Ph.D. thesis, University of Birmingham, Aston, U.K., January 1988. 15. G. E. Totten, G. M. Webster, H. M. Tensi, and L. M. Jarvis, in Proceedings: Second International Conference on Quenching and Control of Distortion, 4–7 Nov. 1996, eds. G. E. Totten et al., ASM International, Materials Park, Ohio, 1996, pp. 585–593. 16. W. F. Walker, Tooling, December 1968, pp. 13–25. 17. Metals Handbook, vol. 4, 9th ed., ASM, Metals Park, Ohio, 1981. 17a. W. Luty, in Theory and Technology of Quenching, eds. B. Liscic, H. M. Tensi, and W. Luty, Springer-Verlag, Berlin, 1992, pp. 248–340. 18. T. W. Dicken, Heat Treat. Met., vol. 1, 1986, pp. 6–8. 19. D. Paddle, Ind. Heat., January 1998, pp. 40–43. 20. J. A. Hasson, Met. Progr., vol. 128, no. 4, 1985, pp. 67–72. 21. G. R. Weymueller, Met. Progr., vol. 102, no. 1, 1972, pp. 38–50.

13.148

CHAPTER THIRTEEN

22. T. I. Tkachuk, N. Ya Rudakova, B. K. Sheremeta, and M. A. Al’tshuler, Metalloved Term. Obrab. Met., October 1986, pp. 45–47. 23. T. I. Tkachuk, N. Ya Rudakova, B. K. Sheremeta, and R. D. Novoded, Metalloved Term. Obrab. Met., October 1986, pp. 42–45. 24. R. J. Brennan, Ind. Heat., January 1993, pp. 29–31. 25. I. F. Moreaux, J. M. Naul, G. Beck, and J. Oliver, Heat Treatment ’84, The Metals Society, London, 1984, pp. 18.1–18.5. 25a. D. Moore, Heat Treating Progress, June/July 2001, pp. 29–33. 26. J. Bodin and S. Segerberg, Heat Treat. Met., vol. 20, no. 1, 1993, pp. 15–23; in Quenching and Carburizing, The Institute of Materials, London, 1993, pp. 33–54. 27. H. J. Gilliland, Met. Progr., October 1960, pp. 111–114. 28. E. A. Bender and H. J. Gilliland, Steel, December 1957, pp. 56–59. 29. C. A. Barley and J. S. Aarons, eds., The Lubrication Engineer Manual, U.S. Steel Corporation, Pittsburgh, Pa., 1971, pp. 56–57. 30. G. R. Furman, Lubrication, vol. 57, 1971, pp. 25–36. 31. T. Croucher, Heat Treating, December 1990, pp. 17–19. 32. W. H. Naylor, Met. Progr., vol. 92, no. 6, 1967, pp. 70–73. 33. P. E. Cary, Met. Progr., vol. 89, no. 2, 1966, pp. 90–92. 34. R. W. Foreman, Paper presented at ASM Short Course on Quenching Media, American Society for Metals, Metals Park, Ohio, November 1985. 35. Houghton on Quenching, products brochure, E. F. Houghton & Co., Inc., England, 1994. 36. S. O. Segerberg, Heat Treat., December 1988, pp. 30–33. 37. V. Srimongkolkul, Ind. Heat., September 1998, pp. 81–84. 38. J. Hasson, Ind. Heat., November 1995, pp. 45–46. 39. R. K. Evans, Met. Mater., vol. 12, 1978, p. 28. 40. R. F. Kern and M. E. Suess, Steel Selection, Wiley, New York, 1979. 41. K. Funfani and G. E. Totten, in Proceedings: Second International Conference on Quenching and Control of Distortion, 4–7 Nov. 1996, eds. G. E. Totten et al., ASM International, Materials Park, Ohio, 1996, pp. 3–15. 42. J. Igarashi, Jpn. J. Tribol., vol. 35, 1990, pp. 1095–1105. 43. G. E. Totten, G. M. Webster, L. M. Jarvis, S. H. Kang, and S. W. Han, in 17th ASM Heat Treating Society Conference Proceedings, 15–18 Sept. 1997, eds. D. L. Milam et al., ASM International, Materials Park, Ohio, 1997, pp. 443–448. 44. J. A. Hasson, Ind. Heat., September 1981, pp. 21–23. 45. V. Srimongkolkul, Heat Treat., December 1990, pp. 27–28. 46. H. M. Tensi, A. Stich, and G. E. Totten, Steel Heat Treatment Handbook, Marcel Dekker, New York, 1997, pp. 157–249. 47. Z. H. Ou, G. E. Totten, G. M. Webster, and Y. H. Sun, in Proceedings: Second International Conference on Quenching and Control of Distortion, 4–7 Nov. 1996, eds. G. E. Totten et al., ASM International, Materials Park, Ohio, 1996, pp. 517–523. 48. K. H. Kopietz, Heat Treat., vol. 16, no. 9, 1984, pp. 20–26. 49. D. Moore, Heat Treat. Met., vol. 26, no. 3, 1999, pp. 68–71. 50. R. W. Rhines and E. R. Mueller, Met. Progr., vol. 122, no. 4, 1982, pp. 33–39. 51. G. E. Totten, G. M. Webster, and C. E. Bates, in 17th ASM Heat Treating Society Conference Proceedings, 15–18 Sept. 1997, eds. D. L. Milam et al., ASM International, Materials Park, Ohio, 1997, pp. 247–255. 52. R. R. Blackwood, L. M. Jarvis, G. E. Totten, G. M. Webster, and T. Narumi, Metal Heat Treat., May/June 1996, pp. 28–31.

HARDENING AND HARDENABILITY

13.149

53. L. S. Zarkhin, S. V. Sheberstov, N. V. Panfilovich, and L. I. Manevich, Russ. Chem. Rev., vol. 122, no. 4, 1982, pp. 33–39. 54. R. T. von Bergen, Heat Treat. Met., vol. 2, 1991, pp. 37–42. 55. S. Segerberg, 4th Int. Cong. Heat Treat. Mater., Berlin, 3–7 June 1985, vol. II, pp. 1252–1265. 56. T. Hibi, Netsu Shori, vol. 25, no. 1, 1985, pp. 46–50. 57. D. Diaz, H. Garcia, and B. Bautista, Ind. Heat., June 1993, pp. 43–45. 58. N. Kobayashi, Netsu Shori, vol. 25, no. 1, 1985, pp. 51–54. 59. G. E. Totten, R. R. Blackwood, and L. M. Jarvis, Heat Treating, March 1991, pp. 17–18. 60. R. D. Howard and G. E. Totten, Metal Heat Treat., September/October 1994, pp. 22–24. 61. Heat Treat., October 1983, pp. 40–41. 62. E. H. Burgdorf, Ind. Heat., October 1981, pp. 18–25. 63. Cassels Salts for Heat Treatment and Case Hardening, Birmingham, England. 64. Heat Treat., vol. 11, no. 3, 1979, pp. 22–28; vol. 11, no. 4, 1979, pp. 42–44. 65. R. W. Foreman, Ind. Heat., March 1993, pp. 41–47. 66. Liquid Salt Baths, product bulletin, E. F. Houghton & Co., Valley Forge, Pa., December 1981. 67. G. P. Dubal, Proceedings: 18th Conference, 12–15 Oct. 1998, Heat Treating, eds. R. A. Wallis and H. W. Walton, ASM International, Materials Park, Ohio, pp. 192–198; Adv. Mats. & Process., December 1999, pp. H23–H28. 68. G. P. Dubal, M. T. Ives, A. G. Meszaros, and J. Recker, 1st International Automotive Heat Treating Conference, 13–15 July 1998, eds. R. Colas, K. Funatani, and C. A. Stickels, ASM International, Materials Park, Ohio, pp. 90–95. 69. Metall. Met. Form, vol. 38, 1971, pp. 322–323. 70. R. Wilson, Metallurgy and Heat Treatment of Tool Steels, McGraw-Hill, London, 1975. 71. R. W. Foreman, Heat Treat., vol. 12, no. 10, 1980, pp. 26–29. 72. Heat Treat., vol. 15, no. 10, 1985, pp. 22–32. 73. B. Harrison, Metallurgia, vol. 45, no. 3, 1978, pp. 145–146. 74. D. Grieve, Metall. Mater. Technol., vol. 7, no. 8, 1975, pp. 397–403. 75. J. A. Schliessman, Meta. Treat., vol. 22, no. 6, 1971, pp. 3–8. 76. A. Creal, Heat Treat., vol. 18, no. 10, 1986, pp. 26–28. 77. R. W. Foreman, Paper presented at ASM National Heat Treating Conference, Chicago, September 1988. 78. L. Rosseau, Metallurgia, vol. 49, 1954, pp. 27–33. 79. B. Liscic, in Quenching and Carburizing, 3d International Seminar, International Federation for Heat Treatment of Surface Engineering, Melbourne, Australia, 1991; The Institute of Materials, 1993, pp. 1–32. 80. J. A. Lincoln and A. Keogh, U.S. Patent 4,431,464, February 1984. 81. C. Skidmore, Heat Treat. Met., vol. 2, 1986, pp. 34–38. 82. G. Wahl, Carburizing—Processing and Performance, ASM International, Materials Park, Ohio, 1989, pp. 41–56. 83. T. W. Ruffle and E. R. Byrnes, Heat Treat. Met., vol. 6, no. 4, 1979, pp. 81–87. 84. J. E. Pritchard, G. Nurnberg, and M. Shoukri, Heat Treat Met., vol. 23, no. 4, 1996, pp. 79–83. 85. P. F. Stratton, N. Saxena, and R. Jain, Heat Treat. Met., vol. 24, no. 3, 1997, pp. 60–63. 86. W. R. Jones, Heat Treat., September 1985, pp. 34–35. 87. J. M. Neidermann, Heat Treat., vol. 16, no. 12, 1984, pp. 20–23.

13.150

CHAPTER THIRTEEN

88. S. Segerberg and E. Troell, Heat Treat. Met., vol. 24, no. 1, 1997, pp. 21–24. 89. E. Edenhofer, Heat Treat. Met., vol. 26, no. 1, 1999, pp. 1–5. 90. J. G. Conybear, Adv. Mats. & Processes, Feb. 2, 1993, pp. 20–21. 91. R. E. Andrews, J. Grooves, and M. H. Jacobs, in Heat Treatment ‘84, The Metals Society, London, 1984, pp. 41.1–41.8. 92. J. W. Bouwman, Metallurgia, vol. 52, no. 2, 1985, pp. 41–45. 93. J. Kowaleski and J. Olejnik, Ind. Heat., October 1998, pp. 39–44. 94. R. Hill, Jr., High Pressure Gas Quenching Typical Oil Hardening Grades of Steel, SME 1998 Technical Paper, CM98-207, pp. 1–18. 95. R. B. Dixon, Heat Treat. Met., vol. 24, no. 2, 1997, pp. 37–42. 96. Product brochure, Fluidtherm Technology Ltd., Madras, India. 97. M. A. Delano and J. Van den Sype, Heat Treat., December 1988, pp. 34–37. 98. R. W. Reynoldson, Heat Treatment in Fluidized Bed Furnaces, ASM International, Materials Park, Ohio, 1993, p. 47. 99. Z. Kulin and H. Genlian, Proc. 4th Int. Cong. Heat Treatment of Materials, vol. 2, 3–7 June 1985, Berlin, pp. 1293–1320. 100. A. J. Hick, Heat Treat. Met., vol. 21, no. 3, 1994, pp. 53–64. 101. G. Beck, in Heat and Mass Transfer in Metallurgical Systems, eds. D. B. Spalding and N. H. Afgan, Hemisphere Publishing, 1981, pp. 509–525. 102. G. Li, Proc. 4th Ann. Conf. Heat Treat., Nanjing, 25–31 May 1987, Chinese Mechanical Engineering Society, pp. 171–175. 103. A. J. Hick, Heat Treat. Met., vol. 26, no. 3, 1999, pp. 53–62. 104. L. Marshall and J. Canner, Ind. Heat., June 1994, pp. 51–52. 105. N. I. Kobasko, M. A. Aronov, G. E. Totten, and A. V. Sverdin, Proc. 18th Conference: Heat Treating 1998, eds. R. A. Wallis and H. W. Walton, ASM International, Materials Park, Ohio, 1998, pp. 616–621. 106. G. F. Malloy, Hardening of Steel, Metals Engineering Institute, ASM, Metals Park, Ohio. 107. T. D. Atterbury, Metall. Met. Form., August 1971, pp. 210–215. 108. C. E. Bates and G. E. Totten, Heat Treat. Met., vol. 19, no. 2, 1992, pp. 45–48. 109. B. Kovacs, in ASM Handbook, vol. 4: Heat Treating, 10th ed., ASM International, Materials Park, Ohio, 1991, pp. 670–681. 110. K. B. Rundman, in ASM Handbook, vol. 4: Heat Treating, 10th ed., ASM International, Materials Park, Ohio, 1991, pp. 682–696. 110a. N. Shimizu and I. Tamura, Trans. ISIJ, vol. 16, 1976, pp. 655–663; Trans. ISIJ, vol. 17, 1977, pp. 469–476. 110b. N. Shimizu, Ph.D. thesis, Kyoto University, 1985 (in Japanese). 110c. K. Arimoto, D. Huang, D. Lambert, and W. T. Wu, Heat Treating Conference and Exposition, Oct. 9–12, 2000, St. Louis, Missouri. 110d. B. Liscic, G. Grubisic, and G. E. Totten, Proc. 2d International Conference on Quenching and Control of Distortion, Cleveland, Ohio, 1996, pp. 47–54. 110e. B. Liscic and G. E. Totten, Advanced Materials and Processes, vol. 9, 1997, pp. 180–184. 111. P. E. Reynolds, Heat Treat. Met., vol. 17, no. 3, 1990, pp. 69–72. 112. N. I. Kobasko, ASM 18th International Conference: 12–15 Oct. 1998, eds. R. A. Wallis and H. W. Walton, ASM International, Materials Park, Ohio, 1998, pp. 613–616; Adv. Mats. & Process., December 1999, pp. H31–H33. 112a. M. A. Aronov, N. I. Kobasko, J. F. Wallace, D. Schwam, and J. A. Powell, Ind. Heat., April 1999, pp. 59–63. 113. R. F. Kern, Heat Treat., vol. 18, no. 9, 1986, pp. 19–23.

HARDENING AND HARDENABILITY

13.151

114. H. Webster and W. J. Laird, Jr., in ASM Handbook, vol. 4: Heat Treating, 10th ed., ASM International, Materials Park, Ohio, 1991, pp. 137–151. 115. J. L. Yarne, Met. Progr., vol. 84, no. 4, 1963, p. 105. 116. W. D. Lankford and H. E. McGannon, eds., Making, Shaping, and Treating of Steel, 10th ed., USS Corporation, Pittsburgh, 1985. 117. L. F. Spencer, Met. Treat., vol. 11, no. 3, 1960, p. 55. 118. E. C. Bain and E. S. Davenport, Trans. ASM, 1930, p. 289. 119. J. R. Keough, W. J. Laird, Jr., and A. D. Godding, ASM Handbook, vol. 4: Heat Treating, 10th ed., ASM International, Materials Park, Ohio, 1991, pp. 152–163. 120. R. L. Siffredini, Heat Treat., vol. 12, no. 1, 1980, pp. 14–19. 121. E. C. Harwood, Metall. Met. Form., vol. 31, February 1964, pp. 82–83. 122. R. Creal, Heat Treat., vol. 18, no. 8, 1986, pp. 24–27. 123. B. J. Hart, Heat Treat., vol. 15, no. 10, 1983, pp. 36–38. 124. B. Liscic, in Steel Heat Treatment Handbook, Marcel Dekker, New York, 1997, pp. 527–662. 125. Metallurgia, vol. 47, no. 3, 1980, pp. 136–138. 126. B. J. Waterhouse, Met. Progr., vol. 106, no. 3, 1974, p. 71. 127. D. Nicholson, S. Ruhamann, and R. J. Wingrove, Heat Treatment of Metals, Special Report 95, Iron and Steel Institute, London, 1966, p. 180. 128. V. K. Chandhok, A. Kasak, and J. P. Hirth, Trans. ASM, vol. 59, 1966, p. 288. 129. Ductile Iron Data for Design Engineers, QIT-Fer et Titane Inc., Chicago, Illinois, 1990. 130. R. A. Harding, Met. Mater., vol. 2, no. 2, 1986, pp. 65–72. 131. R. A. Blackmore and R. A. Harding, J. Heat Treat., vol. 3, no. 4, 1984, pp. 310–325. 132. R. B. Gundlach and J. F. Janowak, Met. Progr., vol. 128, no. 2, 1985, pp. 19–25. 133. V. K. Sharma, J. Heat Treat., vol. 3, no. 4, 1984, pp. 326–334. 134. J. A. Lincoln, Heat Treat., vol. 16, no. 12, 1984, pp. 30–34. 134a. R. Elliott, Heat Treat. Met., vol. 24, no. 3, 1997, pp. 55–59. 135. J. Mallia and M. Grech, Mats. Sc. & Tech, vol. 13, May 1997, pp. 408–414. 136. B. T. Sims and R. Elliott, Mats. Sc. & Tech., vol. 14, February 1998, pp. 89–96. 137. The New Bench Mark Material, Applied Process, Inc. Bulletin, 1999. 137a. M. Bahmani, R. Elliott, and N. Varahram, J. Mats. Sc., vol. 32, 1997, pp. 5383–5388. 138. H. Bayati and R. Elliott, Mats. Sc. & Tech., vol. 13, April 1997, pp. 319–326. 139. R. C. Voigt and C. R. Loper, Jr., J. Heat Treat., vol. 3, no. 4, 1984, pp. 291–309. 140. B. V. Kovacs, Sr., Modern Casting, March 1990, pp. 38–41; J. R. Keough, Foundry, Manage., and Tech., October/November 1995. 141. J. R. Laub, Heat Treat., March 1992, pp. 18–23. 141a. Y. S. Lerner and G. R. Kingsbury, Jn. Mats. Engrg. and Performance, vol. 7, no. 1, 1998, pp. 48–52. 142. A. Owhadi, J. Hedjozi, and P. Davami, Mats. Sc. & Tech., vol. 14, March 1998, pp. 245–250. 143. K. Boiko, Heat Treat., vol. 17, no. 9, 1985, pp. 22–24. 144. J. Race and L. Stott, Heat Treat. Met., vol. 18, no. 4, 1991, pp. 105–109. 145. J. Hemanth, Mats. Sc. & Tech., vol. 15, August 1999, pp. 878–884. 146. E. Dorazil, High Strength Austempered Ductile Cast Iron, Ellis Harwood, New York, 1991. 147. R. R. Blackwood and L. M. Jarvis, Ind. Heat., March 1991, pp. 28–31.

13.152

CHAPTER THIRTEEN

147a. H. W. Rayson, Constitution and Properties of Steels, vol. ed. F. B. Pickering, VCH, Weinheim, 1992, pp. 583–640. 148. T. Kunitake and S. Sugisawa, The Sumitomo Search 5, May 16, 1971. 149. H. Ohtani, in Materials Science and Technology, vol. 7: Constitution and Properties of Steel, vol. ed. F. B. Pickering, VCH, Weinheim, 1992, pp. 147–181. 150. R. Arpi, Jernkont. Ann., no. 115, 1931, p. 75. 151. E. C. Bain and H. W. Paxton, Alloying Elements in Steel, 2d ed., ASM, Metals Park, Ohio, 1966. 152. R. E. Reed-Hill and G. J. Abbaschian, Physical Metallurgy Principles, PWS-Kent Publishing, Boston, 1992. 153. Metals Handbook, vol. 1, 9th ed., ASM, Metals Park, Ohio, 1978. 154. M. A. Grossman and E. C. Bain, Principles of Heat Treatment, 5th ed., ASM, Metals Park, Ohio, 1964. 155. ASTM A255–99, ASTM, Pittsburgh, Pa., 1999. 156. Metals Handbook, vol. 1, 8th ed., ASM, Metals Park, Ohio, 1961. 157. R. M. Grange and T. M. Garvey, Trans. ASM, vol. 37, 1946, pp. 136–174; R. Grange and T. Mitchel, Trans. ASM, vol. 53, 1961, pp. 157–185. 158. C. J. McMahon, Jr., Met. Trans., vol. 11A, 1980, pp. 531–535. 159. J. M. Tartaglia and G. T. Eldis, Met. Trans., vol. 15A, 1984, pp. 1573–1583. 160. C. Skena, T. Prucher, R. Czarnek, and J. M. Jo, The Int. Powder Met., vol. 33, no. 7, 1997, pp. 25–35. 161. A. Jones and P. E. Evans, Heat Treat. Met., vol. 20, no. 4, 1993, pp. 99–100. 162. W. Hewitt, Heat Treat. Met., vol. 8, no. 2, 1981, pp. 33–38. 163. C. T. Kunze, in Hardenability Concepts with Applications to Steel, eds. D. V. Doane and J. S. Kirkaldy, Proceedings Symposium 1977, TMS-AIME, Warrendale, Pa., 1978, pp. 290–305. 164. G. F. Malloy, P. R. Slimomon, and P. Kvaale, Met. Trans., vol. 4, 1973, pp. 2279–2289. 165. O. M. Askelsen, O. Grong, and P. E. Kvaale, Met. Trans., vol. 17A, 1986, pp. 1529–1536. 166. T. Lund, Scand. J. Met., 1990, pp. 227–235. 167. R. Kumar, Physical Metallurgy of Iron and Steel, Asia Publishing House, Bombay, 1968. 168. K. E. Thelning, Steel and Its Heat Treatment, Butterworths, London, 1984. 169. H. Burrier, Jr., in ASM Handbook, vol. 1, 10th ed., ASM International, Materials Park, Ohio, 1990, pp. 464–484. 170. C. A. Siebert, D. V. Doane, and D. H. Breen, The Hardenability of Steels, ASM, Metals Park, Ohio, 1977. 171. R. F. Mehl and W. C. Hagel, Prog. Met. Phys., vol. 6, 1956, p. 74. 172. W. F. Smith, Structure and Properties of Engineering Materials, McGraw-Hill, New York, 1981. 173. Methods of Determining Hardenability of Steels, SAE J406, February 1995, 2000 SAE Handbook, vol. 1, Society of Automotive Engineers, Warrendale, Pa., pp. 1.25–1.46. 174. G. Krauss, Steels: Heat Treatment and Processing Principles, ASM International, Materials Park, Ohio, 1990. 175. B. Liscic, in Steel Heat Treatment Handbook, Marcel Dekker, New York, 1997, pp. 93–156. 176. A. F. DeRetana and D. V. Doane, Met. Prog., vol. 100, 1971, p. 65. 177. A. Moser and A. Legat, Harterei-Techn. Mitt., vol. 24, no. 2, 1969, pp. 100–105. 178. C. F. Jatczak, Met. Trans., vol. 4, 1973, pp. 2267–2277.

HARDENING AND HARDENABILITY

13.153

179. I. R. Kramer, S. Siegel, and J. G. Brooks, Trans. AIME, vol. 167, 1946, p. 670. 180. R. A. Grange, Met. Trans., vol. 4, 1973, pp. 2231–2244. 181. D. V. Doane, in Hardenability Concepts with Applications to Steel, Proceedings Symposium 1977, TMS-AIME, Warrendale, Pa., 1978, pp. 351–378. 182. W. E. Jominy and A. L. Boegehold, Trans. ASM, vol. 26, 1938, p. 574. 183. R. F. Kern and M. E. Suess, Steel Selection, Wiley-Interscience, New York, 1979. 184. T. Brown, in Hardenability Concepts with Applications to Steel, Proceedings Symposium 1977, TMS-AIME, Warrendale, Pa., 1978, pp. 273–288. 185. S. I. Arkhangel’skii, E. S. Miroshnik, and I. V. Tikhonova, Ind. Laboratory, vol. 11, 1991, pp. 1349–1351 (English translation). 186. L. H. Van Vlack, Materials Science for Engineers, Addison-Wesley, Reading, Mass., 1971. 187. S. B. Lasday, Ind. Heat., 1992, pp. 25–27. 188. A Rose and L. Rademacher, Weiterentwicklung des stirnabschreckversuches zur Prufung der Hartbarkeit von tiefer einhartenden stahlen, Stahl Eisen, vol. 76, no. 23, 1956, pp. 1570–1573 (in German). 189. C. F. Jatczak, Trans. ASM, vol. 58, 1965, p. 195. 189a. B. M. Kapadia et al., Trans. AIME, vol. 242, 1969, p. 1689. 190. D. T. Llewellyn and W. T. Cook, Metals Technol., December 1974, p. 517. 190a. T. T. Llewellyn, Ironmaking and Steelmaking, vol. 20, no. 5, 1993, pp. 338–343. 191. E. M. Grinberg, in 21st Century Steel Industry 1994, pp. 170–173. 192. J. Field, Met. Prog., vol. 43, no. 3, 1943, p. 402. 193. L. J. Boyd and J. Field, Contributions to the Metallurgy of Steel, no. 12, American Iron and Steel Institute, New York, 1946. 194. E. Just, Met. Prog., November 1969, pp. 87–88. 195. J. S. Kirkaldy, G. Pazionis, and S. E. Feldman, Heat Treatment, 1976, The Metals Society, London, 1976. 196. J. S. Kirkaldy, Met. Trans., vol. 4, 1973, pp. 2327–2333. 197. J. S. Kirkaldy, in Hardenability Concepts with Applications to Steel, Proceedings Symposium 1977, eds. D. V. Doane and J. S. Kirkaldy, TMS-AIME, Warrendale, Pa., 1978, p. 491. 198. D. J. Keith, J. T. Sponzilli, V. K. Sharma, and C. H. Walter, in Hardenability Concepts with Applications to Steel, Proceedings Symposium 1977, eds. D. V. Doane and J. S. Kirkaldy, TMS-AIME, Warrendale, Pa., 1978, pp. 493–516. 199. P. Maynier, B. Jungmann, and J. Dollet, in Hardenability Concepts with Applications to Steel, Proceedings Symposium 1977, eds. D. V. Doane and J. S. Kirkaldy, TMS-AIME, Warrendale, Pa., 1978, pp. 518–544. 200. M. Larsson, B. Jansson, R. Blom, and A. Melander, Scand. J. of Metall., vol. 19, 1990, pp. 51–63. 201. S. E. Feldman, in Hardenability Concepts with Applications to Steel, Proceedings Symposium 1977, eds. D. V. Doane and J. S. Kirkaldy, TMS-AIME, Warrendale, Pa., 1978, pp. 546–567. 202. J. S. Kirkaldy and S. E. Feldman, J. Heat Treat., vol. 7, 1989, pp. 57–64. 203. J. S. Kirkaldy, in Metals Handbook, vol. 4, 10th ed., ASM International, Materials Park, Ohio, 1991, pp. 20–34. 204. E. R. Kuch, in ASM Handbook, vol. 1, 10th ed., ASM International, Materials Park, Ohio, 1990, pp. 451–463. 205. ASTM A304-96, ASTM, Pittsburgh, Pa. 206. ASTM A400-1995, ASTM, Pittsburgh, Pa.

CHAPTER 14

TEMPERING

14.1 INTRODUCTION An attractive combination of strength, toughness, and ductility for a given application can be produced in steel when fully “as-quenched” martensite is reheated to a suitable temperature below the lower critical temperature A1. Such a heat treatment, following quench-hardening, is termed tempering. Tempering involves redistribution of carbon atoms; precipitation of fine, coherent carbides; decomposition of retained austenite; coarsening and spheroidization of cementite (associated with recovery and recrystallization of the matrix); and precipitation of alloy carbides, in the presence of strong carbide-forming elements, from a supersturated as-quenched martensite.1 This ability of tempering to produce a wide range of mechanical properties, relieve quenching and/or other residual stresses, and ensure dimensional stability was the principal reason for the lasting interest of the metallurgist in the structures of iron-carbon martensites and their various decomposition products. Crystallographic and kinetic theories have been developed to account for the formation of decomposition products of martensite, and numerous studies have identified several stages of decomposition during the tempering of martensite and the effect of undesirable decomposition products in causing various types of embrittling phenomena. In this chapter we start our discussion with the structural and mechanical property changes involved in tempering and the role of alloying elements upon the early-stage behavior and that of precipitation and growth of carbides. Then we present secondary hardening, tempering parameter, strengthening mechanisms of tempered martensite, and various types of embrittlement phenomena such as tempered martensite embrittlement, temper embrittlement, secondary hardening embrittlement, aluminum nitride embrittlement, hydrogen damage, and metalinduced embrittlement associated with tempering. Finally, we describe maraging steels in depth.

14.2 STRUCTURAL CHANGES ON TEMPERING As-quenched steels are usually mixtures of martensite and retained austenite, with the former constituent in abundance. Both these constituents are thermodynamically unstable and slowly transform, at least in part, if left at ambient temperature; the retained austenite (gR) may convert to martensite (a¢), and the martensite (a¢) undergoes transformations which are described here. 14.1

14.2

CHAPTER FOURTEEN

Note at this point that as-quenched low-carbon steels at room temperature may be considered to have undergone the pre-precipitation (tempering) process during quenching because of its high Ms temperature and the high mobility of interstitial atoms. This is called autotempering and is evidenced by the precipitation of fine carbides or the carbon atom clustering, which precedes such precipitation. In general, structural changes occurring during conventional tempering of martensitic steel can be divided into five distinct stages; but the temperature ranges are approximate, resulting in the overlapping stages of tempering and concurrent development of other transformation products. This fact, together with the occurrence of reactions on a very fine scale, produces complex structures of the tempered martensite. The first stage (T1), occurring in the temperature range of 100 to 250°C (212 to 482°F), corresponds to the decomposition of supersaturated martensite to transition carbide, epsilon(e)-carbide [or eta(h)-carbide, as described below], and lowcarbon martensite (a ≤). The second stage (T2), occurring between 200 and 300°C (392 and 572°F), represents the decomposition of retained austenite to bainite. The third stage (T3) occurs in the temperature range of 250 to 350°C (482 to 662°F) and constitutes the transformation of the reaction products of first and second stages into ferrite and [chi (c) and/or] cementite [theta (q)] constituents. The fourth stage (T4), occurring above 350°C, comprises the growth and spheroidization of cementite. The fifth stage (T5) (see Sec. 14.5) holds primarily to alloy steels. In this stage, intermetallic precipitates and alloy carbides are formed.2,3

14.2.1 Aging Reaction Stage Recent electrical resistivity,4 x-ray diffraction,5,6 transmission electron microscopy (TEM),7 atom probe field ion microscopy (APFIM),8 Mössbauer spectroscopy, and 13 C NMR studies have resulted in the incorporation of an additional stage prior to the conventional first stage of tempering;9–13 this is called a preliminary stage, aging reaction stage, or pre-precipitation process, and it refers to a phenomenon that precedes T1, for example, during storage of the martensite specimen at room temperature.4,7,9 The kinetics of this stage are such that aging is already accomplished in most “as-quenched” commercial steels with Ms temperature above room temperature.10 These studies have recognized the development of four steps of structural arrangements of carbon atoms during the aging process:10–12 (1) the formation of small clusters (of carbon doublets) into isolated carbon multiplets along a ¢ directions; (2) coarsening by ordering of these carbon multiplets in (100)a ¢ monolayers; (3) thickening along [100]a¢ into a multilayer structure, called the extended multiplet; and (4) appearance of a superperiod of 12 lattice parameters along [001]a ¢, associated with antiphase domains by long-range interaction. Table 14.1 shows a sequence of precipitation of Fe9C4 (e- or h-) carbide during the four-step aging of Fe-C martensite.12 The TEM was carried out by Nagakura et al.,7 Genin et al.,11–13 and Ohmori and Tamura14 on high-carbon Fe-C alloys, while other studies were carried out by Taylor et al.15 and other workers4–6,8 on Fe-Ni-C alloys with subzero Ms temperatures. A1 Stage. During the first substage of aging reaction A1 of ferrous martensites, a clustering reaction of carbon atoms occurs in the c-oriented octahedral sites, associated with the carbon trapping by lattice defects (or Bain correspondence). This is in good agreement with the foregoing experimental techniques used.

TABLE 14.1 Sequence of Precipitation of Fe9C4 e- or h-Carbide during the Aging of Fe-C Martensite12 Steps of aging Short-range order: clustering of C atoms into isolated multiplets

14.3

Mössbauer spectroscopy

Clustering

13

Clustering

Electron diffraction

Thickening along [100]a’ into the multilayer B2 monoclinic structure

Long-range interaction: 12 ao superperiod along [001]a ¢ and antiphase domains

Clustering and coarsening

Resistivity C NMR

Ordering of multiplets in (100)a’ monolayers

Coarsening into ordered multilayer Ordering

Thickening

Diffuse Scattering Streaks

Satellites

TEM APFIM X-ray analysis

Modulations of composition Clustering (Atomic static displacement of Fe in matrix) Carbon depletion in the matrix

Each experimental technique is sensitive only to some specific aspects of the process and can reveal only some of the steps. Courtesy of the Metallurgical Society, Warrendale, Pa.

C phase or k phase

14.4

CHAPTER FOURTEEN

The electrical resistivity measurements of Fe-Ni-C martensites containing 18 to 24% Ni and 0.003 to 0.62% C were first made following quenching in liquid nitrogen to form virgin martensite and then after up-quenching to temperatures between 80 and 350°C for tempering. Figure 14.1a shows a schematic resistivity versus aging time/temperature curve with the structural changes and regimes produced during the aging and first-stage tempering of virgin Fe-Ni-C martensite.4 At subambient temperatures (in regime I), the first small drop in resistivity is attributed to isothermal transformation of a small amount of retained austenite to martensite; this may be important for “cryogenically” treated steels. In regime II, resistivity increase to peak and its subsequent decrease are attributed to the martensite aging where the resistivity increase to peak value corresponds to the carbon atom rearrangement prior to transition carbide formation. An analysis of integrated intensity changes of Fe-Ni-C martensite (002) x-ray diffraction peak profiles during subambient aging stage (Fig. 14.2) has shown that clusters are associated with two to four carbon atoms and leads to the formation of isolated multiplets. This figure also shows the dramatic changes of martensite (002) peak intensity for early and later stages of tempering up to 450°C. Mössbauer spectroscopy, 13C NMR, and atom probe analysis also support the development of carbon atom clustering (and the depletion of carbon in the martensite matrix). Evidence of carbon atom clustering has been shown in TEM by the presence of diffuse intensity spikes around the fundamental electron diffraction spots. Nagakura et al.7 have demonstrated from dark-field contrasts that the clusters were about 1 nm in size, which agreed well with the results obtained from x-ray diffraction studies.5 On the basis of APFIM study of Fe-Ni-C alloy, the formation of a≤-Fe16C2 structure (similar to a≤-Fe16N2 in Fe-N alloy) has been suggested as the pre-precipitation stage.16 However, according to x-ray diffraction experiments using high-intensity synchrotron radiation, a≤-like structure does not seem to be formed either upon aging at room temperature or upon tempering at higher temperatures.17 Taylor and his coworkers have proposed spinodal decomposition mechanism as the preprecipitation stage.16 The Taylor-Cohen review paper18 discusses the aging in terms of development of structural modulations (possibly spinodal decomposition mechanism). Figure 14.1b presents a pictorial summary of the aging of initially virgin martensites, as developed in this review, showing increasing amplitude in A1 and increasing wavelength in A2 (A3).18 A2 Stage. The second stage of aging, A2, occurring between subzero temperature and about 70°C, involves a fine modulated tweed microstructure containing carbon-deficient regions and carbon cluster regions (varying from 0.2 to 11 at%).4,7,18a,19 Figure 14.3a shows the dark-field micrograph of a modulated A2 structure in Fe-1.31%C martensite tempered for 1 hr at 70°C, taken with the (002) fundamental spot and its satellites. This illustrates the fringes parallel to the (102) and ( 102) planes. Figure 14.3b is a high-resolution electron micrograph of an Fe-1.39%C martensite tempered for 1 hr at 70°C, taken with a beam parallel to [0 10], which illustrates that the vertical fringes are ( 101) lattice fringes and that the white patches are due to the modulated A2 structure.7 These observations suggest that the carbon atom clusters about 1 nm in size concentrate at random in the {102} planes spaced periodically at about 1 nm apart with the intervening carbon-depleted regions. Such a modulated structure formation was also provided by Tanaka and Shimizu20 and Taylor et al.15,16,18 The x-ray diffraction and electrical resistivity measurements are unable to distinguish the A2 structure; the former, however, shows peak shifts in the tempering temperature range corresponding to the formation of A2 structures (Fig. 14.2),

FIGURE 14.1 (a) Schematic resistivity versus aging time/temperature curve identifying the regimes which occur during the aging and tempering stages of initially virgin martensite.4 (b) Summary of the structural changes that occur during aging and tempering stages of initially virgin martensite, with respect to the trends in electrical resistivity r and flow stress s. Aging is identified with spinodal decomposition of the martensite, which produces the regime II resistivity peak and an increase in strength. The respective increases in amplitude Dc and wavelength l of the carbon-concentration modulations that develop coherently are designated as substages A1 and A2; possible secondary ordering of the carbon atoms constitutes substage A3. Alternatively, substages A1 and A2 have also been associated with the formation of randomly distributed carbon clusters and subsequent evolution of the modulated structure by stress-induced alignment of these clusters. Carbide precipitation—e-carbide in Stage T1 and cementite in Stage T3—marks the completion of aging and the subsequent sequence of tempering. Mechanical strength continues through a maximum near the onset of T1, while resistivity continues to decrease in Regime III. Structural changes are illustrated by electron micrographs obtained during the aging of initially virgin Fe-Ni-C martensites (magnifications for these micrographs are all the same).18 (Source: K. A. Taylor, “Aging Phenomena in Ferrous Martensites,” ScD thesis, Massachusetts Institute of Technology, Cambridge, 1985.) (Courtesy of K. A. Taylor.)

14.5

(b)

FIGURE 14.1 (Continued) (a) Schematic resistivity versus aging time/temperature curve identifying the regimes which occur during the aging and tempering stages of initially virgin martensite.4 (b) Summary of the structural changes that occur during aging and tempering stages of initially virgin martensite, with respect to the trends in electrical resistivity r and flow stress s. Aging is identified with spinodal decomposition of the martensite, which produces the regime II resistivity peak and an increase in strength. The respective increases in amplitude Dc and wavelength l of the carbon-concentration modulations that develop coherently are designated as substages A1 and A2; possible secondary ordering of the carbon atoms constitutes substage A3. Alternatively, substages A1 and A2 have also been associated with the formation of randomly distributed carbon clusters and subsequent evolution of the modulated structure by stress-induced alignment of these clusters. Carbide precipitation—e-carbide in Stage T1 and cementite in Stage T3—marks the completion of aging and the subsequent sequence of tempering. Mechanical strength continues through a maximum near the onset of T1, while resistivity continues to decrease in Regime III. Structural changes are illustrated by electron micrographs obtained during the aging of initially virgin FeNi-C martensites (magnifications for these micrographs are all the same).18 (Source: K. A. Taylor, “Aging Phenomena in Ferrous Martensites,” ScD thesis, Massachusetts Institute of Technology, Cambridge, 1985.) (Courtesy of K. A. Taylor.)

14.6

TEMPERING

14.7

FIGURE 14.2 The (002) x-ray diffraction peak profiles for martensite in an 18Ni-0.98C-Fe alloy at successively higher temperatures, as indicated for times between 1 and 2.5 hr. (Reprinted from P. C. Chen and P. G. Winchell, Met. Trans, vol. 11A, 1980, p. 1333; reprinted from a publication of The Metallurgical Society, Warrendale, Pa.)

thereby illustrating a gradual decrease in tetragonality of the martensite, whereas the latter exhibits carbon depletion in the matrix.12 The entire aging process is diffusion-controlled and produces a depletion of carbon in the martensite lattice with the attendant conversion to low tetragonal martensite. Carbon clustering, the formation of isolated multiplets, and the development of modulated structure are insensitive to martensite morphology (provided the effect of “autotempering” during the quenching of alloys with high Ms tempertature is ignored).20a A3 Stage. During the third substage of aging reaction A3, an ordering reaction, perhaps intimately related to the prior clustering, may occur,18 leading to tetragonal Fe4C carbide particles. The A3 structure may be identified by superstructure spots or satellites appearing in the electron diffraction pattern of martensite tempered at about 60 to 80°C. This structure is also exhibited by the carbon-free regions between the antiphase domains, called k phase, C-phase, or low-carbon martensite, which is perhaps coherently strained martensite adjacent to the transition carbide particles.12 However, the recent TEM and APFIM results indicate that A1 and A2 may be the early and advanced stages, respectively, in the same overall process of spinodal decomposition. Furthermore, if A3 is associated with spinodal ordering, then A1, A2, and A3 stages could each represent different aspects of a single decom-

FIGURE 14.3 (a) Dark-field electron micrograph of an Fe-1.31C alloy martensite tempered for 1 hr at 70°C taken with the (002) fundamental spots and its satellites showing modulated A2 structure. (b) High-resolution electron micrograph of an Fe-1.39C martensite tempered for 1 hr at 70°C taken with beam parallel to [01¯0]. The vertical fringes are (1¯01) lattice fringes, and the white patches are due to the modulated structure.7 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

14.8

TEMPERING

14.9

position process.10 However, other workers19 have discounted the A3 stage and hence the A3 structure.21

14.2.2 The First Stage of Tempering (T1) The tempering reactions of martensite formed in low-carbon steels (390

>735

Very rapid heating; special fixturing is required (high density)

Reprinted by permission of ASM International, Materials Park, Ohio.

continuous operation of nearly similar parts. Large volumes of air at a controlled temperature are passed at high velocities over the workload. For temperatures of 550 to 750°C (1022 to 1382°F), alternatively radiant heating may be employed because of the greater transfer of radiant heat (or efficiency). Oil baths are used in the temperature range of 120 to 250°C (250 to 480°F) and are preferred where long exposure time is desired. (Typical tempering time varies between 0.5 and 4 hr.) However, it is necessary to utilize stirring for temperature uniformity and longer oil life, use special ventilation for fume extraction, and avoid overheating for preventing fire hazard and rapid decomposition of the oil. For tempering temperatures in excess of 204°C (400°F), a salt bath is preferred over an oil bath.54,70 Oils for tempering are characterized by oxidation resistance with a flash-point much above the operating temperature. The preferred oils include high-flash-point paraffin oils containing antioxidant additives. [See also “Martempering of Steel” (Sec.13.7).] Salt baths for tempering are normally used in the temperature range of 160 to 750°C (320 to 1380°F) and are favored where a rapid and greater uniform heating and low- to medium-volume production are desired. They should not be employed for complex parts or parts with small or blind holes, since there are difficulties in cleaning them. It is necessary to remove all moisture from parts prior to immersion in the molten salt to prevent violent reaction of hot salt with moisture. The parts must be clean and oil-free; otherwise, salt contamination will take place which will require more frequent rectification with chemicals or gaseous compounds. The introduction of cyanide salts or other reducing agents into nitrite tempering baths must be avoided to prevent violent explosion.70 Thorough cleaning of tempered parts exiting the bath is also important because the adhering salts are hygroscopic and may result in severe corrosion. Table 14.7 lists the compositions and operating temperature ranges for salt baths used in tempering.70 The reader is referred to the military specification MIL-S-10699A (Ordnance), mentioned above, for the chemical and other control

TABLE 14.7 Compositions and Operating Temperatures for Salt Baths Used in Tempering70

Operating temperature

Composition of bath, % Class

NaNO2

Fuming temperature

NaNO3

KNO3

Na2CO3

NaCl

KCl

BaCl2

CaCl2

°C

°F

°C

°F

14.56

1

37–50

0–10

50–60

...

...

...

...

...

165–595

325–1100

635

1175

2

...

45–57

45–57

...

...

...

...

...

290–595

550–1100

650

1200

3

...

...

...

45–55

...

45–55

...

...

620–925

1150–1700

935

1720

4

...

...

...

...

15–25

20–32

50–60

...

595–900

1100–1650

940

1725

4A

...

...

...

...

10–15

25–30

40–45

15–20

550–760

1025–1400

790

1450

Source: Military Specification MIL-S-10699A (Ordnance). Reprinted by permission of ASM International, Materials Park, Ohio.

TEMPERING

14.57

procedures applicable to the different bath compositions. (See Chapter 13 for more details on molten salts.) Molten lead bath for tempering is used above 390°C (735°F) and is useful for rapid local heating and selective tempering. Because of its high density, special fixtures are required to hold down the parts in the bath during tempering. Above 480°C (900°F), granulated charcoal may be used as a protective cover. For certain applications, lead-base alloys with lower melting point are used. Vacuum tempering. Although a high-temperature vacuum tempering process can be carried out (along with a vacuum hardening), this practice proves to be inefficient and costly and provides less productivity than the lower-temperature tempering furnace design. When processed in a separate low-temperature tempering furnace, significant savings are achieved in production rates and utility consumption.94 Selective or localized tempering methods are employed either to temper specific areas of fully hardened parts or to temper areas that were locally hardened previously. The objective of this treatment is to increase the toughness and machinability in the selected area, while maintaining high yield strength for resistance to localized deformation and high hardness for wear resistance in the remaining area. Selective tempering is also used in preheating and postheating of weld areas when a decrease of hardness in the heat-affected zone (HAZ) is needed. Special processes such as steam treatment, induction heating coils, special flame heads, protective atmospheres, defocused laser, or electron beam heating are often used to achieve certain desired properties. For some steels, the tempering mechanism is augmented by cyclic heating and cooling. A special significant method uses cycles between subzero temperatures and the tempering temperatures to increase the transformation of retained austenite. Multiple tempering is primarily used for the following reasons: (1) to relieve residual stress induced during quenching and straightening in complex-shaped carbon and alloy steel parts, thereby minimizing distortion; (2) to bring the retained austenite content to an acceptable level and improve dimensional stability; and (3) to increase both the yield strength and toughness (impact strength) without sacrificing hardness.70

14.9.1 Induction Tempering Induction tempering has been widely adapted to automation in an inline manufacturing system such as for pipe, tube, chain, ball screw, shaft, and bar to produce specific mechanical properties. For induction tempering, mostly a different induction coil from hardening is used. The reason is that in induction hardening, heating and hardness pattern are different because of the component’s shape, and the energy density in the hardening process is much higher than with tempering. In induction tempering the surface is heated at a much slower rate to obtain low-temperature gradient from surface to case depth.95 Advantages of induction tempering include the following: 1. There is a possibility of integration with production lines to avoid excessive handling of the workpieces, thereby minimizing labor cost. 2. The tempering cycle is very short compared to the furnace tempering cycle (i.e., a minute or less instead of 1 hr) which often requires an increase in induction tempering temperature to achieve similar hardness with a considerable reduction in residual stresses. Local induction tempering in high-stress areas on hardened parts can result in doubling of the component strength.

14.58

CHAPTER FOURTEEN

FIGURE 14.39 Variation of roomtemperature hardness with tempering temperature for furnace and induction heating.70 (Reprinted by permission of ASM International, Materials Park, Ohio.)

3. There is greater energy efficiency. Energy is not needed to heat up and hold at temperature, as with a conventional furnace. 4. Much less floor space is required than for a conventional tempering furnace. 5. Material handling is minimized. In some instances, the same mechanism can load the green part and unload the finished hardened and tempered product. 6. There is precise control of power, monitoring of the final temperature of individual part, and enhanced operator environment. 7. There is a possibility of incorporation of options to track the parts through the hardening and tempering stations and cool down of the tempered parts before exiting the system.96 Figure 14.39 shows the increase in tempering temperature required to produce a specific hardness with the decrease in tempering time from 1 hr (furnace tempering) to 60 s and 5 s (induction tempering) in AISI 1050 steel quenched in brine from 855°C (1575°F). Small-section parts should be air-cooled immediately after reaching the tempering temperature, whereas large-section parts need slower heating rates or slow periods of time at temperature (5 to 60 s) before cooling to allow heat penetration. In scanning, however, the time of tempering is determined by the power density, the travel speed, and the length of the inductor.70 Equation (14.3) can be employed for short-time induction tempering to calculate the high temperature required with induction to produce the similar hardness to that in furnace tempering.97 EXAMPLE PROBLEM 14.3 To illustrate the application of tempering parameter P, let us consider the required Rc 58 hardness in the root fillet, where P (for conventional tempering for 1 hr at 400°F) = (400 + 460)(18 + log 1) ¥ 10-3 = 15.48; calculate the corresponding induction temperature T ¢ (°F) to achieve the above hardness level.

15.48 = (T¢ + 460) [18 + log (5/3600)] ¥ 10-3, or 15,480 / 15.142668 = T ¢ + 460, or T ¢ = (1022.3 - 460)°F = 562.3°F.

Answer.

TEMPERING

14.59

14.10 STRENGTHENING MECHANISMS OF TEMPERED MARTENSITE AND BAINITE The factors contributing to the strengthening of tempered martensite (and bainite) in ferrous systems are the following:98,99 (1) grain size (i.e., cell or lath size) hardening, (2) Peierls stress, (3) interstitial and substitutional solid-solution hardening, (4) substructure (twins and dislocations) hardening, and (5) precipitation hardening. Grain Size (Cell or Lath Size) Hardening. The Langford-Cohen100 model for cell size hardening is based on the assumption that the dislocation sources in the cell or lath walls are activated more easily than in grain boundaries, and it ascribes the hardening to the stress needed to expand dislocation loops across the cell or lath. This can be represented by the following yield stress sy(0.2) expression s y (0.2 ) = s 0 + KM -1

(14.4)

where s0 is a friction stress, K is the slope coefficient of the yield strength versus inverse of the average cell/lath width plot, and M is the average lath width or its transverse thickness. This equation is applicable to tempered lath martensitic 0.4% C steel (e.g., AISI 4340) with prior heavy cold-work treatment to cause nonequiaxed and finer structures where yield strength is usually independent of packet size. However, a Hall-Petch type relation is observed to exist, which can be expressed by s y (0.2 ) = s 0 + ky d -1 2 + k sM -1 2

(14.5)

for packet size and for lath width or thickness, and has been found to be operative for tempered low-carbon martensite in low-carbon 9% Mn steel101 and 5% Ni and 9% Ni steels102 where the lath thickness remains remarkably constant. In the above expression, ky and ks are grain size coefficients for high-angle and low-angle boundaries, respectively. Peierls Stress. The stress required to move a dislocation through an otherwise perfect lattice is called the Peierls stress. This stress, based on a sinusoidal force relationship, is estimated to have a value of 10-4 m, where m is the shear modulus. Smith and Hehemann99 have adopted a Peierls stress of 41.2 MPa (4.2 kg/mm2) based on the estimate of Speich and Swann103 for AISI 4340 steel. Interstitial Solid-Solution Strengthening. According to Smith and Hehemann, in 4340 steels where complete precipitation of cementite occurs after tempering at 300°C or above, the equilibrium carbon content in the matrix remains at about 0.01%, which is estimated to contribute 137.3 MPa (14.0 kg/mm2) to s0. However, according to Cox,104 a considerable amount of interstitially dissolved carbon was retained in a 0.3 to 0.38% carbon martensite even after tempering above 550°C, and he therefore noted that 60 to 70% of the strength of tempered martensite was attributed to interstitial strengthening over the entire tempering temperature range. Substitutional Solid-Solution Strengthening. According to Lacy and Gensamer, the assumption of additive effects of various substitutional elements on the yield strength of a 4340 alloy amounts to 165.7 MPa (16.9 kg/mm2). This contribution remains nearly the same up to a tempering temperature of 540°C, provided that the solute distribution is not markedly changed.98

CHAPTER FOURTEEN

14.60

Substructure Strengthening. In a tempered carbon-free Fe-Ni martensite, Speich and Swann103 have predicted the total contribution of twin substructure formation to the yield strength to be about 137.3 MPa (14.0 kg/mm2). On the other hand, Kelly and Nutting105 and Cox104 predicted the substructure contribution of 206 MPa (21 kg/mm2) and 68.7 to 206 MPa (7 to 21 kg/mm2), respectively, to the flow stress in low-carbon and 0.3 to 0.38% carbon martensites. The contribution Ds of substructure dislocation strengthening in 0.42% carbon steel is given by106 Ds = 2 Dt = 2ambr 1 2

(14.6)

where Dt is a flow stress contribution (consistent with the usual terminology), r is the dislocation density per square centimeter, and a is the dislocation strengthening coefficient, usually between 0.33 and 0.4. Since both the densities of internal twinning and of dislocations fall substantially with an increase in tempering temperature, their strengthening contributions to the yield strength must be reduced at elevated temperatures. This is supported by the observations made by Cox104 and Malik and Lund.106 The work-hardening effect that occurs in attaining a 0.2% offset yield strength has been assumed to contribute a constant value of 68.6 MPa (7 kg/mm2) to s0 for all structures in 4340 steel. Precipitation Hardening. The contribution of aging of lath martensite in lowcarbon steel is negligibly small. However, the contribution of aging after the quench in twinned martensite is significantly large and can cause an increment in hardness, compared to the virgin martensite. The most well-recognized general expression for calculating the dispersion hardening strength Ds (= sp) due to the presence of carbide particles (dispersed phase) has been given by Kelly and Nicholson and Ashby and Orowan107 as Ds = s p =

mb È 1 Ê 1 ˆ˘ D 1+ ln p ÍÎ 2 Ë 1 - v ¯ ˙˚ 2b

(14.7)

and sp =

D 0.015 D ln = 5.9 f 1 2 ln ls 2b 2.5 ¥ 10 -4

(14.8)

where m is the matrix shear modulus, b is the Burgers vector of a dislocation in the matrix, D is the average precipitate particle diameter intersecting the slip plane, ls is the effective carbide (interparticle) spacing (mm), f is the volume fraction of carbide precipitate, and the term within the parentheses is an average Poisson’s ratio n for edge and screw dislocations. Onel and Nutting107a have estimated the contribution of carbide precipitation in a heavily tempered plain carbon steel occurring (1) solely at the ferrite grain boundaries and (2) also within the tempered ferrite grains. In the former case, the yield stress, without exhibiting additional strengthening contribution, is given by s y = 108 + 18.2d -1 2

(14.9)

In the latter case, the yield stress, with additional strenghthening effect sp, is given by s y = 77 + 23.9d -1 2 + s p

(14.10)

TEMPERING

Stoichiometric ratio

Bainitic structure

60

40 Martensitic structure 20

0 0

0.2

0.4 0.6 wt. % V (a)

0.8

1.0

Hardness after tempering at 650°C, H.V.

Intensity of secondary hardening H.V.

80

14.61

600 Stoichiometric ratio 550

500

450

400 0

1.0

2.0 wt. % V (b)

3.0

4.0

FIGURE 14.40 The effect of vanadium on (a) the intensity of secondary hardening in tempered 0.1% C steels consisting of prior martensitic or bainitic microstructures and (b) the overaged hardness of 0.6% C martensitic microstructures tempered for 1 hr at 650°C, both showing the maximum effect at the stoichiometric ratio.108 (Courtesy of F. B. Pickering.)

where sp =

0.015 D ln ls 2b

(14.11)

and all terms have the same meaning as mentioned above. A significant aspect of secondary hardening alloy steels is that the 0.2% yield strength is a maximum at steel compositions corresponding to the stoichiometric metal/carbon ratio (wt%) for the precipitating alloy carbides. This takes place for both prior martensite and prior bainite microstructures, as shown by hardness peaks in Fig. 14.40a; this effect continues even into overaged microstructures (Fig. 14.40b).108 The reason for this phenomenon is that at the stoichiometric ratio, the temperature dependence of the solubility is a maximum109 so that an increased volume fraction of alloy carbides provides increased precipitation strengthening according to Eq. (14.8) or (14.11) (see Fig. 14.41).110 Final Expression of Yield Strength. In general, it is considered that the contributions from the carbide precipitates, from the grain or lath boundaries, and from the dislocation substructures are additive. For example, the yield strength of tempered martensite in an Fe-0.42C-1.1Mn-0.22Si-0.13Mo-0.07S-0.011P steel can be expressed as106 s y (0.2 ) = s 0 + 2ambr 1 2 +

mb Ê 1 ˆ D 1+ ln 2pD Ë 1 - v ¯ 2b

(14.12)

where s0 = 343 MPa (35 kg/mm2) and a 艑 0.4. On the other hand, some researchers are of the view that contributions from the dislocation substructure and the carbide precipitates are mutually dependent

14.62

CHAPTER FOURTEEN

FIGURE 14.41 The dependence of the intensity of secondary hardening on the volume fraction of VC carbides precipitated during tempering of 0.35 to 0.50% C martensitic microstructures.110 (Courtesy of F. B. Pickering.)

because both carbide dispersion and dislocation substructures (1) are encountered simultaneously by the mobile dislocations in a slip plane, (2) undergo considerable changes with increasing tempering temperature in the range of 250 to 550°C, and (3) account for 80% of the steel strength on tempering at 250°C and ~70% of the steel strength on tempering at 550°C in AISI 4340 steel. In this situation, Smith and Hehemann99 and Daigne et al.111 derived the yield strength of tempered martensite (MPa) in 4340 steel in the form s y (0.2 ) = s 0 ( = 572) + 126M -1 + 68D-1

(14.13)

where s0 is the sum of (1) the Peierls stress, (2) interstitial and substitutional hardening, (3) work hardening, and (4) the internal dislocation substructure hardening. In order to account for the intergranular carbide particles, it was assumed that the intrinsic average resistance of lath boundaries to shear propagation increased from a value of K per unit area of clean boundary to a large value of sp per unit area of boundary covered with carbides. This assumption leads to an expression s y (0.2 ) = s 0 + 2 KM -1 + 1.28(s p - K ) f 1 2 D-1

(14.14)

where f is the volume fraction of carbides and the other terms have the usual meanings as stated above.

14.10.1

Toughness of Tempered Martensite and Bainite

There is sufficient evidence in the literature which suggests that the transition temperature T of tempered martensite and bainite increases with an increase in carbon content. Figure 14.42 shows the effect of tempering on the relationship between 0.2% yield stress and the ductile-brittle transition temperature (DBTT) T for lowcarbon bainitic microstructures, illustrating the effect of different strengthening

TEMPERING

14.63

FIGURE 14.42 Effect of tempering on the relationship between the 0.2% yield stress and the DBTT for low-carbon bainitic steels, illustrating the effect of several strengthening mechanisms.108 (Courtesy of F. B. Pickering.)

mechanisms. This observation also holds for tempered martensite. In both cases, at lower tempering temperature, T decreases with the decrease in the yield stress. However, at higher tempering temperatures, below A1, the recrystallization and grain growth of the ferrite matrix cause an increase in T, even with a decrease in sy (as evidenced in Fig. 14.42).108 Like martensite and bainite, experimental evidence suggests that an increase in prior austenite grain size of tempered martensite increases T.112 Tomita and Okabayashi have observed that, in martensitic and lower bainitic (low-alloy structural) steels, the packets are the principal constituent within the prior austenite grains, and the strength and toughness of the steels are increased with decrease in the packet diameter. Likewise, in upper bainitic microstructures, the well-defined blocks (i.e., matrix of laths having low-angle boundaries) rather than packets, control the yield stress and DBTT.113 These observations, however, indicate that, provided no embrittlement occurs during tempering, an analysis based on Eq. (7.44) might be employed to relate the microstructure to the transition temperature in tempered martensite. However, modifications pertaining to the effects of precipitation and dislocation strengthening, and omission of the pearlite term would be essential, as in Eq. (9.13), in conjunction with the fracture facet or martensite packet size. It would also be essential to consider a factor for the thickness of the carbide particles t. Thus, finally we can use the modified Eq. (14.15) for DBTT: T (∞C) = 46 + 0.45s p + 131 t 1 2 - 12.7 d -1 2

(14.15)

14.10.2 Effect of Embrittlement on Toughness Figure 14.43a and b shows the secondary hardening embrittlement during tempering of a 0.4C-5Cr-Mo-V steel, exhibiting minimum impact energy and KIC values,

CHAPTER FOURTEEN

14.64

20

600

15

500

10 Impact energy

400

300

5

H.V.

400

450 500 550 600 650 Tempering temperature in °C (a)

700

0

Fracture toughness KIC in MNm–3/2

700

Impact energy at 20°C in J

Hardness H.V.

70 60 50 40 30 20 10

70 60

HRC

55

50 50 40 45 30

KIC

Hardness HRC.

Fracture toughness KIC in MNm–3/2

60

40

20 35 10

200

300 400 500 Tempering temperature in °C (b)

600

FIGURE 14.43 Effect of secondary hardening embrittlement during the tempering of a 0.4C-5Cr-Mo-V steel on (a) impact energy and (b) fracture toughness parameter KIC.108 (Courtesy of F. B. Pickering.)

respectively. Similar effects are also found in tempered martensite embrittlement, reversible temper embrittlement, and upper nose embrittlement, which depend on the tempering time.32,108, 110 Note that embrittlement always accompanies an increase in T where the brittle fracture mechanism usually changes from cleavage to various types of intergranular cracking, which again operates against an overall quantitative analysis.108 Some information on the effect of microstructure on the fracture toughness KIC value suggests that a decrease in strength by carbide coarsening and a reduction in dislocation density result in about the same increase in KIC.110,114 However, a general relationship between microstructure and KIC parameter is lacking, because KIC is not a unique material property; rather, it is a function of both testing temperature

TEMPERING

14.65

TABLE 14.8 Effect of Sulfur and Volume Fraction of MnS on the KIC Value for a Quenched and Tempered Medium-Carbon Steel108,115 Wt%, S 0.008 0.016 0.025 0.049

Volume fraction, MnS 4.3 8.6 1.3 2.6

KIC, MNm-3/2

¥ 10-4 ¥ 10-4 ¥ 10-3 ¥ 10-3

72 62 56 47

FIGURE 14.44 Effect of nonmetallic inclusion content (vol%) on the fracture toughness parameter KIC for a quenched and tempered high-strength steel, showing the anisotropy of fracture toughness.108,116

and fracture mode. As expected, an increase in carbon content usually decreases KIC.108,114 Table 14.8 shows that an increase in sulfur content and volume fraction of MnS in quenched and tempered high-strength steels lowers the KIC value.115 On the other hand, other workers have shown that the basic features of the inclusion distribution which have a bearing on ductility and Cv values also control KIC in high-strength steels when a ductile fracture mode occurs. Figure 14.44 shows the effect of nonmetallic inclusion volume fraction on the anisotropy of KIC; at low volume fraction, there is a little anisotropy because of its small effect on KIC for the cleavage fracture. Note that Cv decreases exponentially with increasing strength in tempered martensite, because perhaps this represents a decrease in dispersion strengthening sp rather than coarsening of grain size.108

14.11 THERMALLY INDUCED EMBRITTLEMENT PHENOMENA Low-alloy quenched and tempered steels are prone to different types of embrittlement phenomena, which usually fall into two categories: (1) thermally induced

14.66

CHAPTER FOURTEEN

embrittlement phenomena, resulting from structural changes produced during thermal treatment (including tempering or slow cooling) or elevated-temperature service in a specific temperature range, and (2) environmentally induced embrittlement phenomena, resulting from the interaction of the environment with the quenched and tempered structures. The former types of embrittlement comprise tempered martensite embrittlement, temper embrittlement, secondary hardening embrittlement, thermal embrittlement, and embrittlement due to aluminum nitride formation; and these are discussed mostly in this section, and one in Sec. 14.14.5. The latter types include hydrogen damage and metal-induced embrittlement, comprising liquid-metal embrittlement and solid-metal embrittlement and are discussed in subsequent sections. In many cases, an overlap between the two types of embrittlement occurs.

14.11.1

Tempered Martensite Embrittlement

A sudden loss of room-temperature toughness of quenched commercial highstrength low-alloy steels during tempering in the temperature range of 250 to 400°C (482 to 752°F), in spite of a loss in strength, has been variously referred to as tempered martensite embrittlement (TME), 350°C or 500°F embrittlement, and one-step temper embrittlement. This has produced severe difficulties in the development of ultrahigh-strength steels. As a result of this embrittlement, this tempering range is usually avoided in commercial practice. Steels with tempered lower bainitic microstructures are also prone to TME, whereas steels with pearlite/ferrite and upper bainite microstructures are not embrittled by tempering in this region.117–119 14.11.1.1 Characteristics of TME32,120–122 1. There is a minimum in absorbed fracture energy (usually measured by roomtemperature Charpy impact test), its magnitude indicating the extent or severity of embrittlement. The drop in CVN impact energy associated with TME can be explained by a change in three energy components associated with initiation zone, stable growth (or fibrous fracture) zone, and post-unstable (or termination) zone with tempering temperature.117 2. The ductile-brittle transition temperature (DBTT) increases through a maximum representing the minimum fracture energy (or Charpy toughness). In other words, it can be characterized by a trough in the plot of Charpy impact energy as a function of tempering temperature for a particular test temperature. Figure 14.20a shows room-temperature Charpy V-notch impact energy for AISI 4130 steels at two different phosphorus levels, 0.003 and 0.03%, as a function of tempering temperature. 3. Compared to temper embrittlement (discussed later), TME is a rapid process; the amounts of impurity segregation needed for intergranular fracture are much lower, and the intergranular fracture required to produce the toughnessminimum is relatively small.122a 4. TME is induced during tempering in the critical temperature range within the usual 1-hr time period and is independent of section dimension and/or cooling rate after tempering.

TEMPERING

14.67

5. Fractures of test samples within the critical temperature range do vary, with transgranular fractures usually observed in the tempering temperature range of 200 to 300°C (390 to 570°F) and intergranular fracture mostly observed at 350°C (660°F) ± 50°C (122°F). These differences may be related to the differences in carbon, alloy, and impurity content, as well as in strength level, nature of the test, test temperature, and grain size. Impurities seem to affect coarse-grained steels to a larger extent than fine-grained steels.119

14.11.1.2 Types of TME Based on Fracture Mode According to the fracture mode, the TME can be classified into two types, namely, transgranular TME and intergranular TME, both of which depend on the microstructure (mainly carbide formation) and variation in strain- (or work-) hardening rates induced by tempering.117 Both ductile and brittle microcracks commence and develop at carbide particles, depending on the size, morphology, and distribution of carbides and on the strain-hardening rate. High work-hardening rates and high volume fraction of carbides (such as in 4150 steel) favor the brittle or intergranular fracture modes of TME. The transgranular TME or the ductile mode of TME is observed in 4130 steel as a result of the decrease in both the work-hardening rates and carbide distribution.117 The interlath (i.e., the area between the parallel martensite laths), intralath (i.e., the area across the martensite laths), and the grain boundary carbides formed by the decomposition of retained austenite or during the third stage of tempering, as well as the undissolved carbides, take part in fracture initiation. The transition carbides do not play any role in crack initiation; however, they influence the strain-hardening rate.117 The transgranular TME occurring at interlath or translath has been associated with (or ascribed to) the following: 1. The decomposition of interlath-retained austenite to interlath carbide films (or stringers) during tempering.43,60,123 This microstructure is similar to that observed for upper bainite.124 2. The subsequent deformation-induced transformation on loading of remaining interlath-retained austenite which has become mechanically unstable due to the carbon depletion caused by this carbide precipitation.125 3. The formation of coarse (or thick) interlath cementite film as a result of the third stage of tempering in steels containing a very small volume fraction of retained austenite,126 and/or the formation of coarse interlath Fe2N. This is usually observed in high purity or lower-carbon quenched and tempered steels such as 4130 (Figs. 14.20a and 14.45a). As usual, crack initiation and crack growth take place by microvoid coalescence around carbide particles retained after austenitizing and formed during the second stage of tempering (Fig. 14.45a). Transgranular TME has been found to develop upon tempering underhardened M2 (containing 0.8% C, 6% W, 5% Mo, 4% Cr, 2% V) and Vasco-MA tool steels (containing 0.5% C, 2% W, 2.75% Mo, 4.5% Cr, 1% V) austenitized at 1040°C (1900°F) or 980°C (1800°F) with a simultaneous decrease in hardness.128 The intergranular TME is ascribed to the decomposition of lath-boundaryretained austenite and the subsequent formation of long and platelike (or interlath)

14.68

CHAPTER FOURTEEN

carbide (M3C or FexC) precipitates at prior austenite grain boundaries which are already weakened by segregation of impurity elements such as P and S, which occur during austenitization120,129 (Fig. 14.45b), and which facilitate the alternate crack nucleation sites or easy crack path to follow. This is also linked with stress concentration susceptibility at the grain boundaries, which is associated with the matrix toughness. In most steels (e.g., 4340 and silicon stabilized M-300), the lowest KIC and Charpy V-notch impact energy at room temperature is related to the increasing amount of intergranular fracture along prior austenite grain boundaries. However, in some cases, it has been suggested that the minimum in fracture energy may be due to cleavage fracture favored by the formation of carbides during tempering.126 14.11.1.3 Influence of Variables. In general, the level of toughness, the severity of TME (as indicated by the depth of the trough), the temperature range of embrittlement, and the mode of failure mechanisms depend on the complex interactions between steel composition, grain size, heat-treating condition, test temperature, and the testing methods.117 Composition. An increase in carbon content from 0.3 to 0.5% in 41xx series steels decreases the CVN absorbed energy and the apparent severity of TME (Fig. 14.46).117 This increase in carbon content causes the crack initiation, ahead of the notch, to change from ductile to brittle mode; however, this depends on test temperature also.129a Capus and Mayer130 were the first to report the influence of impurities on TME. They found the absence of TME in the high-purity steels, and they showed the embrittlement in the steel which contained P, N, Sb, Sn, Mn, or Si alloying addition. Briant and Banerjee131 found the occurrence of intergranular rupture of some heattreating conditions in steels doped with 0.01% N which was believed to be due to precipitation of Cr2N along prior austenite grain boundaries. The severity of TME developed by the addition of P remains the same (i.e., uniform decrease in impact toughness occurs) in both low (0.002% P) and nominal (0.02% P) 41xx series steels (Fig. 14.46).117 It has been seen that Mo addition reduces the P-induced embrittlement, whereas Mn and Cr are potent embrittlers in both P-free and P-doped (0.03%) alloy samples.132 These elements probably change the ability of P to cause embrittlement by changing the chemical bond it can form at the grain boundaries.131 For the low-phosphorus steels (containing 0.003%), interlath cementite is the contributor in initiating cleavage fracture across the martensite laths (Fig. 14.45a), while for high-phosphorus steels (containing 0.03%), a combined effect of carbide formation and P segregation is responsible for intergranular fracture (Fig. 14.45b). However, the plane strain fracture toughness test data exhibited TME only in the high-phosphorus steels, while the Charpy V-notch data displayed TME in both steels. Sulfur is a more potent embrittler (Fig. 14.47);122 in many steels, however, it is precipitated as a sulfide, which does not segregate to the grain boundaries. It has been observed that if S is precipitated as chromium sulfide, embrittlement does not occur in the quenched and tempered steel with 3.5 Ni, 1.7 Cr, 0.3 C, 0.004 S, provided that the specimen is austenitized at or below 1000°C (1832°F). However, above an austenitizing treatment of ~1050°C (1920°F), sulfides dissolve and embrittlement is observed. It is apparent that alloying elements such as Ni (up to 2%), Si (up to 2.5%), and Al, which retard the conversion of e-carbide to cementite within martensite laths132a and/or the precipitation and growth of cementite, can delay the onset of TME to a higher temperature. On the other hand, Mn and Cr promote the decomposition of

TEMPERING

14.69

(a)

(b)

FIGURE 14.45 Fractography of AISI 4340 steel specimens that failed due to tempered martensite embrittlement. Specimens were broken by impact loading at room temperature. (a) Flat cleavage facets in a specimen containing 0.003% P after tempering at 350°C (662°F). (b) Intergranular fracture in a specimen containing 0.03% P after tempering at 400°C (752°F).60 (Reprinted by permission of the Metallurgical Society, Warrendale, Pa.)

retained austenite at a lower temperature; therefore, steels containing Cr or Mn are highly susceptible to TME, and this is probably due to the precipitation of chromium or manganese nitride at the grain boundaries.124 For example, the decomposition of retained austenite is completed for 1 hr at 300°C (572°F) in Mn-containing steels and

14.70

CHAPTER FOURTEEN

FIGURE 14.46 Room-temperature CVN impact energy versus tempering temperature for 4130, 4140, and 4150 steels austenitized at 900°C and tempered for 1 hr at the temperatures shown.117 (Reprinted by permission of Pergamon Press, Plc.)

for 1 hr at 400°C in Ni-containing steels.43 The extent of TME increases with the increase of Mn and Si contents corresponding to 4340 steel. Elimination of Mn and Si from high-purity steels results in the elimination of most of the susceptibility to TME. This may be interpreted in terms of the effects of these elements on impurity segregation. Grain Size Effect and Test Variables. Reduced grain size reduces the TME only when the embrittlement is less severe. Fine-grained 0.01% P-doped steels do not show any embrittlement at room temperature but exhibit embrittlement at a lower test temperature (Fig. 14.48).122 The embrittlement trough becomes more pronounced when the test temperature is below the DBTT of samples near 350°C. The four-point bend test method with the Griffith-Owen elastic-plastic stress analysis is more sensitive to TME than is the Charpy test, and the data obtained have a clear physical interpretation.120 14.11.1.4 Control of TME. TME in low-alloy high-strength steels can be minimized by118

TEMPERING

14.71

FIGURE 14.47 A comparison of TME in steels doped with either 0.01 wt% phosphorus or 0.01 wt% sulfur. The two different graphs represent two different austenitizing treatments carried out for samples with the base composition 3.5 Ni, 1.7 Cr, 0.3 C steel. (a) 䉭, 0.01 P; 䊊, 0.01 S (at 850°C/1 hr). (b) 䉱, 0.01 P; 䊉, 0.01 S (at 1200°C/3 hr). (Source: Met. Trans., vol. 12A, 1981, p. 317.)

FIGURE 14.48 The fracture energy at different test temperatures plotted as a function of tempering temperature. Composition of steel: 3.5 Ni, 1.7 Cr, 0.3 C, 0.01 P. 䊉, -80°C; 䊏, -40°C; 䉱, -5°C; 䉭, 23°C; 䊊, 50°C; ⵧ, 90°C. (Source: Met. Trans., vol. 10A, 1979, p. 1732.)

1. Development of special steels with delayed onset of TME 2. Development of steels with faster rates of martensite tempering 3. Use of steels capable of transforming fully to upper bainite at the required strength level and section thickness 4. Avoiding tempering in the critical temperature region 5. Use of the lowest possible carbon concentration consistent with the required strength level and reduction of Mn and Si contents129a

14.11.2 Temper Embrittlement When a quenched alloy steel is tempered or slowly cooled through or isothermally heated in the critical temperature range, usually 300 to 600°C (572 to 1112°F), it

14.72

CHAPTER FOURTEEN

suffers an upward shift in DBTT,† a progressive reduction in impact toughness, and an increasing tendency toward intergranular (brittle) fracture;133,134 this metallurgical phenomenon is termed temper embrittlement (TE), reversible temper embrittlement (RTE), 500°C embrittlement, or two-step temper embrittlement.135–137 TE is reversible; reheating an embrittled steel beyond 600°C (1112°F) and fast cooling eliminate most of the embrittlement.122a Note that TE occurs not only after the tempering in the embrittling temperature range (e.g., 550°C) but also on slow cooling (extending several hours) from high temperatures.138 Although plain carbon steels are not embrittled by P138a and immune to TE, the substantial addition of Mn causes susceptibility to this problem. This phenomenon is an extraordinarily complex metallurgical problem which had been known since the 19th century, and it was long considered as a metallurgical mystery. It has caused a potential concern in heavy-section components such as heavy armor plate, turbine rotors, and pressure vessels, which are slowly cooled through the embrittling temperature range following the tempering treatment during fabrication or experience service temperature operating within this embrittling range. Another area of concern is the prolonged use of such steels in the embrittling temperature range such as low-pressure rotors and disks of large steam turbines where concern over temper embrittlement has restricted the maximum use temperature, to the detriment of thermodynamic efficiency.136,137 TE can also substantially promote the susceptibility of a steel to: hydrogen-induced embrittlement;139 stress corrosion cracking;140 fatigue crack propagation, especially at high maximum stress intensity;141 and stress-relief cracking in heat-affected zones of a constrained weldment, particularly in creep-resistant low-alloy steels.137,142 Table 14.9 lists a variety of TE studies.143–147 14.11.2.1 Metallurgical Variables. TE studies on numerous steels have shown that the extent of TE depends primarily on the following metallurgical variables: intergranular concentration of embrittling metalloid impurities (also Mn); composition; grain size;148 hardness;143 and thermal treatment, including the rate of cooling. Impurity Grain Boundary Segregation. The general conclusions drawn from TE studies on impurity grain boundary segregation are as follows: 1. This phenomenon (i.e., grain boundary decohesion) is attributed to the grain boundary cosegregation, during tempering, of certain metalloid impurities (P, Sn, Sb, As) at prior austenite grain boundaries (Fig. 14.49a) or Fe3C/martensite interfaces with alloying elements such as nickel, chromium, silicon, or manganese. TE will not occur in the absence of impurity segregation. 2. The experimental data on 3340 (Ni-Cr) steel doped individually with P, Sn, Sb, and As indicate that the relative potency of these impurities are in the order Sb > Sn > P > As.149‡ 3. The embrittled grain boundary in a quenched and tempered steel provides a continuous fracture path (Fig. 14.49b). Embrittlement occurring at the Fe3C/martensite interface boundaries, in the tempered martensite, provides a discontinuous fracture path.137

† The DBTT can be assessed in three ways: (1) the temperature for 50% ductile and 50% brittle fracture (50% fracture appearance transition temperature, or 50% FATT), (2) the lowest temperature for 100% ductile fracture (100% FATT), and (3) transition temperature based on absorbed energy values. Among these, the first one is most common, and the last one is normally not used. ‡ Later work by McMahon, Jr., has not found any evidence of As as an embrittler.129a

TABLE 14.9 Examples of a Two-Step Temper Embrittlement

Type of steel

Reference

Composition, wt%

Embrittling heat treatment

Change in transition temperature, °C

Segregating impurities

210

Si, P, Sn, N

144

0.11C-0.80Mn-0.003P0.006S-0.0355Si-4.95Ni0.53Cr-0.50Mo-0.08V

480°C for 1000 hr

3340 + 0.06 P

143

3.5Ni-1.7Cr-0.06P-0.4C

480°C for 100 hr

175

P

Plain carbon

145

0.2C-2.0Mn-0.05Si0.0815Sb-0.006P-0.011S

500°C for 480 hr

Transition temperature not measured but fracture surface was 90% intergranular

Sb

A533B with high Cu

146

0.18C-0.25Si-1.39Mn0.007P-0.01S-0.37Cu0.63Ni-0.19Cr-0.55Mo0.016Al-0.0108N-0.009Sb0.008Sn-0.013As

500°C for 100 hr

120

P, Cu(?)

Cr-Mo base

147

0.34C-3.1Cr-0.59Mo0.89Mn-0.33Si-0.18Ni0.036S-0.03P-0.015Sn0.029As-0.008Sb

Furnace cool after temper

130

?

14.73

HY130

CHAPTER FOURTEEN

14.74

(a)

(b)

FIGURE 14.49 (a) Quenched and tempered Ni-Cr steel containing segregated phosphorus at the prior austenite grain boundaries, revealed by a picric acid etchant with a wetting agent. (b) Fracture surface of the same specimen showing the occurrence of brittle fracture along the phosphorus-contaminated prior austenite grain boundaries.137 (Courtesy of C. J. McMahon, Jr.)

4. The impurity segregating at the grain boundaries appears to be an equilibrium and a reversible phenomenon. However, the results of Faulkner et al.150 on combined quenching and tempering induced P segregation to grain boundaries in a 0.077% P doped 2.25Cr-1Mo steel indicate that combined equilibrium and nonequilibrium segregation plays a significant role in TE of the steel caused by direct tempering after quenching. Nonequilibrium segregation needs the formation of adequate amounts of vacancy-impurity complexes, and their migration to grain boundaries is of vital importance in the segregation. 5. The equilibrium grain boundary concentration of impurities increases with decreasing aging temperatures. However, aging time at the low aging temperature plays a vital role. For example, in a study made on 3.5Ni-1.7Cr-0.4C steel, it has been found that segregation of P is rapid and reaches equilibrium within 50 to 100 hr at 480 to 560°C; Sn and Sb require much longer to attain equilibrium. The rate of segregation falls off sharply below 400°C because of diffusional kinetics and above 560°C because of entropy effect.132 6. Among the impurity elements, P (as well as Mn) is the most common grain boundary embrittler in commercial alloy steels because of (a) its segregation during austenitization, tempering, and aging; (b) its rapid segregation even at low aging temperature; and (c) its larger concentration than those of other embrittling impurity elements in commercial steel. Sb is rarely present at the grain boundaries, and sulfur is usually precipitated as manganese or chromium sulfides. The next very important impurity elements are Sn and As. It is also evident that the complex sequence of precipitation of carbides in steels has an important influence on the embrittling process by affecting the impurity segregation.122 Composition: Alloying Elements. The embrittling impurity elements produce increased TE in Ni-Cr steels compared to the Ni and Cr steels. Mo additions are effective in decreasing or eliminating TE when impurity elements are present and retard the kinetics of TE of Fe-Mo-P alloy steels. However, to be more effective, Mo must be dissolved in the ferrite matrix and not tied up in carbides, with a maximum effect

TEMPERING

14.75

at a concentration of around 0.7%. The very strong interaction between Mo and P results in the precipitation of (Mo,Fe)3P or Mo-P atom cluster which prevents the segregation of P to grain boundaries.32 It has been shown that C and P compete with each other in grain boundary segregation, in that a strong repulsive interaction between C and P may cause hindrance to P segregation. When more Mo is tied up in carbides, its beneficial effect diminishes. An increase in Cr concentration increases the embrittlement resulting from a fixed amount of P in Ni-Cr and other steels.143,151,152 This influence of Cr on embrittling behavior is attributed to the reduction of carbon activity. The similar effect is found with any carbide formers such as Nb.151† Cr addition also raises the potency of Sn and Sb.152–154 When the addition of Ni and Cr is made alone, the grain boundary segregation becomes large and small, respectively. However, the addition of both Ni and Cr in Sb-doped Ni-Cr steel leads to a very large Sb and Ni segregation compared to their cumulative individual effects.122 The interaction between Mo and C seems to be the dominant factor that has a bearing on the grain boundary cohesion in the Fe-Mo-P alloys.154a Mn and Si also increase the susceptibility to embrittlement. The addition of 0.3% Mn to a NiCrMoV rotor-type steel with 0.02% P was found to produce a large increase in susceptibility of TE, compared to a Mn-free steel. However, when only 30 ppm P was present in a NiCrMoV steel, the presence of Mn did not cause TE. A study of the effect of 1% Mn in a high-purity, decarburized iron showed the segregation of Mn to the grain boundaries, associated with a large reduction in the intergranular fracture stress at 133 K. When P was added to the Fe-1%Mn alloy, it raised the amount of Mn segregation due to their attractive interaction and vice versa (with a Mn-free Fe-P alloy). Based on these findings, it has been concluded that Mn addition has two effects with respect to TE. One effect is attributed to its interaction (i.e., cosegregation effect) with P which increases the segregation level of P, thereby enhancing the embrittling potency of P. The other is an intrinsic embrittling effect due to segregation of Mn to the grain boundaries, which causes reduced intergranular fracture strength.155 However, some workers have not observed the enrichment of Mn to the austenite grain boundaries in 10Mn-P steels.156 However, the presence of both Mn and S in the steel leads to the scavenging of S from solid solution by incorporation into existing sulfides, thereby its inability or less availability to segregate.132,157 Similarly, Ti has a beneficial effect in the lowcarbon Fe-Ni-Cr-Sb alloy, which is a scavenger of Sb.132 Steels containing Mo, W, and/or V exhibit delayed temper embrittlement due to the slow precipitation of alloy carbides of increasing stability.137 The mechanisms of delayed TE in Mo-bearing steels can be explained in the following way:137,158 The reduction of Mo (or C) activity in the ferrite matrix causes reduced segregation to grain boundaries, presumably due to the formation of Mo2C carbides and the release of locked up impurities such as P into solid solution, which allows increased P segregation, finally reaching the level expected in Mo-free steels. In Cr-Mo steel containing P as an impurity element, the kinetics of P segregation and hence the embrittlement is quite different from that in Ni-Cr steel as a result of the initial interaction of P and Mo. One mechanism to explain the influence of Mo is that it serves as a scavenger for P but that the resulting Mo-P compound is less stable than Mo-rich carbides. The other mechanism is Mo-P cosegregation which tends either to reduce greatly the embrittling potency of P at the grain boundaries or to exert an additional intrinsic strengthening effect due to the increased grain bound† According to McMahon, Jr., and Yu-Qing, Cr is mainly a scavenger of carbon which indirectly influences in reducing carbon segregation and allowing greater P segregation, both of which provide increased intergranular embrittlement.154a

14.76

CHAPTER FOURTEEN

ary cohesion. Experimental evidence favors both mechanisms.132,158,159 The addition of Mn and Si in the Cr-Mo base alloy increases the potency of P. The presence of V retards the rate of TE in Mo-bearing steel by a factor of 10; the mechanism, whether it is scavenging element or whether it retards the formation of Mo-rich carbides, has not yet been made clear. The P-doped NiCrMoV steel has been found to show the least TE and the commercial Ni-Cr steel the most. The P-doped NiCrV steel is more prone to TE than P-doped NiCrMoV steel. This difference is due to Mo-enhanced grain boundary cohesion;160 however, it is a questionable proposition.129a Microstructure. Matrix microstructures are also very important because they control the toughness of both embrittled and nonembrittled steels. In general, tempered martensite is more susceptible than tempered bainite to TE; however, tempered bainite is more susceptible than ferrite-pearlite structures.161 Grain Size and Hardness. It has been shown in a quenched and tempered steel that for a fixed grain size and hardness, the extent of TE is a function of the metalloid element concentration in the grain boundaries. Similarly, for a fixed metalloid impurity concentration in the grain boundaries, the extent of embrittlement is a function of the alloying element, hardness, and grain size of the steel where the DBTT increases with either hardness or grain size or both. Thus, for a fixed level of impurities and constant embrittling temperature and time, there is a greater coverage of the grain boundaries in a coarse-grained steel than in a fine-grained steel. However, the distance over which the impurities must diffuse increases with the increase of grain size. The degree of embrittlement increases with an increase in prior austenite grain size, and in steels with duplex grain structures, the size of the largest grains controls the deterioration in toughness. The reason is that less energy is needed to initiate and propagate cracking on the embrittled boundaries of a coarse-grained steel.162 In the case of CrMoV steels with P as the impurity element, susceptibility to TE during service is practically eliminated when the prior austenite grain size is ASTM number 9 or above. However, a decrease in ASTM grain size from 4 to 0 increases the shift in FATT by 61°C (110°F).161 Figure 14.50 shows the effect of prior-austenite grain size on the TE of a Ni-Cr steel (containing 0.33 C, 0.59 Mn, 0.03 P, 0.031 S, 0.27 Si, 2.92 Ni, and 0.87 Cr) that was heat-treated to produce coarse- and fine-grain size. It is quite clear that coarsegrained specimens were severely embrittled compared to fine-grained specimens.163 It is thus inferred that fine-grain size leads to a decrease in the amount of embrittlement for any given grain boundary segregation. The improvement induced in this case is much larger than in the TME. In general, the grain size effect increases with the potency and concentration of the embrittling element and with hardness.144 Increasing hardness (or yield strength) increases the extent of TE for a fixed amount of grain boundary segregation. This is quite evident in plain carbon steels which are not embrittled by P because of being too soft.138a In an experimental study on Ni-Cr steels doped with P and Sn, it was found that the transition temperature could be expressed as a Taylor series involving as variables the grain size, hardness, and average metalloid impurity concentration on the fracture surface. Figure 14.51 shows the variation of transition temperature for various prior austenite grain size and hardness values in these steels when doped with P and Sn as a function of Auger peak-height ratio (PHR) with respect to Fe in a Ni-Cr steel.164 In more complex steel, other factors must be taken into consideration, such as a real fraction of hard carbide or nitride particles on the grain boundaries and presence of alloying elements, such as Mn, Ti, Mo, and so forth, cosegregated with the metalloid impurity elements.137

14.77

400

200 Coarse grain 150

300

100

200 Fine grain

50

100

0 0 –50 –100 –100 0

2

4 6 Time at 500°C (930°F), h (a)

8

10

450

250 Shift in 100% fibrous FATT, D°C

Actual 100% fibrous FATT temperature, °F

250

200

360 Coarse grain

150

270 Fine grain

100

180

50

90

0 0

2

4

6

8

Shift in 100% fibrous FATT, D°F

Actual 100% fibrous FATT temperature,°C

TEMPERING

0 10

Time at 500°C (930°F), h (b)

FIGURE 14.50 Effect of prior-austenite grain size on the temper embrittlement of a Ni-Cr alloy steel that was heat-treated to produce coarse- and fine-grain sizes. The alloy was tempered at 650°C (1200°F) and aged for various times at 500°C (930°F). (a) Actual 100% fibrous FATT. (b) Change in 100% fibrous FATT.163

14.11.2.2 Detection and Measurement of TE. Auger electron spectroscopy (AES) and DBTT have been the widely accepted tools used to detect and measure the embrittlement susceptibility. AES application to embrittlement studies allows the direct chemical analysis of impurity segregants on the intergranular fracture surfaces of embrittled specimens and alloy element segregation such as Ni at these boundaries, which act as stimulant for impurity element segregation to the prior austenite grain boundaries. The degree of enrichment of impurity elements may be 100 to 1000 times the bulk concentration, while the concentration of alloying elements may be only 2 to 3 times that of the bulk concentration, and the concen-

14.78

CHAPTER FOURTEEN

FIGURE 14.51 Variation of transition temperature with intergranular concentration of (a) phosphorus and (b) tin, expressed in terms of Auger peak-height ratio with respect to iron in Ni-Cr steel. The prior austenite grain size (C, coarse; M, medium; F, fine) and the Rockwell C hardness were also varied.164 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.; after C. J. McMahon, Jr.)

tration profile from the grain boundary into the grain interior is usually much shallower than for the impurity element. Figure 14.52a shows an example of Auger analysis of Sb, S, and P segregated to either fracture grain boundaries or free surfaces.165 These results have been accomplished by alternate argon ion sputtering (depth profiling) and analysis.165 An alternative method is the measurement of area fraction of intergranular facets on the fracture surface; all these measurements are directly proportional to the concentration of segregated impurities on prior austenite grain boundaries.166 Monitoring acoustic emission activity may also be a very sensitive method of detecting the occurrence of TE in A 533B, which is a nuclear-grade MnMoNi low-carbon low-alloy steel.167 To identify RTE in the in-service embrittled CrMoV steel turbine bolts, Bulloch and Hickey168 have developed an embrittlement estimative diagram (EED) by plotting grain size and grain boundary area Sv versus percent P (Fig. 14.52b). This exhibited two distinct regimes, i.e., the embrittled regime and nonembrittled regime, which were separated by a critical embrittled-nonembrittled interface which would be represented by the following equations: d(%P) = CRTE and

Sv = 21.4 ¥ (%P)

where d is the grain size in micrometers and Sv is the grain boundary per unit volume in mm-1. CRTE is called the RTE constant and is in the range of 0.12 to 0.59; an increased value of CRTE represents an increase in the resistance of a series of bolts

Distance, in. ¥ 10–6 0

1.0

0.02

0.06

S on AISI 5140 S on Fe-0.6Sb P on AISI 5140 Sb on AISI 3340 Sb on Fe-2.2Sb Sb on Fe-0.6Sb

0.8

Normalized intensity

0.04

0.6

0.4

0.2

0

0

0.5

1.0 Distance, nm (a)

1.5

Average grain size d, mm

Grain boundary area/unit volume Sv, mm–1 200 400 70

LEGEND

60

Embrittled Nonembrittled Embrittled Ref.11

Embrittled region

50

40

30

d=

0.18 %P

Grain boundary area/unit vol. Sv = 6 d

20 Nonembrittled region

10

0

0.005

0.01 0.015 % Phosphorus

0.02

(b)

14.79

FIGURE 14.52 (a) Normalized intensities of Auger peaks (as a function of depth below the surface) from Sb, S, and P segregated to grain boundaries or free surfaces (depth profiling by argon ion sputtering).165 (b) Relationship between grain size and bulk phosphorus level for the IP and HP turbine bolts with CRTE value of 0.28.168 (Note: Ref. 11 after J. J. Hickey and J. H. Bulloch, unpublished data, Electricity Supply Board, Dublin, Ireland, 1986–1987.) (Courtesy of J. H. Bulloch.)

14.80

CHAPTER FOURTEEN

to RTE during service at elevated temperatures. That is, RTE only occurs in bolts with larger grain sizes and bulk P levels. Essentially, it was illustrated that the extent of grain boundary area available for P segregation was the principal factor controlling turbine bolt embrittlement. Figure 14.52b shows the relationship between grain size and bulk phosphorus level for the intermediate-pressure (IP) and highpressure (HP) turbine bolts. Later work by Bulloch on steam-turbine casing steel bolts [composition: (0.36–0.44)C-(1.24–1.36)Cr-(0.56–0.81)Cr and (0.23–0.24)C-(1.28–1.37)Cr-(0.80– 0.84)Mo-(0.25–0.28)V], undergoing operating temperatures of 450 to 490°C (842 to 914°F) for service times from 60,000 to over 200,000 hr, led to the development of a general embrittlement law for the occurrence of RTE in the form b

d ¥ (%P) ¥ (%e ) = a where e is the accumulated service strain and b and a are scaling constants equal to 0.64 and 0.0772, respectively. When the left-hand side product becomes greater than a, bolt embrittlement is envisaged; and when less, no service embrittlement is expected.168a 14.11.2.3 Control and Prediction of TE. TE can be controlled by reducing susceptibility, which is primarily achieved by maintaining small concentration of embrittling impurities by control of raw materials and melting practice.118 However, it is not possible, from both technical and economical grounds, to lower the concentrations of some metalloid impurities to completely harmless levels. An alternative method is given in the next section. To predict the extent of TE of a particular steel such as a Ni-Cr steel, first the TE equation of a P-doped Ni-Cr steel and the McLean equation for equilibrium intergranular segregation of P in the same steel are derived experimentally and then used to construct the two-dimensional TE diagram.169 An embrittling treatment referred to as step cooling is sometimes employed to estimate the influence of extensive exposures. Step cooling comprises cooling the sample through the embrittling range in a series of steps from about 600 to 300°C (1100 to 572°F), with the time increasing progressively with the lowering temperature.118 The extended step-cooling treatment (ESCT) provides more accurate indication of the degree of RTE for new 1CrMoV rotor steels during their service lifetime.168 14.11.2.4 The Kinetics of TE. The embrittling kinetics follows the C-curve behavior with tempering time and temperature, with a minimum time for embrittlement at about 550°C (1022°F); these curves are plotted by using the Auger analysis of monolayers of P segregation at the prior austenite grain boundaries. 14.11.2.5 Segregation Theory of TE. The use of AES, a semiquantitative analytical tool for measuring segregated elements at the embrittled grain boundaries, has confirmed the simultaneous segregation (or cosegregation) of metalloid impurities and alloying elements (Mn and/or Cr) to high-angle austenite grain boundaries in the temper-embrittled condition.170 This cosegregation theory can be employed satisfactorily to elucidate the embrittlement after long exposure times of steels at 450 to 500°C, but it does not interpret well the two main features of TE, namely, (1) susceptibility to cooling rate after tempering and (2) the capacity to return to the ductile (i.e., deembrittled) state from the embrittled state on

TEMPERING

14.81

reheating the steel to 600 to 700°C (1112 to 1292°F) followed by rapid (water) quenching. Site competition theory addressed a lack of evidence for the cosegregation model in the ternary Fe-based alloys. According to this mechanism, a site competition between C and P at the grain boundary is assumed. Consequently, activities of carbon in the ferrite matrix, precipitation of carbides and phosphides, partitioning of alloying elements between matrix and carbides, and presence of dislocation around precipitates should be the primary factors that influence the P segregation in the Fe-C-P-M system. This theory agrees well with the experimental results observed by Janovec et al.171 in CrMoV steels. Lei et al.172 have concluded that En steels such as 40 Mn2Mo, 40Cr, and 40 CrNi steels are susceptible to TE and that this embrittlement is always associated with lowering of the Köster peak heights of the internal friction curves. The Mo addition to the steels as in 40 CrNiMo and 40 Mn2Mo, and NiCrMoV significantly improves grain boundary cohesion by interacting with carbon,167 which inhibits the TE and simultaneously decreases the Köster peak heights in slowly cooled (brittled) or aged (embrittled) states. The reversibility of TE in En steels such as 30 CrMnSiNi 2 and 40 CrNi steels is closely associated with the reversibility of the internal friction behavior. Room-temperature impact toughness values are linearly related to the Köster peak height. This suggests the aging mechanism of asolid solution to be the cause of TE, which causes the dead pinning of dislocations by ultrafine Fe3C(N) precipitate particles on slow cooling after tempering or on isothermal embrittling treatment. The dissolution of these ultrafine particles into the a-phase occurs on reheating and holding just a few minutes at about 600°C or above. Subsequent rapid cooling to below about 300°C (570°F) eliminates the TE; that is, ductility and impact toughness values of the embrittled steels are restored. The aging mechanism can be employed to elucidate the short-time TE, while McMahon’s segregation mechanism is adequate to explain the TE after long-time exposure of boiler or steam engine parts at temperatures higher than 450 to 500°C (842 to 932°F).172

14.11.3 Secondary Hardening Embrittlement The embrittlement occurring after tempering in the secondary hardening range is referred to as secondary hardening embrittlement (SHE). SHE in tungsten- and molybdenum steels is of two types, intergranular and transgranular. Intergranular SHE is associated with the impurity segregation and leads to easy intergranular fracture while the transgranular SHE is caused by coarse boundary carbides, resulting in easy transgranular fracture.173

14.11.4 Aluminum Nitride Embrittlement When an Al-killed plain carbon steel (or Al-stabilized low-alloy steel) containing increased levels of aluminum and nitrogen is slowly cooled from high austenitizing temperature ~1300°C or from solidification, precipitation of long and dendritic- or plate-shaped AlN phase occurs along grain boundaries covering a large fraction of the grain boundary area. AlN dendrites form from the liquid near the completion of solidification and may act as nucleation sites for plate-like AlN that precipitates after solidification and appears as small, shiny fracture surface facets. This leads to

14.82

CHAPTER FOURTEEN

FIGURE 14.53 Thin aluminum nitride particles (arrows) extracted from the intergranular fracture surface of a medium-carbon steel casting. Thick/dark particles are carbides. Extraction replica electron micrograph. 82,500X; reduced to 75%. (Courtesy of G. Krauss.)

a drastic reduction in toughness and can cause intergranular fracture along the prior austenite grain boundaries which have been weakened by the existence of AlN. The resulting intergranular fracture is sometime called rock-candy fracture because the coarse intergranular facets of the castings produce a macroscopically crystalline appearance.119,174 This can result in catastrophic failures in castings, panel cracking in ingots, and reduced hot ductility. Figure 14.53 shows the thin aluminum nitride and thick carbide particles extracted from the intergranular fracture surface of a medium-carbon steel casting.32 The minimum amounts of AlN necessary to produce intergranular fracture for plain carbon and alloy steels are 0.004 and 0.002%, respectively. AlN embrittlement in castings, in both the as-cast and heat-treated conditions, can be minimized or eliminated by119,175–177 1. Additions of Ti, Zr, B, S, Mo, Ni, or Cu 2. Use of the lowest possible amount of nitrogen (0.005%) and minimum requisite amount of aluminum (0.015 to 0.030%) for deoxidation 3. Increased cooling rate after solidification 4. Faster cooling rate in the range of 1150 to 700°C (2100 to 1290°F) after solutionizing at high austenitizing temperature to control the amount and size of AlN precipitation Panel Cracking in Ingots. Panel cracks are longitudinal surface cracks on the side face of an Al-killed (0.4 to 0.7%) carbon steel ingot (or a low-carbon alloy steel ingot) that usually form (possibly below 850°C or 1560°F) near the center of the face and extend up to the midradius of the ingot.178 These carbon levels produce ferrite grain boundary network film containing mostly pearlitic matrix structure. The extent of susceptibility to panel cracking is a function of the melt practice and aluminum and nitrogen contents. Thus, the severity of panel cracking decreases in the following order: electric arc furnace steel, basic open-hearth steel, basic oxygen furnace steel, and acid open-hearth steel. Panel cracking does not occur with less than 0.015% Al and 0.005% N. Small ingots are less prone than large ingots. Stripping of the ingot at a permissible high temperature perhaps reduces the susceptibility.119

TEMPERING

14.83

FIGURE 14.54 Elevated-temperature tensile test results for five plain carbon steels with various levels of aluminum nitride. The nitrogen level (in ppm) of the steels in the form of aluminum nitride was: A, 80; B, 70; C, 72; D, 2; E, 1. (Source: Ref. 181.)

Reduced Hot Ductility.179–183 Increased levels of Al (>0.03%) and N (~ > 0.01%) have been found to degrade hot ductility in low-carbon steels, En 36 alloy steels, and Cr-Mo-V turbine rotor steels in the temperature range where the volume fraction and size of the AlN precipitates are maximum. These trends were improved with the decrease of AlN particle size.119 Figure 14.54 shows the reduction in area for hot tensile tests over a temperature range for steels with different levels of soluble AlN and high levels of N present as AlN.181 The presence of high residual impurity contents, mainly Cu and Sn, played vital roles in decreasing the hot ductility in medium- and high-carbon steels.183 These impurities tend to segregate at the grain boundary ferrite networks. However, the addition of Ti and/or B in rotor steels has been observed to improve the hot ductility in the test temperature range of 800 to 1000°C (1470 to 1830°F) where high N contents are deleterious.179 It has been shown that Sn reduces the solubility of copper in austenite by a factor of 3; hence, in the presence of Sn, molten Cu can form at the surface at lower bulk Cu contents. Sn and Sb are extremely harmful to Cu-induced hot shortness. Ni reduces Cu-induced hot shortness, Mn and Cr slightly increase hot shortness, and As is slightly more harmful than Mn.119,182

14.84

CHAPTER FOURTEEN

14.12 HYDROGEN DAMAGE OF STEELS Hydrogen damage is a term that describes a number of processes in metals by which the load-carrying capacity of the metal is reduced because of the presence of hydrogen, usually in combinations with residual or applied tensile stresses. Hydrogen damage can develop in a wide variety of environments and circumstances and, in one form or another, can largely limit the use of certain materials. Although it is more predominant in carbon and low-alloy steels, many metals and alloys can exhibit this phenomenon. Hydrogen can be retained in steels and other metals internally as a result of melting and casting practices (supersaturated) and/or present externally in the atmosphere around the alloy material as a gas or a constituent of gas as a result of pickling, electroplating, cathodic process, sour environment, contact with water or other hydrogen-containing liquids or gases, and so forth. Hydrogen damage can be classified into:184 (1) hydrogen embrittlement, (a) hydrogen-assisted cracking, (b) delayed failure, (c) sulfide stress cracking, and (d) hydrogen-induced ductility loss; (2) hydrogen attack; (3) shatter, cracks, flakes, and fish eyes; (4) blistering; (5) metal hydride formation; (6) microperforation; and (7) degradation in flow properties. These are discussed below.

14.12.1 Hydrogen Embrittlement Hydrogen embrittlement has been found in various metallic materials. It is of great concern for the use of advanced high-strength materials in a wide range of hightechnology applications. Hydrogen embrittlement (HE) is a process resulting in the degradation in any one or more of a number of mechanical properties such as ductility, work-hardening rate, tensile and yield strengths, fracture toughness, and so on, depending on its application.185,186 In general, a degradation of these mechanical properties occurs through a hydrogen-induced change in either the plastic behavior or the fracture behavior of the alloy, primarily the latter. The effect of hydrogen on the plastic behavior of an alloy is direct and is due to some hydrogen-dislocation interactions. On the other hand, the effect of hydrogen on the fracture behavior of the alloy is far less direct and can comprise any one or more of the hydrogen-metal interaction mechanisms (discussed later). Hydrogen is capable of influencing all three stages of fracture, namely, initiation, slow crack growth, and the onset of rapid, unstable fracture. All the classical modes of fracture, such as intergranular fracture, transgranular ductile fracture, cleavage fracture, or a mixed fracture, can occur in the presence of hydrogen. The extent of a particular mode depends on the specific alloy and its application. For example, mixed modes of fracture can be produced in a single specimen of high-strength steels. The segregation of impurity elements such as S, As, and Sb at the grain boundaries in the nickel alloys and steels can promote intergranular HE by changing their fracture mode and the stress intensity necessary for the occurrence of HE.185 Ti-, Nb-, or Zr-based alloys, which can form stable hydrides, tend to fracture by HE, exhibiting a stress-induced hydride formation and cleavage mechanism. HE in a-b Ti-alloy has been associated with slow tensile strain embrittlement and sustained load cracking.187 HE in the Ti-24Al-11Nb alloy is a function of microstructure, H content, and type of hydrides formed in the microstructure.188 Hydrogen-induced embrittlement in tensile tests has been reported for L12 intermetallic compounds such as those based on Ni3Al, Co3Ti, Ni3Si, and (Fe,Co,Ni)3V.189

TEMPERING

14.85

HE of high-carbon austenitic stainless steels is mostly attributed to phase instability of austenitic phase with respect to cathodic hydrogen charging.190 HE susceptibility in stainless steels has been associated with stacking fault energy (SFE) through its effect on planarity of slip and deformation twinning.191 HE is more likely a combination of several elementary steps of hydrogen, namely, surface entry or absorption, transport through the structure, accumulation or trapping, and decohesion, each being described by a different mechanism. 14.12.1.1 Hydrogen Entry, Transport, Trapping, and Resultant Embrittlement. Hydrogen may be derived from hydrogen gas molecules, dissociated hydrogen molecules or atoms, or hydrogen-containing molecules such as H2S, H2O, or methanol. The slow crack growth behavior or severity of embrittlement observed in a high-strength martensitic steel is quite different when exposed to these three hydrogen-containing environments. When hydrogen originates in the bulk of the alloy, hydrogen transport is a simple process and is most often controlled by the lattice diffusion process, which takes place by the movement of a screened proton that has given up its electrons to the electron gas of the metal. When hydrogen originates from an external environment, it is required to adsorb on an external surface, chemisorb, and enter the metal lattice as a screened proton. Of all interstitial elements, hydrogen migrates fastest in metals and alloys, particularly in iron and steel. Hydrogen transport in dislocation cores or as associated Cottrell atmosphere may be several orders of magnitude faster than lattice diffusion. The transport process is important in certain phenomena such as the development of nonequilibrium internal gas pressure.192 Grain boundary diffusion has also been suggested in the modeling of H transport during hydrogen-induced cracking of Ni and Ni-based alloys.193 Hydrogen trapping occurs at various depths and at a wide variety of locations in a microstructure, such as grain boundaries, carbide/matrix interfaces, inclusions, precipitates, voids, dislocations, dislocation arrays, and solute atoms. These traps can be either reversible or irreversible according to their binding energy for H atoms. Microadditions of alloy elements have a vital effect on H diffusion in steels through trapping mechanisms. A large volume fraction of coherent e-Ti(C,N) precipitates in microalloyed steels is the most appropriate state for the formation of irreversible trapping sites of H atoms.194 According to Takahashi et al.,195 the fine coherent TiC particles 85 kJ/mol). Any of these locations which are most sensitive to fracture probably controls the magnitude of the hydrogen-induced changes. In many instances, trapping can be beneficial due to reduction of local concentration potential crack nuclei.192 A summary of the generalized processes which occur during HE is shown in Fig. 14.55.192 These processes are influenced by temperature, stress state, and microstructure.197 14.12.1.2 Theories of Hydrogen Embrittlement. Hydrogen embrittlement theories abound. Most theories involve the generation of atomic H at the crack tip with its subsequent absorption into the material and volume diffusion to specific microstructural sites. The accumulation of H at these sites results in a reduction in the work to fracture.198 Hydrogen-metal interaction mechanisms can be grouped into the following categories:186,192,199,200

CHAPTER FOURTEEN

14.86

H + + e– = H

H2 = 2H

HX + M

MX + H

[H]

Dislocation transport

Diffusion

Lattice

Grain boundaries

Cleavage

Incoherent precipitates

Dislocation tangles

Intergranular

Voids and pores

Coherent precipitates

Ductile

FIGURE 14.55 Summary of hydrogen sources, transport, and microstructural locations with corresponding end processes. The dashed line at the bottom refers to cleavage of hydrides.192 (Reprinted by permission of Pergamon Press, Plc.)

1. Local hydrogen pressure mechanism. Hydrogen-dislocation interaction either hinders dislocation movement or provides localized hydrogen pressure at grain boundaries, thereby embrittling the lattice (by reducing the stress required to initiate voids or increasing the growth rate of voids). Initially it was suggested by Bastien and Azou201 and was later modified by Tien et al.202 2. Hydrogen-enhanced local plasticity (HELP) mechanism or slip softening mechanism. Absorption of hydrogen is used to improve the generation of dislocations, mobility of dislocations, or both. This model depends on a decrease in general

TEMPERING

14.87

plasticity (or flow stress) due to the localization of H at the crack tip, where it is absorbed or drawn (or migrated) into the crack tip region by the hydrostatic stress field or perhaps by dislocations moving inward from the crack tip. This HELP mechanism of embrittlement, initially reported by Beachem,203 and later expanded by Lynch204 and Birnbaum and coworkers,205,206 differs, in general, from the previous mechanisms, in which hydrogen has been assumed to improve deformation behavior locally rather than actually embrittling the lattice.207 3. Decohesion mechanism.208–210 Interaction of dissolved hydrogen reduces the cohesive force between the atoms at high-stress region near the crack tip, leading to a decrease in stress required for fracture and eventually causing breaking of the atomic bond of the lattice at a crack tip, matrix/particle interface, or grain boundary. This model was proposed by Troiano208 and later extended by Oriani.209 Although observations of plasticity at the crack tip are consistent with this mechanism, the amount of plastic zone is assumed to decrease with the cohesive strength. The decohesion mechanism also holds for interfacial fracture. For the microvoid coalescence process, the decohesion mechanism could be employed to characterize the reduction in cohesive strength of the carbide/matrix interface. According to this theory, the embrittlement effect of hydrogen is related to its concentration in the region of the steel where brittle cracking takes place. The equilibrium concentration of hydrogen CH in the stressed crystal lattice depends on the hydrostatic tensile stress sh [= (s11 + s22 + s33)/3], according to the thermodynamic relation211 CH = C0 exp

s hV RT

(14.16)

where sii denotes the three principal stresses; V is the partial molar volume of hydrogen in the solid solution, assumed to be 2.1 cm3/mol for iron and steel,212 1.72 cm3/mol for alloy 690, and 1.8 cm3/mol for X-750 alloy;213 R is the universal gas constant; T is the absolute temperature in Kelvin; and C0 is the equilibrium hydrogen concentration in the unstressed lattice (region), which is related to the square root of hydrogen pressure pH2, for a gaseous atmosphere, in the form214 C0 = 0.00185( pH 2 )

12

-Qs ˆ expÊ Ë RT ¯

(14.17)

where Qs is the heat of solution of hydrogen in iron, assumed to be 28.6 kJ/mol.215 To estimate sh in front of a precrack in terms of tensile yield stress sy of the material, the following relation based on the elastic-plastic stress analysis of Rice and Johnson216 should be used: s h = 2.42s y

(14.18)

This theory predicts that in a given steel the Kth value for detectable cracking is a function of CH, which, in turn, is a function of hydrogen pressure pH2 or yield stress sy. This dependence is analogous to the metalloid impurities effect observed in many steels. The metalloid impurities, however, segregate during heat treatment rather than during the application of stress at room temperature.217 4. Surface energy mechanism. Reduction in surface energy due to hydrogen adsorption needed to form a crack is due to Petch.218

14.88

CHAPTER FOURTEEN

5. Internal pressure mechanism. There is recombination or precipitation of molecular hydrogen as a gas bubble and associated development of pressure at internal defects such as microvoids at the inclusion/matrix interface. This pressure, when added to the applied stress, leads to a decrease in apparent fracture stress. This mechanism, first suggested by Zapffe and Sims,219 favors premature stressassisted fracture. 6. Hydride formation. There is precipitation of brittle, less dense, metal hydride and its subsequent cracking near the crack tip. In actual practice, a considerable overlap exists between these mechanisms; more than one or different combinations of mechanisms may dominate for different mechanical or environmental conditions.192 For example, hydride can form along dislocations in the same manner as Cottrell-type hydrogen atmosphere and can change the ease and character of deformation in a structural alloy. Similarly, hydrides can preferentially form in front of a crack tip or a notch at the point of increased stress. Several authors have provided brief reviews of the theories of HE and have shown that no one simple theory is able to explain the hydrogen degradation. In a broader sense, the decohesion theory encounters most of the observed phenomena. 14.12.1.3 Forms of Hydrogen Embrittlement. 1. Hydrogen-Assisted (or -Induced) Cracking, Hydrogen-Stress Cracking, and Static Fatigue. In general, hydrogen-induced cracking (HIC) occurs when the hydrogen atoms which are produced on the steel surface penetrate into the steel and precipitate at the inclusion/matrix interfaces. It has been suggested that large inclusions such as elongated MnS and clusters of oxide promote the HIC susceptibility. This type of cracking has the following characteristics:220 (1) Cracks occur at sustained loads below the yield strength of the material and mostly in low-strength, ductile steels (or alloys). (2) All cracks are planar-oriented and occur with exhibiting tensile stress. (3) They are associated with brittle fracture and produce sharp singular cracks when compared to the extensive branching developed in stress corrosion cracking.184 When a steel structure containing dissolved hydrogen and/or exposed to hydrogen environment is subjected to a sustained load in service applications, it may fail at a stress level far below its tensile strength, as measured in a short-time notch tension test. This behavior is variously termed delayed failure, delayed low-stress brittle failure, hydrogen-stress failure, hydrogen-induced delayed failure (or cracking), and hydrogen-induced cracking under static loading. It is a manifestation of decreased fracture stress (or strain at fracture) in the presence of hydrogen, as noted above. This involves the fracture initiation in regions of highly localized stress where hydrogen is concentrated due to increased diffusion of hydrogen into triaxially stressed sites. HIC under static loading has been recognized as the single most important design constraint for such structural applications as pressure vessels, pipelines, fasteners, and power equipment. This is also observed in many weldments. In reality, HIC (or HAC) in steel weldments is grouped into (1) weld metal (HAZ) cracking, from the location of initiation; (2) macro- (or micro-) cracking, from its size; (3) longitudinal (or transverse) cracking, from the direction of its propagation to the welding direction; (4) root cracking, heel cracking, toe cracking, or underbead cracking, from the weld location of initiation and propagation; and (5) restraint (or distortion) cracking, from the restraint condition.220a

TEMPERING

14.89

RATE OF SLOW CRACK GROWTH

STAGE III

STAGE II

STAGE I

K th

K IC

APPLIED STRESS INTENSITY, K I (a) 10–5

Crack growth rate, m/cycle

Decreasing ␯

10–7

10–9

Increasing R

10–11

K Tmax

HYDROGEN GAS Increasing R

AIR(independent of ␯ and R)

AIR(independent of ␯)

DK th(H2) DK th(air) Log (cyclic stress intensity) (b)

FIGURE 14.56 (a) Usual form of the hydrogen-induced slow crack growth rate for mediumto-high-strength steels as a function of applied static stress intensity.186 (b) Crack growth rate behavior in gaseous hydrogen and in air as a function of applied cyclic stress intensity.223 [(a) Reprinted by permission of Academic Press, Inc., New York; (b) reprinted by permission of The Institute of Metals, England.]

HIC can be controlled by reducing the amount of segregation of impurity elements at grain boundaries and inclusions and by accomplishing inclusion shape control.220,221 Hydrogen-induced slow crack growth under static loading in steels and other alloy systems exhibits the classical form of stress-intensity dependence, as shown in Fig. 14.56a,186 where applied stress intensity for mode I (tensile) loading, represented by KI, the stress intensity factor, is given by K I = constant ¥ s a

(14.19)

14.90

CHAPTER FOURTEEN

where a is the effective crack length for a given applied stress s. This plot consists of four regions. First, there is the threshold stress intensity level Kth below which crack growth is either very slow or nonexistent. The Kth value is believed to represent the maximum hydrogen degradation attainable in a material for a specific set of conditions. In high-strength steels, Kth is a function of the strength level of steel, temperature, and hydrogen pressure or fugacity, which indicates an equilibrium relationship between the hydrogen about or around the crack tip and that in the bulk of the alloy. In an a-phase titanium, however, Kth is independent of temperature, which indicates a maximum degree of degradation attainable by another hydrogeninteraction mechanism [category (6) above “hydride formation”].222 Second, there is a region where slow crack growth rate increases rapidly with KI and competes with the time-dependent hydrogen transport process (Stage I). Third, there is a region where the crack growth rate is nearly constant over a substantial range of KI (Stage II). This is the result of the hydrogen transport process where slow crack growth is controlled by the rate of hydrogen transport to the area near the crack tip. Fourth, there is a region where crack growth rate increases rapidly as KIC is reached (Stage III). In this regime, a failure criterion holds because of the increased mechanical contributions of the applied stress, and unstable fracture follows.186,192 Lower-strength steels were originally considered to be resistant to hydrogeninduced brittle crack growth under static loading.221 However, both low-strength and high-strength steels have been found to exhibit crack growth under cyclic loading ratio Kmin/Kmax = R. Figure 14.56b223 shows the general trends in the fatigue behavior of lower-strength steels in hydrogen gas when compared with air. Hydrogen tends to influence two important regions of crack growth behavior: (1) thresholdreducing DKth and (2) growth measured by the Paris law, causing a steep rise in growth rate above a critical Kmax level. The latter effect is usually associated with the onset of intergranular fracture.223,224 Yurioka and Suzuki224a have provided a review on hydrogen-assisted cracking in C-Mn and low-alloy steel weldments. 2. Delayed Failure. Above a tensile strength level of 1241 MPa (180 ksi), most high-strength low-alloy steels, notably AISI 4130 and 4340, and precipitationhardening stainless steels are prone to hydrogen stress cracking in marine environments, when the applied or residual tensile stresses are quite high, and cracking normally takes place as a form of delayed failure.184 The delayed-failure behavior of steel has been extensively studied using static loading of precharged notched sample. Figure 14.57199 shows a schematic representation of delayed failure in a high-strength steel, illustrating two stress-time curves, one for crack initiation and another for failure. These curves are bounded by two stress levels, an upper critical stress (UCS) and a lower critical stress (LCS), which correspond to the maximum and minimum stress levels, respectively, causing delayed fracture. At stresses above the UCS, failure takes place without a time delay; and at stresses below the LCS, hydrogen is not harmful. At intermediate stress level, failure occurs as a series of events consisting of crack initiation following an incubation period and crack propagation. The important factors responsible for the delayed-fracture behavior of precharged steels are the hydrogen concentration (or potential or fugacity), the strength level or composition of steel, notch acuity, temperature, and grain boundary impurity concentration. An increase in either hydrogen concentration or the strength level usually causes a decrease in both the UCS and the LCS and leads to shorter delay times for failure.199 Delayed failure of high-strength ferritic steels in hydrogen is believed to occur by diffusion of internal or external hydrogen to a region of high local constraint ahead of the crack tip.

TEMPERING

14.91

FIGURE 14.57 Schematic illustration of delayed-failure behavior of notched tension high-strength steel specimens containing hydrogen.199 (Reprinted by permission of ASM International, Materials Park, Ohio.)

Delayed hydrogen cracking behavior has also been observed in zircaloy-2 pressure tubing during reactor service, where H is generated by corrosion processes and absorbed by these materials and eventually leads to precipitation of zirconium hydride platelets.225 grain boundary impurity concentration. It is generally recognized that hydrogen-induced delayed failure (or cracking) occurs in quenched and tempered medium- and high-strength steels when stressed in a relatively low-hydrogen fugacity. It has been found that the stress intensity Kth required to produce a detectable amount of crack extension decreases with the increasing grain boundary impurity (such as P, S, and Sb) concentration, which lowers the intergranular cohesion. This is accompanied by a shift in cracking mode from displacement-controlled transgranular fracture at a high Kth level (in the unembrittled condition) to the stresscontrolled intergranular fracture mode at a low Kth level (in the embrittled condition).226,227 For a given yield strength and hydrogen pressure (or concentration), this type of hydrogen-induced delayed failure is believed to be mainly an impurity effect. The hydrogen and impurity effects tend to be simply additive. This is the minimum effect which can be expected, but no such details have yet been established.228,229 experimental results. In a study of HY130-type steel (composition: Fe-5Ni0.5Cr-0.5Mo-0.1V-0.1C) using static loaded edge-notched and precracked cantilever beam specimens, Yoshino and McMahon226 demonstrated that TE by step cooling reduced—drastically the Kth for crack growth from — a high level, ~104.5 MNm-3/2 (95 ksi÷in .), to a quite low level, ~22 MNm-3/2 (20 ksi÷in .) (in a cathodically charged hydrogen atmosphere), in a 0.1 N H2SO4 solution and shifted the mode of cracking from transgranular fracture to a completely intergranular mode. They interpreted their results, in terms of Oriani’s theory of HE, to occur from an interaction between

14.92

CHAPTER FOURTEEN

FIGURE 14.58 (a) Examples of crack growth data for samples aged for 50, 200, and 1000 hr, showing the decrease in crack growth rate as Kth is approached. (b) Crack growth rate data plotted on an expanded time scale show clearly the discontinuous nature of crack growth.228 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

hydrogen-induced (delayed) cracking and weakening of grain boundaries by embrittling impurities. In another study, Briant et al.228 used precracked bolt-loaded, wedge-opening-loaded (WOL) HY130 steel samples (1T-WOL) at fixed displacements, to determine the crack length near the Kth level. They measured the variation in Kth and crack length with time in these specimens aged for various times (i.e., grain boundary concentrations) and stressed to various initial K levels at a fixed temperature and hydrogen pressure (Fig. 14.58). These data show that hydrogeninduced (delayed) cracking proceeded in a stepwise fashion. All these results support the hypothesis that hydrogen-induced (delayed) cracking at low Kth levels in these quenched and tempered steels is stress-controlled intergranular decohesion

TEMPERING

14.93

FIGURE 14.59 Variation of crack growth rate with stress intensity obtained from bolt-loaded WOL specimens (segmented curves) and from CT specimens (individual data) points.228 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

which is essentially related to the presence of embrittling impurities in the grain boundaries. In the absence of the impurity effect, cracking would occur by displacement-controlled transgranular mode at high Kth levels, often referred to as quasi-cleavage. The hydrogen-plus-impurity effect can be rationalized in terms of the Oriani theory. To measure the crack velocity at K > Kth, compact test (CT) specimens were used at a fixed load, and the stress intensity was increased with an increase in crack length. Figure 14.59 shows macroscopic average crack velocity V versus stress intensity, illustrating well-defined Kth values below which the crack growth rate dropped below 10-6 m/s. Thus Kth values definitely decreased with the increased aging times (i.e., increased grain boundary impurity concentration). At K values just above Kth, the region of steeply rising crack velocity V is called the Stage I crack growth, while the regions where crack velocity appears to level off represent the Stage II crack growth.228 In 4340 and 300 M steels, following TME, similar studies have shown that highpurity Ni-Cr-Mo-C steel exhibits transgranular fracture mode at a higher Kth value, representing the intrinsic effect of hydrogen in these steels; the commercial grades, however, show greater susceptibility to hydrogen-induced delayed cracking at low Kth ( 1% are not suggested for service in sour environments.232 The SSC susceptibility of weldments seems to be greater than that of the base metal, and the high hardness and residual stresses arising from welding are believed to enhance the susceptibility. Recently, National Association of Corrosion Engineers (NACE) has issued a series of guidelines for safe operations in sour environments (NACE MR-0175-93). Accordingly, the strength of corrosion-resistant alloys (CRAs) for sour service has been restricted to maximum yield strength of 690 MPa and maximum hardness of 22 Rc.232 Guidelines for dealing with hydrogen stress cracking that occurs in refineries and petrochemical plants have also been developed by the American Petroleum Institute (API).184 High-strength low-alloy steels such as C-90, C-100, and C-120 grades with improved SSC resistance have been used for the last two decades. These steels have minimum yield strengths of 620 MPa (90 ksi), 690 MPa (100 ksi), and 830 MPa (120 ksi), respectively; and they are used as oil- and gas-well tubing, casing and coupling, and advanced tool joints for deep sour-well drilling and production.233 Recently, the SSC resistance of API X-80 steel has been improved by microstructural modification involving water quenching from annealing temperature (852°C) and tempering for 1 hr at 600°C. This improvement is attributed to the increased number of H traps provided by the resultant fine distribution of (Nb,Ti)-containing carbides, which causes a tortuous crack path along ferrite interlath grain boundaries.234 The development of X-100 linepipe steel has also been reported by Okatsu et al., based on optimum microstructural control by thermomechnical treatment.234a 4. Hydrogen-Induced Ductility Losses. The presence of dissolved hydrogen can often result in a loss of ductility, as measured by a decrease in the reduction of area (RA) value in a smooth tensile test specimen. Such ductility losses may take place with or without a change in the fracture mode, compared to a hydrogen-free test. This mode of failure is mostly observed in lower-strength alloys, and the embrittlement index, often quoted by the percent RA loss, is given by RA loss =

RA - RA H ¥ 100 RA

(14.20)

where RA and RAH denote the values for uncharged and hydrogen charged specimens, respectively.

TEMPERING

14.95

In addition to the effect of the hydrogen concentration, the measured tensile properties are affected by the temperature and strain rate of the test, the stress concentration of the specimen, the extent of prior cold work or work-hardening rate, and the strength level of the steel. Loss in tensile ductile behavior is more pronounced when strain rate decreases, from a normal strain rate (i.e., at 10-4 s-1) to a slow strain rate (i.e., on the order of 10-7 s-1 and below) testing. Hence, this behavior can pose the potential service problems under static loading and quasi-creep conditions,192 but not during impact tests, such as the Charpy V-notch test. The temperature of minimum ductility has been found to increase with the increase in strain rate. This sensitivity to strain rate and temperature clearly demonstrates that the mechanism which leads to the ductility changes involves transport and trapping processes.199 As the notch acuity increases, the reduction in notch strength at a given value of the other factors will be greater due to increased strain localization. Although a function of temperature, the work-hardening rate increases in the presence of hydrogen, with a maximum influence near ambient temperature.192 In higher-strength steels, comparatively small hydrogen concentrations can produce large property changes, while in lower-strength steels the effect of hydrogen decreases. 14.12.1.4 Control of Hydrogen Embrittlement. Hydrogen embrittlement (HE) of structural alloys is very complex and very specific, and it can be influenced by a large number of variables. To control HE in any structural applications, the HE process must be understood from the start to the end.186 Numerous steps are used to control the HE: 1. Melting and casting practices and subsequent finishing operations such as pickling, plating, heat-treating, and welding should be ensured so that steel is free of residual internally dissolved hydrogen. For example, flaking can be eliminated by using vacuum stream degassing in steelmaking where the hydrogen content is held below 2 ppm. 2. To keep steel surface away from the accelerated entry of hydrogen, all hydrogen produced on the surface should be prevented from entering the material in deep sour oil and gas wells, which produce improved resistance to SSC.235 3. To control HE in applications involving aqueous pit and crevice corrosion, hydrogen concentration should be reduced by means of electrochemical reactions.217 4. Embrittlement becomes more severe with an increase in strength level, and strengths below 700 MPa do not show any marked embrittlement. Improved alloy design (rather than relying only on external hydrogen control) such as C-90, C-100, and C-120 high-strength steels should be used in tempered condition in petrochemical and ammonia industries, which will substantially decrease the susceptibility of steel to SSC. 5. C, Mn, and Si shift the crack-tip metal solution potential toward cathodic values so they increase steel sensitivity to HE. Hence, unnecessary increase of C, which induces low Kth values, should be avoided, and Mn and Si should be used with caution.122a Cr addition, which promotes grain boundary segregation of P, should be reduced. Mo and Ti should be used, which scavenge P and Al, which scavenges N. Increased Ni content, which promotes inherent toughness of the material, should be used.230 Concentrations of sulfur and trace impurities in high- and low-strength steels should be reduced, and the addition of rare earth elements, such as La and

14.96

CHAPTER FOURTEEN

Ce, in amounts above 0.15 wt% should be used to improve substantially the resistance to HE. 6. Microstructural control. The microstructure and phase distribution plays a key role in determining the kinetics of HIC.236 Fine and uniform structures of quenched and tempered steel and bainite are the most resistant to HE; spheroidized structure, intermediate resistant; ferritic or pearlitic (normalized) structure, somewhat less resistant; and untempered martensitic structure, the least resistant. The banded microstructure in ferrite-pearlite steel, being rich in Mn, is susceptible to HE. For example, carbon steels in the quenched and tempered conditions are more resistant to hydrogen attack (HA) than those in the annealed condition. This suggests that the presence of discrete carbide particles rather than continuous carbide films along the boundaries decreases the susceptibility to HA.237 Among the quenched, ferritic, and austenitic structures, the austenitic structures are just a little susceptible to HE. This is due to their higher H solubility, low H permeability, usual low sensitivity to notch effect, and low yield strength.238 SSC resistance is increased by (1) fine prior austenite grain size, (2) fully asquenched martensitic structure, (3) tempering at high temperature, (4) rapid cooling from tempering temperature,239 and (5) thermomechanical treatment. Shape control of sulfide inclusions from crack-like flat to globular form increases substantially the steel’s resistance to hydrogen-induced blistering and SSC.239 7. Metallurgical modification. Methods for mitigating HE by metallurgical modification include alloy additions which (a) favor adhesion at the solid/solid interfaces [e.g., by segregation of elements which tend to improve cohesion (B, C, or N) by displacing decohesion-improving elements (P, S, Sn) and consequently increase the critical H level needed to produce interfacial decohesion, or by interfacial segregation of elements which displace H from interfaces],240 (b) reduce H entrance, thereby reducing the dissolved H concentration in the lattice,241 or (c) force a redistribution of H trap that is partitioned between lattice sites and trapping sites, so that the critical H content required to induce interfacial decohesion is not easily reached at the same H fugacity.242 In this way, Pd addition up to 1 wt% is shown to significantly change the hydrogen-assisted cracking phenomenon of PH 13-8 Mo stainless steel by suppressing intergranular hydrogen cracking.243 Other workers have also obtained similar results in other steels (e.g., quenched and tempered AISI 4130 steel).244 8. Adsorption of O2, SO2, and CO, as well as organic inhibitors, appears to be effective in some metallic systems to hinder gaseous HE. 9. Stresses. A change in deformation mode, extent of cold-work, and increased applied stress can have a large influence on HE.245 10. Passive oxide coating consisting of TiO2 and Al2O3 on Ti-45 at% Al [twophase titanium aluminide, Ti3Al (a2) + TiAl (g)] and Ti-50 at% Al alloys has been observed to be effective in preventing H penetration (or occlusion).246 In the case of iron aluminide, small addition of Cr is effective in minimizing HE and improving ductility. 11. Thermomechanical treatment (TMT). High-temperature thermomechanical treatment (HTMT) reduces the susceptibility of steel to HE by favoring a change from intergranular to intragranular fracture mechanism. Presumably, impurity concentration on grain boundaries falls during the hot deformation step of HTMT due to its redistribution between boundaries and the substructure, which improves the resistance to development of grain-boundary crack.247 It has also been shown that TMT plays a significant role in improving the mechanical properties of Fe-25Al-1B intermetallic alloy.248 Agarwal et al.249 demonstrated that partially recrystallized

TEMPERING

14.97

microstructure achieved by TMT in Fe-25Al iron aluminide also prevented H ingress through grain boundaries, thereby minimizing HE. 12. Reversibility of HE. HE, attributed to plating and cleaning, is reversible in that damage can be eliminated by baking treatment which disperses, diffuses out, or removes H, and consequently the time to failure and the lower critical stress limit increase. Appropriate baking time is a function of material hardness, plating processes, coating type, and coating thickness.250 14.12.2 Hydrogen Attack Prolonged exposure of pressure vessel or piping steels to high-pressure hydrogen at elevated temperatures (>200°C) such as in petrochemical plant equipment† and hydrosulfurization reactors can develop a network of internal cracks and result in external surface decarburization by contact with H and internal decarburization by H permeation within the steel and consequent reaction with iron carbides to form methane gas bubbles, mainly along the grain boundaries. Driven by the (internal) methane pressure, cavities grow due to grain boundary diffusion and dislocation creep. This results in a progressive development or linking up of cavities to form intergranular fissures (or cracks) at grain boundaries or enlarged pores in the metal matrix. In cold-worked steels, bubbles can form within grains, probably initiating at voids formed at carbide/ferrite interfaces.250a,251 This link-up of bubbles or cavities ultimately produces a substantial amount of swelling (breakaway stage) or blistering and premature failure at elevated temperature. This mode of material degradation, occurring at elevated temperatures, is called hydrogen attack (HA) and produces irreversible damage. The severity of hydrogen attack is a function of temperature, hydrogen partial pressure, exposure time, stress level, steel composition, and microstructure of the steel, especially its alloy carbides MxCy. Internal decarburization and surface decarburization both result in marked reduction in strength; however, the former will tend to decrease the ductility of a steel, while the latter will tend to increase the ductility. Thus hydrogen attack can be a limiting design problem in both petroleum and synthetic ammonia industries.192 This process of decarburization may accelerate cavity growth. Fissures by HA are initially microscopic, and in advanced stage a large number of fissures lead to a large reduction in mechanical properties. A complete decarburized and fissured steel component may have its UTS of 41.3 MPa (60 ksi) reduced to 172.4 MPa (25 ksi) and its test bar elongation reduced from 30% in 2-in. (51-mm) sample to nil. Figure 14.60a is a microstructure of a C-0.5Mo steel sample damaged by HA comprising initial decarburization and cracking when exposed to service conditions of 421°C (790°F) at hydrogen partial pressure of 2.93 MPa (0.425 ksi) for approximately 65,000 hr in a catalytic reformer.252 The conditions under which different steels can be employed in hightemperature hydrogen service are listed in the American Petroleum Institute document 941, the January 1997 edition. Figure 14.61 shows the recent and modified Nelson curves developed by API. The main data are presented by a set of empirical operating curves on a T-P plot, known as Nelson curves which have been used by the industries for over 66 years252,252a to keep the hydrogen attack under control. These solid curves253 provide the operating limits for plain carbon and low-alloy steels for various combinations of temperature and hydrogen partial pressure, above † They handle hydrogen-hydrocarbon streams at pressures and temperatures up to 21 MPa (3 ksi) and 540°C or 1000°F, respectively.

CHAPTER FOURTEEN

14.98

(a)

(b)

(c)

FIGURE 14.60 (a) Microstructure of a C-0.5Mo steel (ASTM A 204–A) specimen showing internal decarburization and fissuring in hydrogen attack. Service conditions: 65,000 hr in a catalytic reformer at 421°C (790°F) and 2.95 MPa (425 psi) absolute hydrogen partial pressure; nital etch. 520X.252 (Courtesy of American Petroleum Institute.) (b) Fisheyes in as-welded E7018 tensile specimen tested at room temperature (4X).184 (c) Stepwise cracking of a low-strength pipeline steel exposed to hydrogen sulfide (6X, shown here at 65%).184

which HA associated with internal decarburization and internal cracking will be observed and below which operations may be safely conducted in periods of plant operations. Although the curves establish a useful guideline, a safer approach uses alloys with (1) the addition of carbide stabilizers such as Cr, Mo, W, Ti, V, and Nb, thereby making methane formation thermodynamically and kinetically more difficult; (2) decreased carbon content to increase its resistance to hydrogen attack; (3) elimination of slags, segregated impurities, stringer-type inclusions, or laminations; and (4) neither inclusion in welds nor HAZs since they are more prone to hydrogen attack than the base or weld metal. However, HAZ susceptibility to HA can be minimized after tempering at 690°C for 1 hr for 2.25Cr-1Mo steel.254 Additional methods to minimize hydrogen attack are as follows:

14.99 FIGURE 14.61 Operating limits developed from industrial experience for various steels exposed to hydrogen-containing environments at elevated temperatures to avoid decarburization and fissuring.252 (Courtesy of American Petroleum Institute.)

14.100

CHAPTER FOURTEEN

1. Addition of ethylene (up to 0.5%) and ethane decreases hydrogen permeation at high temperature (>200°C) in iron by a factor of 50 according to a reversible process. 2. The 1-2 dibromoethylene is even better, because it decreases the permeation rate by a factor of 90. 3. Use of low-alloy Cr-Mo-containing steels, for example, 2.25Cr-1Mo steels under hydrogen pressure of 9.8 to 24.9 MPa (100 to 300 kg/cm2) at temperatures below 450°C, which is the critical temperature of a Nelson curve, provided that the weld joints have been completely stress-relieved.255 In the case of carbon steel, Shewmon and Xue have experimentally observed that (1) high-pressure H significantly increases the rate of crack growth at elevated temperature near the Nelson curve; (2) when KI is reduced to zero, the crack growth ceases; (3) a spheroidized structure is much more resistant to hydrogen-assisted crack growth than is a normalized (ferrite + pearlite) structure; and (4) metallographic examination reveals that the fracture comprises a mixed grain boundary and transgranular mode and displays little branching.251 14.12.3 Shatter Cracks, Flakes, and Fisheyes Shatter cracks, flakes, fisheyes, hairline cracks, and white spots are common features of hydrogen damage in forgings, castings, and weldments. These internal defects in heavy sections are attributed to hydrogen pickup during steelmaking where moisture is present in the atmosphere and additives. These defects appear when steel cools below ~200°C (~392°F), at which temperature the amount of hydrogen degradation is operative.199 A thermal flake is a tight crack formed by the combined action of H and stress, always fully contained within a steel section, usually appearing as a disk or hairline crack in the central portion unless directional stresses or localized weaknesses change the shape. Flakes usually form during cooling after the first rolling or forging and not during cooling following solidification. Flakes are normally oriented within the forging grains or segregated bands. Flake susceptibility increases with the increase of H content. Medium and heavy flaking can be identified by ultrasonic pattern, using longitudinal waves.256 It is believed that hairline cracks appear in those steels melted in nonvacuum furnaces when measures such as slow cooling or isothermal annealing of slabs after rolling have not been carried out. It has been reported that HSLA steels with (1) higher Ni and Mn levels and/or (2) segregation of Mn, V, and Si are usually flakesensitive at low-carbon levels. The flakes observed in such HSLA plates are present along grain boundary and other banded low-temperature transformation microstructures.256a Fisheyes, being another form of localized HE, are described as small shiny spots frequently found on the fracture surface of tension specimens taken from steel forgings, plates, or welds (Fig. 14.60b) containing a high H level, which have the propensity to reduce tensile ductility. Fractographic examination generally exhibits fracture-initiation sites such as pores or inclusions, related to fisheyes. Baking or extended room-temperature aging of tension specimens mostly removes fisheyes and restores tensile ductility. When this type of hydrogen damage takes place in welding, it is termed underbead cracking or delayed cracking, which develops in the HAZ region of the base metal (after several hours or days following welding) and runs nearly parallel to the fusion line. (See also Secs. 3.10.3.4 and 14.12.1.3 for more details.) The factors

TEMPERING

14.101

responsible for this type of cracking are dissolved H (arising from the arc atmosphere by the shielding gas, flux, or surface contamination), low-ductility (martensite) structure, and tensile stresses. The stresses produced by external restraint and by volume changes during transformation can readily develop cracks in this region.184 14.12.4 Blistering Hydrogen-induced blistering is more common in low-strength steels, and it is observed in metals exposed to sufficiently higher hydrogen-charging conditions such as acid pickling, electroplating, cathodic processes, or in-service corrosive environments comprising H2S. If these damage processes occur at the surface or just below it in the interior, the hydrogen gas pressure in the cracks can lift up and bulge-out in blistering. When these blisters are on a line following plane precipitation, it may lead to step cracking, an opening of the wall, and a lower mechanical behavior of the steel part. These irreversible damage processes were a common problem two decades ago in such applications as rails, plated parts, and enameling steels. Proper outgassing methods and compositional control have reduced its occurrence. However, sometimes they occur in line-pipe steels exposed to gases such as H2S and moisture, which can cause high hydrogen fugacities.192 In the refinery, H-induced blistering has been observed most often in vessels handling sour (H2S-containing) light hydrocarbons and in alkylation units where HF acid is employed as a catalyst. Storage vessels containing sour gasoline and propane are most susceptible to blistering; however, sour crude storage tanks are less susceptible to blistering, perhaps due to the corrosion-inhibiting effect of oil film of the heavier hydrocarbon. In storage vessels, blistering usually occurs at the bottom or in the vapor space containing water. Gas-plant vessels in catalytic hydrocarbon cracking units are very susceptible to blistering due to the generation of cyanides by the cracking reaction. Hydrogen-induced blistering also takes place on steel plates as cathodes in industrial electrolysis. Hydrogen blistering is often associated with HE in low-strength steel subjected to H2S-containing environments in the unstressed condition. Internal hydrogen blistering on a microscopic scale along grain boundaries (fissures) can result in hydrogen-induced stepwise cracking (Fig. 14.60c). Cracking advances with the cracking of metal ligaments between adjacent fissures. In general, the severity of hydrogen blister is a function of the severity of corrosion, but even low corrosion rates can generate adequate hydrogen to cause extensive damage. In some instances, hydrogen blistering is restricted to dirty steel with highly oriented slag inclusions or laminations. Vapor/liquid interface areas in equipment frequently exhibit most of the damage, perhaps because NH3, H2S, and HCN concentrate in the thin water films or in water droplets that collect at these areas.184 14.12.5 Other Forms of Hydrogen Damage Microperforation by high-pressure hydrogen takes place in steels at very high hydrogen pressure near ambient temperature. This form of hydrogen damage exhibits a network of small fissures that favor permeation of the alloy by gases and liquids.184 Degradation in flow properties occurs in iron and steel in hydrogen environment at room temperature.184 Metal hydride formation usually occurs in hydride-forming metals such as Ti, Nb, and Zr.184

14.102

CHAPTER FOURTEEN

14.13 METAL-INDUCED EMBRITTLEMENT Metal-induced embrittlement is grouped into (1) liquid-metal embrittlement, where contact with liquid-metal embrittles a solid metal, and (2) solid-metal embrittlement, where contact with a solid metal just below its melting temperature embrittles a solid metal.257

14.13.1 Liquid-Metal Embrittlement The exposure of a normally ductile solid metal or an alloy to a liquid-metal environment results in a thin film of liquid-metal coating, and subsequent stressing in tension may cause a reduction in the ductility or fracture stress together with a catastrophic failure by brittle intergranular mode or transgranular (cleavage) mode.257–259 This phenomenon is referred to as liquid-metal embrittlement (LME). LME has received much less attention than HE. LME was first recognized in abrass (embrittled) by mercury in 1914 by Huntington.260 Since then, LME has been observed in various failure analyses.118 There are four distinct forms of LME:118 1. A sudden fracture of a certain metal occurs under an applied or residual tensile stress as a result of contact with a certain liquid metal. This is the common type of LME. 2. Delayed failure of a certain metal in contact with a particular liquid metal occurs after a fixed time period under static loading below the tensile strength of the metal. This form of LME is due to the liquid-metal penetration along the grain boundary and is not as common as the previous one. 3. Grain boundary penetration of a specific solid metal in the unstressed condition by a specific liquid metal causes the final disintegration of the solid metal. 4. Elevated-temperature corrosion of a solid metal by a liquid metal leads to embrittlement. This form is quite different from the others. LME occurs only in combinations of specific liquid-metal and specific solid metal. For example, liquid mercury embrittles Zn but not Cd; liquid Ga embrittles Al but not Mg; liquid Li, Al, Cd, Cu, brass, bronze, Sb, Te, Ga, In, Zn, and Hg embrittle steel but not Na, Se, and Th. The prerequisites for the occurrence of LME are:118,258 (1) a good intimate contact or wetting between the surface of the solid metal and the liquid metal, i.e., complete coverage by the liquid metal which is usually difficult to remove; (2) an applied or residual tensile stress; (3) some measure of plastic flow and some stable obstacle to dislocation motion (or plastic flow) at the solid/liquid interface; and (4) little or no mutual solubility and absence of intermetallic compound formation. However, there are some exceptions.257 Additional factors that promote LME in solid metal are the presence of a sharp notch or stress raiser, high strain rate, coarse grain size, and the test temperature. LME does not depend on the purity of the liquid or its presaturation with the solid or on the time of exposure to the liquid metal.257 14.13.1.1 LME of Steel. This section provides the information on the embrittlement of a wide range of steels by various liquid embrittlers.

TEMPERING

14.103

Aluminum Embrittlement. Tensile and stress rupture tests on steels exposed to liquid aluminum at 690°C (1275°F) for a short time revealed a selective attack with matrix corrosion, as reported by Radekar and others.257,258 Antimony Embrittlement. Bend tests were conducted by Shottkey et al. on plain carbon and silicon and chromium steels, and it was found that exposure to liquid antimony in the temperature range of 540 to 649°C (1000 to 1200°F) produced embrittlement. Fatigue testing of 4340 steel in the liquid Pb-35Sb at 540°C (1000°F) and in antimony at 675°C (1250°F) exhibited very severe embrittlement.257 Bismuth Embrittlement. Bismuth embrittlement of mild steel was observed by Tanaka and Fukunaga, on testing only at higher temperatures with maximum embrittlement and DBTT occurring at 350 and 550°C (662 and 1022°F), respectively. Embrittlement by Brazing Alloy. Embrittlement of mild steel by brazing alloy was observed by Genders during bend tests at 900°C (1652°F). Riede reported failures in thin-walled steel tubing during dip-brazing operations.258 Cadmium Embrittlement. Delayed failures were observed by Iwata and Asayama in a range of cadmium-plated high-strength steels such as 4130, 4140, 4340, and 18 Ni-maraging steels down to 232°C (450°F), which is about 90°C (160°F) below the Tm of Cd. Radekar reported embrittlement along grain boundaries in a series of steels which were produced by pure cadmium at 350°C (662°F). He noted that the addition of 8 and 36% Zn to the cadmium enhanced the sensitivity to embrittlement at 400°C (752°F), while additions of 0.55% Al or 2% Ni did not produce any significant effect. In the case of electroplated and vacuum-deposited cadmium, fatigue limits of 10 and 60% of room-temperature strength, respectively, have been reported at 300°C (570°F), and the catastrophic failure was associated with a transgranular ductile fracture.257 Cadmium has been recognized as a more potent solid-liquid embrittler than lead, tin, zinc, or indium.257 Copper Embrittlement. During the hot-working of some steels, embrittlement by copper was noticed by several workers258 in the range of 1100 to 1300°C (2010 to 2370°F). This occurred by diffusion-controlled grain boundary permeation of copper and associated dissolved alloying elements such as Ni, Mo, Sn, and As during oxidation. In another study made by Hough and Rolls, copper remarkably changed the creep behavior of notched pure iron specimen and caused intergranular fracture. Embrittlement was of the delayed type and occurred by penetration of copper along the prior austenite grain boundaries. The liquid copper appeared in front of the advancing surface cracks and was believed to favor the initiation and growth by reducing the cohesive strength of the boundaries and promoting grain boundary sliding.118,258 Hot tensile testing in a Gleeble testing machine at high strain rates exhibited severe copper embrittlement in AISI 4340 steel (Fig. 14.62).257 Gallium Embrittlement. Severe embrittlement of iron alloy, Fe-3Si and 4130 steels by liquid gallium has been observed.257 Indium Embrittlement. Liquid indium embrittles pure iron (only above 310°C, or 590°F), carbon steel, and 4130 steel. Embrittlement depends on both the microstructure and the strength level. Lead Embrittlement. Lead embrittlements of steel were found to be of two types: (1) external LME (i.e., embrittlement by molten lead and lead alloys) and (2) internal LME (i.e., the embrittlement of leaded steel, where lead is present as second phase or inclusions). Exposure of both pure lead and lead alloys induced external embrittlement in 4140 and 4145 steels. Additions of Zn, Sb, Sn, Bi, and Cu increase the extent of lead embrittlement. Additions of up to 0.5% Zn, 2% Sb, or 9% Sn increase the embrittlement of AISI 4145 steel; the embrittlement potency

14.104

CHAPTER FOURTEEN

FIGURE 14.62 Copper-plated specimen that was pulled at 1100°C (2010°F) in a Gleeble hot tensile machine, showing liquidcopper embrittlement of 4340 steel.257

increases with the increase of impurity level. The severity of embrittlement or embrittling susceptibility and DBTT were found to increase with an increase in both surface roughness and grain size and a decrease in the amount of cold work.258 Failures of leaded steel parts such as shafts, gear teeth, die block, and compressor disks in jet aircraft and helicopter engines were reported by Breyer and Gordon, who noted that such failures were due to the presence of lead, a temperature range of 200 to 800°C (392 to 1472°F), and lowering of tensile stress toward the yield stress. However, the extent of lead embrittlement can be either eliminated or substantially decreased by controlling sulfide morphology and composition (by the addition of rare-earth elements) and cold-working of steel. Lithium Embrittlement. Tensile tests on lithium exposure were conducted by Cordwell on mild steel and Fe-2.25Cr-1Mo and Fe-9Cr-1Mo steels in the temperature range of 200 to 250°C (392 to 480°F) and at a strain rate of 2 ¥ 10-5 s-1. The tensile ductility of mild steel exposed to lithium at 200°C (392°F) showed drastic reduction of tensile ductility in lithium, with intergranular fracture after 2 to 3% elongation, without affecting its yield stress or the initial work-hardening behavior.257 Mercury Embrittlement. The fracture toughness of notched 1Cr-0.2Mo steel was drastically decreased upon testing in mercury. The additions of solutes such as

TEMPERING

14.105

Co, Si, Al, and Ni to iron reduced the tendency for cross-slip by a decrease in the number of active slip systems and changed the slip mechanism from wavy to planar glide, which enhanced the susceptibility to embrittlement. Iron alloys (such as Fe-2Si, Fe-4Al, Fe-8Ni, Fe-20V, and Fe-49Co-2V alloys) have been reported to be embrittled by mercury in unnotched tensile tests.257 Selenium Embrittlement. Selenium did not exhibit embrittling influence on the mechanical properties of a quenched and tempered steel (UTS ~1460 MPa or 212 ksi) that was subjected to bend test at 250°C (480°F).257 Silver Embrittlement. Silver showed little effect on the mechanical properties of plain carbon steels, silicon steels, and chromium steels that were subjected to bend test at 1000 to 1200°C (1830 to 2190°F). However, a silver base filler metal (composition: 45Ag-25Cd-15Sn) has been found to embrittle A-286 heat-resistant steel in static-load tests above and below 580°C (1076°F), the Tm of the alloy.257 Sodium Embrittlement. Unnotched tensile properties of low-carbon steels were unaffected when tested in sodium in the temperature range of 150 to 250°C (300 to 480°F). Similarly, no embrittlement by sodium was observed in Armco iron, lowcarbon steel, and type 316 stainless steel in the temperature range of 150 to 1600°C (300 to 2910°F).257 Embrittlement by Solders and Bearing Metals. A wide variety of steels have the propensity to embrittlement by molten solders and bearing alloys at temperatures below 450°C (840°F). The extent of embrittlement increased with the grain size and the strength value of the steel, except in the temper-embrittled steels. In general, the embrittlement was associated with a change to a brittle intergranular fracture mode and penetration along prior austenite grain boundaries. Intercrystalline penetration of solder was not observed in (0.14 and 0.77%) carbon steel at 950°C (1740°F).257 Tellurium Embrittlement. Embrittlement by tellurium has been observed in both plain carbon and alloy steels. Hot-shortness has been reported in AISI 12L14 + Te steel, associated with drastic loss in ductility in the temperature range of 810 to 1150°C (1490 to 2100°F) and the most severe embrittlement at 980°C (1795°F) due to the formation of lead-tellurium compound film at the grain boundary, which has Tm = 923°C (1693°F).257 Thallium Embrittlement. Thallium did not exhibit embrittling effect on the properties of a quenched and tempered steel (UTS ~1460 MPa or 212 ksi) which was bend-tested at 325°C (615°F).257 Tin Embrittlement. The embrittlement behavior of tin on a range of steels can be represented by an embrittlement trough in the temperature range of 110 to 400°C (230 to 752°F); the location and extent of the trough were a function of the steel composition and the strain rate, as reported by Tanaka and Fukunaga.258 In another study made on the fatigue properties of tin-embrittled mild steel, 13% Cr steel, and 18-8 stainless steel, it was indicated that in the case of the unnotched specimens the lifetime was crack-initiation-controlled whereas in the notched specimens the lifetime was crack-propagation-controlled. Zinc Embrittlement. In many instances, zinc embrittlement cracks contain zincrich precipitates on fracture surfaces and at the crack-tip and are irregular in nature. Zinc-embrittled austenitic stainless steels are of two types. (1) Type I involves the metal penetration/erosion in the unstressed materials such as 18-8 austenitic stainless steel at 419 to 507°C (786 to 945°F) and above, type 316 stainless steel at 750°C (1380°F), and type 321 steel at 515°C (959°F). (2) Type II involves LME in the stressed materials at temperatures above 750°C (1380°F). According to Radekar, certain ferritic steels having greater thermal stability exhibited the maximum resistance to molten zinc embrittlement.258

14.106

CHAPTER FOURTEEN

FIGURE 14.63 Embrittlement behavior of cadmium-plated 4340 steel. Specimens were tested in delayed failure (a) at 300°C (570°F) and unplated steel in air at 300°C (570°F) and (b) at temperatures in the range of 230 to 360°C (445 to 680°F).262

Zinc embrittlement was also observed in ferritic steels and Armco iron in the temperature range of 400 to 620°C (750 to 1150°F) and in AISI 4140 steel at 431°C (808°F).257

14.13.2 Solid-Metal Embrittlement Solid-metal embrittlement (SME), also called solid-metal-induced embrittlement (SMIE), occurs below Tm of the solid when the embrittling solid is an internal environment, such as inclusion. Although SME has not been suggested or accepted as an embrittlement phenomenon in industrial processes, numerous examples of loss in ductility, strength, and brittle fracture have been noted for electroplated metals and coatings or inclusions of low-melting-point metals below their Tm.261 Asayama has reported delayed failure of cadmium-plated high-strength steel bolt below the Tm of cadmium (Fig. 14.63); consequently, the use of cadmium-plated steel bolts above 230°C (450°F) is not recommended, despite their excellent corrosion resistance.262 Solid-lead embrittlement has been reported in notched tensile specimens of various steels, resulting in considerable loss in ductility below the Tm of lead. This behavior seems to be responsible for several elevated-temperature failures of leaded steel, namely, failure of steel shafts during straightening at elevated temperatures, radial cracking of gear teeth during induction heating, and heat treatment failures of jet-engine compressor disks.263 14.13.2.1 Characteristics of SME. To date, SME has been reported only in LME couples. However, SME can occur without LME. Table 14.10 lists the occurrence of SME, showing that all solid-metal embrittlers are also liquid-metal embrittlers. Since both SME and LME have similar behavior, the prerequisites for SME are the same as those for LME. Thus the prerequisites of SME are (1) intimate contact between the solid and the embrittler, (2) the presence of tensile residual or applied stresses, (3) the presence of the embrittler at the growing crack tip, and (4) crack initiation at the solid/embrittler interface from a barrier such as a grain boundary. Furthermore, the metallurgical variables that increase the brittleness in steels (e.g., grain size, strain rate, yield strength, solute strengthening, and the presence of

TEMPERING

14.107

TABLE 14.10 Occurrence of Solid-Metal Embrittlement in Steels257

Base metal 1041 1041 leaded 1095 3340 4130 4140

4145

4145 leaded 4340

4340M 8620 8620 leaded A-4 A-4 leaded D6ac † ‡ §

Embrittler (melting point) Pb (327°C, or 621°F) Pb In (156°C, or 313°F) Sn (232°C, or 450°F) Pb Cd (321°C, or 610°F) Cd Pb Pb-Bi (NA)§ Pb-Zn (NA) Zn (419°C, or 786°F) Sn Cd Pb In Pb-Sn-Bi (NA) In Sn Sn-Bi (NA) Sn-Sb (NA) In In In-Sn (118°C, or 244°F) Sn In Pb-4Sn (NA) Pb-Sn (NA) Pb-Sb (NA) Pb Pb Cd Cd Cd Zn Cd Pb Pb Pb Pb Cd

Temperature at onset of embrittlement °C

°F

288 550 204 399 100 212 204 399 316 601 300 572 300 572 204 399 Below solidus Below solidus 254 489 218 424 188 370 160 320 Room temperature Below solidus 80 176 204 399 Below solidus Below solidus 110 230 93 199 93 199 204 399 121 250 204 399 204 399 204 399 288 550 204 399 260 500 300 572 38 100 400 752 38 100 288 550 204 399 288 550 204 399 149 300

St, standard tensile test; DF, delayed-failure tensile test. S, smooth specimen; N, notched specimen. NA, data not available.

Test type†

Specimen type‡

ST ST ST ST ST DF DF ST ST ST DF DF DF DF DF ST DF ST ST ST DF DF DF ST ST ST ST ST ST ST DF DF DF DF DF ST ST ST ST DF

S S S N N N N S S S N N N N N S S S S S S S S S S S S S S S N N S N S S S S S N

14.108

CHAPTER FOURTEEN

stress raisers or notches) all seem to increase embrittlement. Susceptibility to SME is a function of the stress and temperature and does not take place below a certain threshold value. Delayed-failure type embrittlement has also been noticed for both SME and LME. The essential differences between SME and LME are 1. Formation of multiple cracks in SME, rather than the propagation of a single crack to failure in LME 2. Propagation-controlled fracture in SME, but their crack propagation rates being about 2 to 3 orders of magnitude slower than in LME 3. Possibility of a change from brittle intergranular fracture to ductile shear mode due to the inability of the embrittler to continue with the propagating crack tip 4. Presence of incubation periods, thereby implying that the crack nucleation process may not be the same as in LME 5. Nucleation and growth as two distinct stages of fracture in SME 6. Both SME and LME attributed to the reduction in the cohesive strength of the atomic bonds at the tip 7. Transport of the embrittler being the rate-controlling variable in SME 8. Crack-initiation attributed to the stress-assisted penetration of the embrittler in the grain boundaries, but crack growth controlled by the surface self-diffusion of the embrittling species, similar to those suggested for LME.257 The study of SME is of importance in eliminating the likelihood of LME that a crack, once formed, may propagate in a brittle fashion in the absence of embrittling species at the crack tip.257

14.14 MARAGING STEELS Maraging steels are a special class of highly alloyed low-carbon iron-nickel martensitic steels which derive their ultrahigh strength not from carbon but from the precipitation of a uniform and dense distribution of various fine intermetallic compounds and Laves phases during aging (or age-hardening) treatment.264–266 The term maraging is adopted because it involves martensite age-hardening, i.e., aging in the martensitic form.267 Furthermore, the presence of s-phase, m-phase, and R¢phase has been reported in the literature, although these appear to be rare. The type and structure of the precipitates are functions of the composition of the material, aging temperature, and time.268 In addition to the extremely high strength, the maraging steels have excellent fracture toughness and ductility. The commonly available maraging steels contain 17–19% Ni, 7.5–12.5% Co, 3–5% Mo, 0.2–1.8% Ti, and 0.1–0.15% Al. Like other martensitic alloys strengthened by intermetallics, maraging steels show the transformational behavior involving austenite formation on heating; martensitic reaction on air-cooling from the solution-annealing (austenitizing) to room temperature to form a soft, ductile, heavily dislocated low-carbon, iron-nickel, bcc lath martensite (Fig. 14.64a) (with no twinning); and decomposition of martensite on aging below the austenite start temperature As (Fig. 14.64b).266,269,270

14.109 FIGURE 14.64 (a) Metastable phase diagram for the iron-rich end of the iron-nickel binary system showing the fcc austenite-to-bcc lath martensite transformation upon cooling and the martensite-to-austenite reversion upon heating.266 (b) Equilibrium phase diagram showing, for higher nickel content, the austenite and ferrite equilibrium phases at low temperatures.266 (Reprinted by permission of ASM International, Materials Park, Ohio.)

CHAPTER FOURTEEN

14.110

TABLE 14.11 Nominal Composition of Commercial Maraging Steels273 Composition,† wt% Grade Standard grades 18 Ni (200) 18 Ni (250) 18 Ni (300) 18 Ni (350) 18 Ni (cast) 13 Ni (400)§ 12-5-3 (180)

Ni 18 18 18 18 17 13 12

Mo

Co

3.3 8.5 5.0 7.75 5.0 9.0 4.2‡ 12.5 4.6 10.0 10.0 15.0 3.0 —

Cobalt-free and low-cobalt-bearing grades Cobalt-free 18 Ni Marage 200 18.5 3.0 Cobalt-free 18 Ni Marage 250 18.5 3.0 Low-cobalt 18 Ni Marage 250 18.5 2.6 Cobalt-free 18 Ni Marage 300 18.5 4.0 Cobalt-free 18 Ni Marage 300 18.0 2.4 Cobalt-free Sandvik 1RK91 9.0 4.0 Cobalt-free C455 9.0 Zc).76 (Reprinted by permission of Pergamon Press, Plc.)

Critical grain size (i.e., microstructural) model for dynamic recrystallization. Numerous workers have shown that the characteristic flow curve changes gradually from the multipeak to the single-peak type with increasing Z, or strain rate, and decreasing temperature. These two types of behavior can be distinguished from the critical grain size mechanism map of Fig. 15.12.75 The vertical coordinate is Z, or its equivalent sp or ss. The horizontal coordinate is the initial grain size D0 or the equivalent stable recrystallized grain size Ds. The full line dividing the crosshatched region from the plain region separates the initial conditions into two types, either multiple-peak or single-peak flow. The full line also indicates the experimentally observed locus of D0 = 2Ds or the Z-2Ds relation. The same curve also denotes the dependence of Zc on D0, where Zc is the critical value of Z. Thus, when the point (D0, Z) lies above the line, Ds < D0 /2 and grain refinement takes place. Conversely, when (D0, Z) lies below the line, Ds > D0 /2 and grain coarsening occurs.77,80 When the critical grain size model was verified in a 0.115% V, a 0.035% Nb, and a 0.040%Nb-0.30%Mo microalloyed steel under both solute retardation and precipitation retardation conditions, these results clearly supported the belief that necklace recrystallization (where grain growth is halted by concurrent deformation) is linked with the single-peak behavior. Conversely, termination of grain growth, under grain-coarsening conditions, by boundary impingement is associated with the cyclic flow curve.75,80 Progressive Lattice Rotation Mechanism. It has been reported in certain materials that new grains with high-angle boundaries may be formed during straining, by the progressive subgrain-rotation with small associated boundary migration. This strain-induced phenomenon comprises gradual rotation of subgrains near the preexisting grain boundaries with the straining of the material. The occurrence of this phenomenon has been noted in Mg83 and in Al containing particular solute additions such as Al-Mg alloys84,85 and Al-Zn alloys.84

15.20

CHAPTER FIFTEEN

In this type of dynamic recrystallization, it is usually observed that the size of grains formed at the old boundaries is only slightly greater than that of the subgrains, and the grain size is, therefore, expressed approximately by Eq. (15.4). The mechanism by which this progressive subgrain rotation takes place is not yet fully understood. However, it is likely that it is associated with (1) inhomogeneous plasticity and accelerated dynamic recovery in the grain boundary regions and (2) grain boundary sliding. Although this phenomenon usually leads to partially recrystallized necklace microstructure, at large strains a fully recrystallized structure may be formed (Fig. 15.9d and e).59,74,84 3. Delay or hold period. Since the structural changes obtained by dynamic restoration are thermodynamically unstable, holding at temperature in the interstand period, particularly during roughing passes, modifies them by the static restoration process. These structural changes play an important role in determining the final microstructures and properties of HSLA steels.86 Static processes. Since the structures obtained by dynamic restoration are thermodynamically unstable, unloading the stress at strains less than ep and/or holding at temperature modifies them by the static restoration process. The static restoration process involves three different types of softening; namely, mode I is static recovery; mode II, classical (static) recrystallization; and mode III, metadynamic (i.e., post-dynamic) recrystallization, with the amount of each depending on the prestrain.87,88 Static recovery occurs at low strains and causes a loss of dislocation density, which, in turn, leads to a small decrease in yield strength or flow stress. This does not produce any detectable microstructural changes under optical microscope. In the case of classical (static) recrystallization, it is clear that recrystallized nuclei start to form after the completion of straining and an apparent incubation period. Classical static recrystallization progresses in the same manner as in cold-worked material. The rate of static recrystallization increases with the strain until the initiation of dynamic recrystallization.89 The distribution of recrystallized nuclei is highly localized and inhomogeneous, occurring predominantly at triple junctions of grains and deformed grain boundaries87,90 and less frequently at the intragranular sites such as twin boundaries and deformation bands. Hence, the initial grain size has an important bearing on the progress of recrystallization and the recrystallized grain size. Like annealing after cold-rolling, the rate of static recrystallization is a strong function of the extent of deformation and the temperature and is a weak function of strain rate. Figure 15.13 illustrates the effects of initial grain size and deformation temperature on the critical deformation required for recrystallization.22 The statically recrystallized grain size is determined primarily by the initial grain size and the extent of deformation, whereas the deformation temperature mainly affects the progress of recrystallization.19 The relationships among the statically recrystallized austenite grain size dsr (mm), prior (initial) grain size (mm), and the applied strain, for (a) both austenitic stainless steel and transformable C-Mn steel and (b) Nb steels, are given by

and

d sr = Kd02 3e -1

C-Mn steels

(15.11)

d sr = K ¢d02 3e -0.67

Nb steels, T > 950°C

(15.12)

where d0 is the initial austenite grain size (prior to deformation, mm), e is the equivalent true (rolling) strain, and K and K¢ are constants. For C-Mn steels, the values

THERMOMECHANICAL TREATMENT

15.21

Temperature

R

P

TCr

Undeformed Acceleration of P Retardation of R Deformed

FIGURE 15.13 Schematic diagram representing the interaction of recrystallization R and strain-induced precipitation P.22

Time

of K have been reported to be in the range of 0.35 to 0.83 (mm)1/3, and a mean value of 0.5 (mm)1/3 may be appropriately used.91 After the establishment of a statically recrystallized structure with an initial grain size dsr, at some interval within the interpass period, then, in the absence of any second-phase pinning particles, grain growth will occur. As stated earlier, in this case, the basic form of the grain growth equation can be given by

or

d n - d srn = kt

(15.13)

d = Kt m

(15.14)

where d is the growing grain size after a time t (s); K and k are constants; m is a temperature-dependent exponential; and n is the grain growth time exponential and in Eq. (15.13) is usually 2; i.e., classical parabolic growth follows. Note that in industrial situations, grain growth is not significant for the interpass time of 5 to 10 s in a plate mill and 0.1 to 10 s in a hot strip mill. However, after the completion of recrystallization at the end of the rolling pass, the grain growth rate within 20 to 40 s is found to be quite high, and, in that situation, Eqs. (15.13) and (15.14) require further modifications.92,93 The time for dynamic or static recrystallization to take place is dependent on the microalloy content due to (1) solute drag effect in retarding recrystallization and (2) the pinning action of stable microalloy carbide, nitride, and/or carbonitride precipitated on migrating grain boundaries. Metadynamic recrystallization, immediately after the termination of hot deformation, depends strongly on the strain rate; and it was reported that this condition occurred by transformation of partially recrystallized grain structure observed just after the deformation process to a more fully recrystallized structure by growth of recrystallized nuclei (grains) formed during deformation93a and, therefore, did not require an incubation period essential for conventional static recrystallization. It

15.22

CHAPTER FIFTEEN

was also found that the rate of softening was a very weak function of temperature, composition, and strain.89 Consequently, it is about an order of magnitude faster than classical static recrystallization. It also produces finer recrystallized grain structures compared with the classical mechanism, primarily due to the formation of higher density of nuclei by dynamic nucleation.77,88,89 4. Finish rolling. The deformation (rolling) of microalloyed austenite in the nonrecrystallized temperature range above or below Ar3 (intercritical rolling) leads to the breaks-up and formation of flattened (pancaked) unrecrystallized austenite grains. The process involving repeated flattening of the grains by repeated deformation below T5 or Tnr is referred to as classical or conventional controlled rolling (CCR). This is usually achieved by Nb additions. The conditions of finish-rolling differ extensively for plate, TRIP, and rod; and the resulting microstructures depend on the relative kinetics of recrystallization and precipitation. This practice is wellsuited for production conditions that require a high recrystallization temperature. Retardation of recrystallization. In controlled rolling of HSLA steel plate, straininduced precipitation of very fine Nb(CN) particles in austenite during recrystallization and at finishing temperatures plays a vital role of effectively stopping or retarding recrystallization of the austenite, which leads to increased ferrite nucleation on cooling due to grain elongation and accumulated strain (or flow stress).94 During hot-rolling, the spontaneous precipitation of MAEs in the austenite occurs with the falling temperature which results in the retardation of recovery and recrystallization. The effect of MAEs on recrystallization can also be observed with respect to the increased recrystallization temperature that takes place after a particular rolling deformation. Figure 15.2 shows the effects of MAEs on the critical temperature for g recrystallization. Under the basic conditions of hot deformation, recrystallization occurs in the region above the curves in each case, whereas below the curves it is strongly retarded. It is thus clear that Nb is very potent in raising the recrystallization temperature. The TMT is characterized by a complex interaction between the retardation of recovery and recrystallization occurring after rolling and a marked acceleration of strain-induced precipitation, if adequate MAEs and C and/or N are present, as shown schematically in Fig. 15.13, occurring below a critical temperature.22 Effect of deformation on ferrite formation. Among the austenitic conditions, the austenite grain structure is of vital importance since it determines the density of sites for ferrite nucleation during subsequent transformation. The nucleation sites are the internal defect structures, namely, grain boundaries, subboundaries, deformation bands, and incoherent annealing twin boundaries. Researchers have found that the austenite grain boundary per unit volume Sv, which is inversely proportional to the austenite grain diameter D, is an effective grain size parameter. Austenite grains with an increased Sv, obtained by producing large rolling reduction and highly elongated grains, favor large ferrite nucleation, while a small Sv causes low nucleation rates and coarse ferrite grains (Fig. 15.14a).18 Although RCR and CCR offer different approaches to austenite conditioning, they both have the same objective of achieving grain refinement in the final plate, beam, coil, or forging. The difference between RCR and CCR can be realized by using Fig. 15.14b which shows the different path employed by each to achieve high Sv values.96 The increase in Sv for the RCR practice is due solely to an increase in grain boundary area per unit volume owing to a decrease in average grain volume, whereas the increase in Sv for the CCR practice is due to the increase in grain

THERMOMECHANICAL TREATMENT

15.23

(a)

(b)

FIGURE 15.14 (a) A relationship between ferrite grain size and Sv.18 (Reprinted by permission of ASM International, Materials Park, Ohio.) (b) Schematic illustration of austenite microstructures resulting from deformation above or below the recrystallization stop temperature of austenite, with corresponding description of Sv. The superscripts GB, DB, TB, and NPD are the contributions to the total Sv from grain boundaries, deformation bands, twin boundaries, and near-planar defects.96 (After E. J. Palmiere, University of Pittsburgh, unpublished research, 1990.)

boundary area per unit volume resulting from a change in grain shape and through the addition of the transgranular twins and deformation bands.95 Precipitation kinetics of microalloying during hot-rolling. In microalloyed steels, the strain-induced precipitation (of carbonitrides) during hot deformation decreases the onset of dynamic recrystallization and the kinetics of static recrystallization when compared to C-Mn steels. Hence, the classical Avrami equation, usually

CHAPTER FIFTEEN

15.24

employed for C-Mn steel, has to be modified to incorporate this retardation effect. Empirically, the strain-induced carbonitride precipitation in microalloyed Nbbearing HSLA steel can be described by the Avrami equation of the form fp = 1 - exp ( - kt n )

(15.15)

where fp is the fraction of strain-induced precipitation at the holding time t (s), K is a constant, dependent on nucleation and growth rates, and n is a constant, near 0.6. Finally, the precipitation kinetics in austenite can be described by the equation fp = 1 - exp ( -0.38e 0.265t 0.6 )

(15.16)

in unrecrystallized austenite at 950°C, or fp = 1 - exp( - Kt 0.6 )

(15.17)

in recrystallized austenite where K = 0.005 to 0.006 or 0.004, respectively, at 900 to 950°C or 1050°C.97 5. Cooling. It has long been understood that the transformation temperature and cooling rate have profound effects on the transformation structures. A simple relationship between ferrite grain size (da, mm), austenite grain size (dg, mm), and some of the process variables has been given by:98

.

.

da = a + bT -1 2 + c[1 + exp( -1.5 ¥ 10 -2 dg )](1 - 0.45e 1 2 )

(15.18)

where T is the cooling rate (°C/s); e is the strain applied during finish-rolling below the recrystallization temperature of g ; and a, b, and c are constants, dependent on the composition of steel and, perhaps, on the processing parameters determining the degree of strained-induced precipitation. Accelerated cooling lowers the transformation temperature and increases the density of effective nucleation sites within the unrecrystallized austenite grains in addition to those activated by deformation bands, which provides both the ferrite grain refinement with uniform grain size and finer carbonitride precipitation during or after transformation to ferrite, thereby further improving room-temperature strength at the expense of toughness.43 In addition, AC modifies the transformed structure by replacing pearlite with an increased volume fraction of finely dispersed bainite as well as produces smaller and more effective microalloy precipitates.19 All these factors enhance the strength of the product.35 15.2.2.4 Recrystallization-Controlled Rolling. To overcome the above shortcomings of the CCR method, attention was directed toward the RCR method of producing very fine ferrite grain size on the order of 5 to 10 mm (0.2 to 0.4 mil). This process involves a relatively low reheating temperature, repeated high-temperature finish-rolling above TR (i.e., up to 1050°C), and retardation of austenite grain growth during and after rolling. In the RCR operation, recrystallization must go to completion within times available between rolling passes (i.e., occurrence of full recrystallization after each pass or after alternate passes to control grain refinement). After the last rolling pass, the fine recrystallized austenite grains transform to very fine ferrite grains. This approach is well-suited for production conditions that must have high finishing temperatures, e.g., in underpowered rolling mills and forging. The RCR process in combination with controlled accelerated cooling is often used to optimize the ferrite nucleation rate and higher precipitation-strengthening increment.

THERMOMECHANICAL TREATMENT

15.25

Advantages of the RCR processes are their high productivity, rapid processing, lower mill loading, energy savings, and easy implementation within existing plants,99 compared to CCR. It also enhances the through-thickness properties of TMCP steel.100 To successfully utilize the RCR approach, it is necessary to determine the relationship between recrystallization behavior (time for static recrystallization and recrystallized grain size) and rolling parameters (strain, temperature, and interpass time).101 The RCR process is especially suited to high-speed, high-finishtemperature bar and section mills, as well as plate and strip mills which are not designed for high mill loads produced at low-temperature CCR. The RCR is well-suited to V-N and V-N-Ti steels. V steels with increased N contents (0.01 to 0.02% N) exhibit austenite grain refinement during high-temperature rolling, as well as increased VN precipitation strengthening when compared to lownitrogen vanadium steels. The negative influence of Nb at higher finish-rolling temperature and the need for a low TR prevent its use in RCR steels. A small Ti addition (0.015%) to V-N steels causes a fine distribution of stable TiN particles, which effectively impede austenite grain growth during reheating and rolling, enhances the uniformity of grain size upon transformation, and contributes to a higher final Sv by hindering growth of recrystallized grains (Fig. 15.15). Figure 15.15 shows the microstructural evolution of austenite grain size in C-Mn and V-TiN steels during RCR.102 This illustrates the occurrence of a significant grain growth in C-Mn steel after the last rolling pass with respect to V-N-Ti steel.102 Properties of low-carbon V-Ti-N steels, processed by RCR in combination with accelerated cooling, are considered as an available alternative to steels processed by lowtemperature CR. The concept of the RCR - AC approach has been successfully applied to produce heavy plates of Ti-V-N microalloyed HSLA steels which is found to be more economical than low-finish-temperature CR. Mechanical properties of the present 0.01Ti-0.08V-N steels in the RCR + AC condition (finish-rolling tem-

FIGURE 15.15 Microstructural evaluation of C-Mn and Ti-V-N austenites.102

15.26

CHAPTER FIFTEEN

perature ~1050°C) were reported to be superior to those after CR with finish-rolling temperature of ~800°C.103 15.2.2.5 Some Applications of Microalloying and TMT. A very wide variety of microalloying applications are clear from the diverse range of carbon in microalloyed steels from 0.02% C in interstitial-free steels to eutectoid composition for wire rod employed for prestressed concrete stands. In these applications, MAEs are used as getters for interstitials, as precipitation strengthener, or as microstructural modifier, depending on the property requirements. Different functions fulfilled by microalloying are described below as some examples.104 1. High-strength linepipe steels. The high-strength linepipe steels (API X60 to API X80) are produced via a judicious choice of (micro)alloy composition and optimization of TMT and subsequent accelerated cooling conditions to obtain a fully bainitic structure, a high level of toughness with a low DBTT.22,105 The economic benefits of these higher-strength pipelines, such as lower pipe procurement cost, lower cost for transport to the site, reduced gas transportation, and reduced welding costs due to smaller-diameter and thinner walls, have led to their increasing use throughout the world for laying pipelines over long distances across the national and international borders and counteracting the consequent rise in operating pressure and exploitation of deposits with extremely sour media. From the weldability point of view, improvement in weldability provides a challenging problem arising from combinations of weld joint, welding procedure and technology, and alloy composition and properties. However, from a metallurgical point of view, weldability is associated with hardenability, HAZ properties, HAZ and weld metal cracking, and ability of steel to respond to post-weld treatments.106 It is possible to achieve an improvement in weldability by using several metallurgical processes, such as inclusion shape control (CaSi-rare earth injection treatment), reduced P and S levels, improvement of steelmaking and continuous casting practices,105 effective exploitation of TiN technology (Ti content between 0.010 and 0.015% and Ti/N ratio £3.42),104 and a reduction of C, Mn, and Si concentrations to minimize segregation of these elements to the centerline during the continuous casting and welding.106 It has been shown by William et al.105 that lower Mn content reduces centerline microstructural banding, leading to low-segregation-ratio hardenability enhancing Mn, which lowers considerably the likelihood of martensite formation in the hot-rolled strip and plate. Taillard et al.107 reported that Si content 1000 MPa), nonmagnetic properties with a slight reduction in ductility and toughness. In this case the strengthening mechanism is due to a combination of the grain size hardening, substructure hardening, workhardening, and solid-solution hardening, all of which are increased by higher N level.117 8. AISI 4130 steel. An optimum combination of strength and toughness (UTS of 660 MPa, KIC of 70 MPa◊m1/2) and microstructure of fine equiaxed (with mean 8-mm diameter) ferrite grains in the finished product has been reported using a TMT involving a reheat temperature of 900°C, and 50% deformation above Ar3, and reducing the finish-rolling temperature to 620°C (i.e., in the g-a region).118

15.2.3 High-Temperature Thermomechanical Treatment (HTMT) This is a highly effective nonconventional heat treatment operation which was developed by Kula and his associate, later followed up in the United States and Russia.1 The usual method, schematically shown in Fig. 15.1a, consists of deforming the stable austenite at temperatures just above Ac3, then direct quenching (to form martensite), followed by tempering. Alternative treatments include deforming the steel on a falling temperature between Ac3 and Ar3 or Ac1 (i.e., between 1000 and 800°C), but in this case the accelerated decomposition into ferrite/carbide structure must be avoided.119,120 As in other TMT processes, the increased strength achieved in this case is due mainly to g grain size refinement (say, from 40 to 60 mm to ~3 mm) and suppression of recrystallization by alloy carbide precipitation. Advantages over Conventional Heat Treatment.3,48 First, HTMT is used to provide superior strength, higher ductility, and improvement in fracture toughness by markedly refining both the initial austenite grains and the subsequently formed martensitic platelets. It improves the low-temperature mechanical properties of ultrahigh-strength steels (such as 4340) when appropriate combinations of deformation temperature with tempering conditions are applied to the steels.120a Second, the martensitic structure produced is less susceptible to quench-induced cracking and is less liable to premature or delayed fracture. It greatly hinders the initiation of defects in the martensitic structure by suppressing the dynamic influence of martensitic phase transformation. Third, tools and dies made of HTMT materials have considerably increased useful life. Fourth, it is simple to apply in industry.121 Advantages over Ausforming Process. First, HTMT produces higher impact ductility, a substantial increase in fatigue limit but slight reduction in strength. Second, it is most practicable because there is greater flexibility in temperature control and reduction in the forming loads. Also, the optimum properties can be attained at a moderate deformation (~40%). Application. It is applied to highly alloyed steels, especially suited to those steels which recrystallize slowly, such as silicon-containing steels and microalloyed 38CrSi steel (containing 0.37% C, 0.62% Mn, 1.33% Si, 1.58% Cr, 0.02% S, 0.02% P) with

THERMOMECHANICAL TREATMENT

15.29

0.18% Ti, 0.11% Nb, or 0.15% V.122 It can be applied to low-alloy high-carbon steels which are unsuitable for the lower-temperature treatment such as ausforming with significant success in improved strength and toughness. The fatigue limit is also found to improve in a wide variety of steels provided the deformation is limited to 25 to 30%.

15.2.4 Low-Temperature Thermomechanical Treatment (LTMT) or Ausforming Ausforming, first suggested in 1954, now consists of quenching the steel from the austenite phase field to a temperature in the metastable bay, usually in the range of 450 to 600°C (i.e., below the pearlite nose), where it is deformed to a considerable amount (up to 80% reduction in cross-sectional area) before its transformation, and then quenching to develop a martensitic structure followed by tempering (Fig. 15.1c). For ausforming, the steel must have a TTT diagram with a deep metastable austenite region and a large bay between the pearlite and bainite noses,123 which can be produced by carbon content of 0.3 to 0.4% and a high concentration of alloying elements (e.g., Cr, Mo, Ni, and Mn) in the ausformed steel. Other alloying elements such as Ti, Nb, and V may be advantageously added to enhance the coarsening resistance during tempering, thereby increasing the strength of the steel.124 A schematic time-temperature-deformation diagram of this process is shown in Fig. 15.1c. The retained austenite is completely removed after double tempering of ausformed steel specimens.125 The important process variables which affect the overall properties of ausformed steels are the austenitizing temperature, amount (total percentage) and temperature of deformation, and deformation schedule. Improvement of mechanical properties due to ausforming is proportional to the amount of deformation. An increase in strength is observed with the decreasing deformation temperature, due, probably, to the greater strain-hardening; however, the temperature selected should be appropriate so that neither recovery and recrystallization nor transformation occurs during the deformation.124 Characteristic Features of Ausforming 1. Strength obtained by this process is usually independent of the prior g grain size and carbon content.126 2. Improved strength (without adverse effect on ductility) and toughness can be obtained. Alternatively, for a given strength, ausformed steels have greater toughness than conventionally heat-treated steels. Large improvements in strength, during ausforming treatment, are brought about by the combined effect of precipitation-hardening due to fine dispersion of alloy carbide precipitates, very high dislocation density (up to 1013/cm2), and martensite transformation associated with a large number of smaller martensite plates and inherent fine dislocation substructures from the deformed metastable austenite.124 Dislocations are partly formed as uniformly distributed pile-up dislocations during deformation and are partly inherited as dislocation cell structure during transformation into martensite. The contribution of inherited dislocation structure is more effective at high deformation temperatures. The explanation is that as the ausforming temperature is increased, the dislocation configuration changes from uniformly distributed pile-up dislocations into dislocation cells.127

15.30

CHAPTER FIFTEEN

3. Steels required for ausforming exhibit serrated yielding during the warm deformation process as a result of dynamic strain aging (see Chap. 4) involving the formation of a fine dispersion of alloy carbides on the dislocation networks.124 4. Higher ductility is presumably related to the formation of a fine untwinned (i.e., twin-free) martensite plate.8 The martensite plate size is considerably smaller than that in similar steels, given a conventional quenching and tempering treatment. This occurs due to the pinning effect of fine alloy carbide precipitates on the prior dislocation configuration (arrays or tangles) which serve as barriers to martensite plate propagation. Subsequent tempering further improves the impact properties because of the precipitation of carbides in a fine spheroidal form.128 5. It is beneficial for those parts which can be shaped to produce simple shapes by the deformation process, e.g., punches,129 leaf springs, and bolts.8 6. The ausformed parts have excellent fatigue strength under severe stress conditions. They also offer better performance at high temperature because of their high strength at elevated temperature and greater ability to withstand heat fatigue.123 Disadvantages of Ausforming 1. Ausforming requires the use of expensive, high-alloying additions (e.g., Cr, Ni, Mo, etc.). 2. It has not been successful in practical use for steels with high concentration of carbon and alloying elements because the deformation resistance of such steels is usually high in ausforming compared to warm-forging, due to the precipitation of a substantial amount of alloy carbides during deformation. 3. High loads on the rolling mill or forming equipment are necessary to produce large deformation; this also limits the treatment to simple shaped parts.8 As a result of these shortcomings, ausforming is not a widely accepted industrial practice. Application. Ausforming is applied to steels with relatively high hardenability. It is especially suitable for high-alloy steels such as high-speed tool steels, AISI H13 tool steels (modified with Nb),130 and 17-4 PH stainless steel131; and for low-alloy steels, especially containing Cr and Mo. The process can be applied to certain martensitic steels (with advantages)—examples are cold heading, cold piercing, extrusion, riveting, and hot piercing of steels.132 Note that the ausforming of 17-4 PH stainless steel does not increase the strength of the product, but increases the ductility. Moreover, ausformed and aged 17-4 PH stainless steels offer a dramatic improvement in the impact value over the temperature range of -100 to 0°C, due to the refined microstructure.131

15.2.5 Isoforming Isoforming is a class II TMT which involves austenitizing the steel, quenching to an intermediate temperature in the pearlite nose region, deforming the metastable austenite during the isothermal transformation to pearlite, followed by air-cooling (Fig. 15.1d). Tempering is not needed at all. It is applied to any low-alloy steel with a suitable TTT diagram.

THERMOMECHANICAL TREATMENT

15.31

Irani et al.133–135 have shown that the isoforming process can significantly improve the toughness of several low-alloy steels relative to the conventionally rolled products. It has been demonstrated that the lamellar morphology of pearlite lowers the toughness of ferrite-pearlite steel, the DBTT increasing with increasing pearlite content. However, isoforming, involving high deformation (>60%) during the transformation of a suitable low-alloy steel between 600 and 700°C, produces the change into a satisfactory morphology and distribution of ferrite and carbide from the prior ferrite-pearlite structure. The final structure of the isoformed steels is characterized by the formation of very fine ferrite subgrains (0.26% copper, inclusion shape control, and minimum Mn segregation. 15.3.2

Mechanical Properties

Hot-rolled steels with ferrite-pearlite microstructures are the most widely used HSLA steels. Commercially available microalloyed ferrite-pearlite HSLA steels have yield strengths in the range up to 700 MPa (100 ksi), which is about 3.5 times greater than 200 MPa (30-ksi) yield strength of conventional hot-rolled plain carbon steels. Tables 15.1 and 15.2 summarize tensile properties of some hot-rolled ferritepearlite HSLA steels.142 These properties depend on the alloying additions and production methods.

15.3.3

Structure-Property Relationships

Strengthening Mechanisms. The major strengthening mechanisms in controlled rolled HSLA steels include grain refinement; precipitation-hardening by strainenhanced precipitation of microalloyed carbonitrides in ferrite or at the interphase; solid-solution strengthening from Mn, Si, and uncombined N; and dislocation substructure (including dislocation tangles and cell walls) strengthening. When controlled rolling is used with a finishing temperature well below the Ar3 temperature, an additional contributor to the overall strength comes from preferred orientation or texture hardening. The observed yield stress of a polycrystalline controlled rolled microalloyed HSLA steel can be expressed by the following Hall-Petch relationship: s y = s 0 + ky d -1 2 = (s i + s ss + s d + s p + s tex ) + ky d

(15.19) -1 2

+ k ¢d

-1 2 s

(15.20)

where s0 is the friction stress and is a function of steel composition; ky is a material constant; d is mean ferrite grain size (diameter); ds is the subgrain diameter; and s0 is divided into several terms: internal friction stress si, solid-solution strengthening sss [as given by Eq. (7.42)], dislocation strengthening sd [as given by Eq. (14.6)], precipitation strengthening sp [as given by Eq. (14.8)], and texture strengthening stex; kyd-1/2 is the contribution to the strength by the ferrite grain size; and k¢ds-1/2 is the subgrain or substructure strengthening term. However, the relative contribution of any individual mechanism presumably varies with the change in steel composition and rolling practice.143 A typical effect of g Æ a transformation temperature for an Fe-0.1C-1Mn-0.2Si steel with a prior g grain size of ASTM 8 (or 20 mm) is shown in Fig. 15.17a. Evidently a decrease in the g grain size and an increase in cooling rate associated with decrease in the transformation temperature will lead to a lower ferrite grain size, as shown by26 da = adg + bT˙ -1 2 + c

.

(15.21)

where da and dg are the ferrite and austenite grain sizes, respectively; T is the cooling rate; and a, b, and c are constants.

TABLE 15.1 Tensile Properties of HSLA Steel Grades Specified in ASTM Standards142

ASTM specification†

Type, grade, or condition

A 242

Type 1

A 572

Grade 42 Grade 50 Grade 60 Grade 65

A 588

Grades A–K

Minimum tensile strength§

Product thickness‡ mm

in. 3

Minimum yield strength§

Minimum elongation,§ % in 200 mm

in 50 mm

MPa

ksi

MPa

ksi

(8 in.)

(2 in.)

Bend radius§ Longitudinal

Transverse

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

15.35

20 20–40 40–100

/4 /4–11/2 11/2–4

480 460 435

70 67 63

345 315 290

50 46 42

18 18 18

... 21 21

150 100 32 32

6 4 11/4 11/4

415 450 520 550

60 65 75 80

290 345 415 450

42 50 60 65

20 18 16 15

24 21 18 17

¶

100 100–125 125–200

4 4–5 5–8

485 460 435

70 67 63

345 315 290

50 46 42

18 ... ...

21 21 21

¶

50 45

... ...

22 22

t t

2t–3t 2t–3t

3

¶ ¶ ¶

¶ ¶

... ... ...

A 606

Hot-rolled Hot-rolled and annealed or normalized

Sheet Sheet

480 450

70 65

345 310

Cold-rolled

Sheet

450

65

310

45

...

22

t

2t–3t

A 607

Grade 45 Grade 50 Grade 55 Grade 60 Grade 65 Grade 70

Sheet Sheet Sheet Sheet Sheet Sheet

410 450 480 520 550 590

60 65 70 75 80 85

310 345 380 415 450 485

45 50 55 60 65 70

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

22–25 20–22 18–20 16–18 15–16 14

t t 1.5t 2t 2.5t 3t

1.5t 1.5t 2t 3t 3.5t 4t

A 618

Ia, Ib, II Ia, Ib, II, III

19 19–38

485 460

70 67

345 315

50 46

19 18

22 22

t–2t t–2t

... ...

A 633

A C, D C, D

100 65 65–100

430–570 485–620 450–590

63–83 70–90 65–85

290 345 315

42 50 46

18 18 18

23 23 23

¶

... ... ...

3 /4 /4–11/2

3

4 2.5 2.5–4

¶ ¶

TABLE 15.1 Tensile Properties of HSLA Steel Grades Specified in ASTM Standards142 (Continued)

ASTM specification†

A 656

Type, grade, or condition

Minimum tensile strength§

Product thickness‡ mm

in.

MPa

ksi

Minimum yield strength§ MPa

ksi

Minimum elongation,§ % Bend radius§

in 200 mm

in 50 mm

(8 in.)

(2 in.)

Longitudinal

Transverse ... ...

E E

100 100–150

4 4–6

550–690 515–655

80–100 75–95

415 380

60 55

18 18

23 23

¶

50 60 70 80

50 40 25 20

2 11/2 1 3 /4

415 485 550 620

60 70 80 90

345 415 485 550

50 60 70 80

20 17 14 12

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

¶

¶

¶ ¶ ¶

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

15.36

A 690

...

100

4

485

70

345

50

18

...

2t

...

A 709

50 50 W

100 100

4 4

450 485

65 70

345 345

50 50

18 18

21 21

... ...

... ...

A 715

Grade 50 Grade 60 Grade 70 Grade 80

415 485 550 620

60 70 80 90

345 415 485 550

50 60 70 80

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

22–24 20–22 18–20 16–18

0 0 t t

t t 1.5t 1.5t

A 808

...

450 450 415

65 65 60

345 315 290

50 46 42

18 18 18

22 22 22

... ... ...

... ... ...

A 812

65 80

585 690

85 100

450 550

65 80

... ...

13–15 11–13

... ...

... ...

A 841

...

485–620 450–585

70–90 65–85

345 310

50 45

18 18

22 22

... ...

... ...

A 871

60, as hot-rolled 65, as hot-rolled

520 550

75 80

415 450

60 65

16 15

18 17

... ...

... ...

†

Sheet Sheet Sheet Sheet 40 40–50 50–65

11/2 1 /2–2 2–21/2 1

Sheet Sheet 65 65–100 5–35 5–20

2.5 2.5–4 3

/16–13/8 /16–3/4

3

For compositions, available mill forms, and special characteristics, see Table 1.13. Maximum product thickness except when a range is given. No thicknesses are specified for sheet products. May vary with product size and mill form. ¶ Optional supplementary requirement given in ASTM A 6. Reprinted by permission of ASM International, Materials Park, Ohio. ‡ §

TABLE 15.2 Mechanical Properties of HSLA Steel Grades Described in SAE J410142 Minimum tensile strength‡ Grade†

15.37

942X 945A 945C 945X 950A 950B 950C 950D 950X 955X 960X 965X 970X 980X

Minimum yield strength‡§

Minimum elongation,‡ %

MPa

ksi

MPa

ksi

in 200 mm (8 in.)

in 50 mm (2 in.)

Bend diameter‡¶

415 415–450 415–450 415 430–483 430–483 430–483 430–483 450 483 520 550 590 655

60 60–65 60–65 60 63–70 63–70 63–70 63–70 65 70 75 80 85 95

290 275–310 275–310 310 290–345 290–345 290–345 290–345 345 380 415 450 485 550

42 40—45 40–45 45 42–50 42–50 42–50 42–50 50 55 60 65 70 80

20 18–19 18–19 19 18–19 18–19 18–19 18–19 18 17 16 15 14 10

24 22–24 22–24 22–25 22–24 22–24 22–24 22–24 22 20 18 16 14 12

t–3t t–3t t–3t t–2.5t t–3t t–3t t–3t t–3t t–3t t–3.5t 1.5t–3t 2t–3t 3t 3t

† For compositions, available mill forms, and special characteristics of these steels, see Table 1.13. ‡ May vary with product size and mill form; for specific limits, refer to SAE J410. § 0.2% offset. ¶ 180° bend test at room temperature. Used for mill acceptance purposes only; not to be used as a basis for specifying fabricating procedures. Reprinted by permission of ASM International, Materials Park, Ohio.

CHAPTER FIFTEEN

15.38

(a) (b)

FIGURE 15.17 The effect of (a) g Æ a transformation temperature on the transformed polygonal ferrite grain size and (b) carbide thickness on the ITT of a ferrite-pearlite HSLA steel.26 (Courtesy of F. B. Pickering.)

Toughness. Both aspects of toughness, such as the ductile-brittle transition temperature (DBTT) or impact transition temperature (ITT) associated with the resistance to brittle cleavage fracture and the Charpy shelf energy (CSE) associated with the resistance to low-energy ductile fractures, must be considered for overall evaluation. The ITT of low-carbon pearlitic microstructures given by Eq. (7.44) can hold for ferrite-pearlite HSLA steels. In addition, both precipitation strengthening sp and dislocation strengthening sd increase the ITT by about 0.25°C per MPa increase in sp and 0.4°C per MPa increase in sd. Similarly, an increase in {111}a texture in the rolling plane also raises the ITT; however, this is very small in hot-rolled or normalized plate or sheet.26 The ITT decreases with the increase of d-1/2, as expressed by ITT = To - ky d -1 2 = f (s 0 ) - ky d -1 2

(15.22)

where To is a function of the lattice friction stress s0 and ky is a measure of the effectiveness of refining the ferrite grain size in order to lower ITT. For a given design stress, the beneficial effect of controlled rolling for improved toughness is attributed to the combined effects of lattice friction stress and grain size refinement; the latter factor affecting the most. For example, for each unit increment in d-1/2 mm-1/2, ITT drops by about 11.5°C. 144 For 0.01Ti-0.08V-N microalloyed HSLA steels in the RCR + AC condition, the ITT may be expressed in the form of an inverse relationship between toughness and yield strength as ITT(∞C) = -217 + 0.33(YS)

(15.23)

This singular relationship is based virtually on constant g grain size and the fact that increased N and V contents and increased cooling rate and decreased finish-cooling temperature all seem to simultaneously increase precipitation hardening while refining the ferrite grain size.103

THERMOMECHANICAL TREATMENT

15.39

FIGURE 15.18 Effect of sulfur content on the Charpy shelf energy of a ferrite-pearlite HSLA steel, showing the anisotropy of the Charpy shelf energy. The sulfur content is proportional to the volume fraction of MnS.26 (Courtesy of F. B. Pickering.)

Finally, if any coarse carbides form in the hot-rolled or normalized HSLA steels, which is unusual, ITT will increase with carbide thickness >2 mm, as shown in Fig. 15.17b. This effect is of great interest in evaluating the toughness of the acicular ferrite microstructure. A high CSE is essential to avoid the likelihood of low-energy ductile fracture as well as to overcome lamellar tearing in welds, particularly in the short transverse or through-thickness direction. There is a significant decrease of CSE with increasing pearlite content (as shown in Fig. 7.27) and increasing nonmetallic sulfide inclusion content in ferrite-pearlite HSLA steels (Fig. 15.18). The remarkable anisotropy in CSE, as evidenced in the short transverse or through-thickness direction, has been eliminated/minimized by inclusion shape control. However, note that inclusion shape control is not advantageous for ITT, which really increases. An equation to describe the transverse CSE is given by145 CSE( J) = 112 - 2.8d -1 2 - 0.18s p (MPa) - 832(%S) - 43(%P) -0.76(%pearlite) + 107(%Zr )

(15.24)

It is clear that increases in pearlite (i.e., C%) and S contents have deleterious effects. Zr offers beneficial effect through inclusion shape control. The deleterious effect of sp and grain refinement denote the lower CSE due to overall strength increase. Al is deleterious because aluminous inclusions formed by Al deoxidation decrease the CSE value.

15.4 TRIP OR MULTIPHASE STEELS Martensite formed by plastic deformation of metastable (or retained) austenite gR is called strain-induced martensite (SIM). This irreversible transformation-induced plasticity (TRIP)146 or strain-induced martensite transformation (SIMT) causes superior mechanical properties such as increased tensile strength (sT), resistance to

15.40

CHAPTER FIFTEEN

local inhomogeneous necking (or strain inhomogenization), work-hardening rate, uniform elongation eu up to extremely high strain rates, and formability (or sT ◊ eT%). The strain- (or stress-) induced transformation to martensite has become the basis to create a new class of ultrahigh-strength metastable austenitic steels, known as TRIP steels. In these steels, deformation of metastable austenite between the Md and Ms temperatures little by little causes formation of martensite up to fracture, elongation well over 100%, and yield strength often greater than 1379 MPa (200 ksi).147,148 The presence of gR in the microstructure also causes higher toughness via localized TRIP leading to crack-tip blunting. Fatigue crack propagation (FCP) studies (controlled DK) have further suggested that the deformation-induced martensitic transformation retards the propagation of a crack in the lower-strength metastable austenite, especially at low DK, and increases the fatigue strength of TRIP steels in smooth bar tests under stress-control situations.149,150 This phenomenon in delaying and stabilizing the necking was first reported by Banerjee.151 Later, Zackay et al.152 found it in highly alloyed austenitic Fe-Cr-Ni stainless steels. However, the large amount of expensive alloying elements imposed severe practical problems for their widespread use, notably, for automotive applications.153 This deformation-induced a ¢ transformation method of improving toughness and ductility is also used in various other materials such as in cast iron,153a (a + b) type Ti-6Al-2Sn-4Zr-6Mo alloy,153b and stabilized zirconia.153c TRIP steels are based on multiphase microstructures comprising initially three phases, ferrite, bainite, and gR, with a fourth one being martensite transformation from gR during deformation such as deep drawing of sheet steel. A typical phase distribution of microalloyed Si-Mn TRIP steel [composition: 0.17C-1.4Mn-1.5Si-(0.020.04)Nb] in the as-shipped condition comprises about 50 vol% ferrite, 40 to 45 vol% bainite, and 5 to 10 vol% gR, but gR transforms to martensite during deformation. In this, Nb retards the isothermal bainite transformation kinetics, which causes an enhanced amount of gR during proper annealing. Nb also enhances the yield and tensile strengths by ~15 MPa per 0.01% addition due to grain refinement.154 The TRIP steels can be classified into (1) high-alloy austenitic TRIP steels, comprising Fe-Cr-Ni-Mo150,155 and Fe-Mn-Si-Al alloys;156 (2) low-alloy Si-Mn TRIP steels with varying amounts of Si and Mn; (3) TRIP-assisted steels (with increased gR stability);153 and (4) microalloyed Si-Mn TRIP steel processed by a TMP route.157 Note that TRIP-assisted multiphase steels are of two types: (a) cold-rolled TRIP-assisted multiphase steels by combination of intercritical annealing of cold-rolled product and isothermal holding at bainite transformation temperatures during a continuous annealing process and (b) hot-rolled TRIP-assisted multiphase steels by austenitic hot-rolling, holding in the ferrite field, followed by coiling in the bainitic range. Unfortunately both these TMT methods need a high Si content to suppress cementite precipitation in order to avoid a loss of gR stability. Also, high Si content causes red scale surface defects, leading to moderate hot-dip galvanizability.158 Table 15.3 lists the typical composition of TRIP steels.150,153–159,161,162,164–168

15.4.1 Advantages of TRIP Steels 1. Compared to conventional high-strength steels, TRIP steels offer higher formability (i.e., strength-ductility combinations) and expand the range of coldformable steels at high-strength values.154 The product of sT ◊ eT percent was reported by Kawano et al.159 in hot-rolled Fe-0.2C-1.5Si-(1.5-1.7)Mn steel in the range of 25,000 to 30,000 MPa◊% and by Sakuma et al.153 in cold-rolled 0.2C-2.1Si-0.94Mn and 0.2C-1.24Si-1.8Mn steels (held at 425°C in the bainite range) in excess of 27,500

TABLE 15.3 Typical Compositions, wt%, of TRIP Steels150,153–159,161,162,164,166–168

15.41

No.

C

Si

Mn

Ni

Mo

Cr

Other

gR%

Steel type

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

0.25 0.30 0.25 0.25 0.2–0.4 0.16 0.2 0.2 0.39 0.18 0.2 0.2 0.20 0.4 0.17 0.16 0.22 0.2

2.0 2.0 2.0 2.0 3.0 — 2.0 1.5 1.37 0.39 2.0 2.1 1.24 1.5 1.5 0.45 1.55 1.5

2.5 2.0 1.7 2.0 15–30 — 1.5 1.5–1.7 1.45 1.33 1.0 0.94 1.8 1.5 1.4 1.4 1.55 1.5

9.0 8.0 8.8 8.8 — 1.5 — — — — 1.0 — — — — — — —

5.0 4.0 5.5 3.0 — — — — — — — — — — — — — —

9.0 9.0 9.0 9.0 — 1.6 — — — — — — — — — — — —

— — — — 3.0 Al — — — — 0.018P, 0.012S, 0.029Al, 0.0069N 1.2Cu 0.0058S, 0.003P, 0.041Al, 0.0027N 0.006S, 0.003P, 0.041Al, 0.0028N 0.018S, 0.015P, 0.036Al 0.002–0.04Nb 0.01P, 0.035Al, 0.045N, 0.005Ti, 0.015Nb, 0.03V 0.035Nb, 0.028Al, 0.002–0.004N 0.035Nb

— — — — — 10 — 15

Austenitic TRIP steel150,155 Austenitic TRIP steel150,155 Austenitic TRIP steel150,155 Austenitic TRIP steel150,155 Austenitic TRIP steel156 15CrNi6 TRIP steel164 Si-Mn TRIP steel167 Si-Mn TRIP steel159 Si-Mn TRIP steel166 Si-Mn TRIP steel161 Mod. Si-Mn TRIP steel167 Mod. Si-Mn TRIP steel153 Mod. Si-Mn TRIP steel153 TRIP-aided DP steel168 Microalloyed Si-Mn TRIP steel154 Microalloyed Si-Mn TRIP steel158 Hot-rolled TRIP-aided steel157 Microalloyed Si-Mn TRIP steel162

— — — — — — — 11 10

CHAPTER FIFTEEN

15.42

Temperature, °C

Ac3

780°C 5 min 730°C

Ac1

430°C 15 s – 30 min 370°C

Water-quenching

Time, s (a)

T fin A3 Temperature

T1,t1 T1’

A1

T2

Time (b)

Temperature

Tnr 820°C 50°C/s

A3 50°C/s¥12 s 620°C 50°C/s

A1

1800s

330°C 10°C/min Time (c)

FIGURE 15.19 Schematic representation of (a) heat treatment condition for multiphase treatment consisting of an intercritical annealing followed by a bainite tempering stage,161 (b) thermomechanical path for TRIP-assisted multiphase steels, and (c) new thermomechanical path for TRIP-assisted multiphase steels.158

THERMOMECHANICAL TREATMENT

15.43

MPa◊%. However, the formability increases with increasing volume fraction of gR, C%, and plastic stability of gR.159,160 2. Through heat treatment (Fig. 15.19a)161 or various thermomechanical paths (Fig. 15.19b and c),158 it is now possible to stabilize considerable amounts of gR at room temperature in low-alloy ferritic [e.g., Fe-(0.1–0.2)C-(1–3)Mn-(1–2.5)Si] steels and microalloyed (e.g., Fe-0.16C-1.4Mn-0.45Si-0.015P-0.035Al-0.0045N-0.005Ti0.015Nb-0.003V) steels, respectively. 15.4.2 Stability of Retained Austenite As a major phase, the stability of gR is very important for enhancing TRIP steels. Large-sized gR is unstable and transforms to martensite at low strains, while smallsized ( Tm/2, where Tm is the melting point). The first description of superplasticity in a metallic material was reported by Bengough in 1912 in a + b brass, which exhibited an elongation of 163% at 700°C.276 However, Pearson277 was credited with the first observation of this phenomenon in 1934, when he dramatically showed the occurrence of exceptionally large (1950%) tensile ductility in eutectic Sn-Bi alloy. His sensational result and further work on superplasticity were largely ignored until 1962, when Underwood278 published a review paper describing the experimental work carried out on superplastic materials in the Soviet Union. Later, Backofen et al. studied Zn-Al eutectoid and Pb-Sn eutectic alloys in 1964.279 Of great significance is that Backofen and his coworkers280 showed that the superplastic Zn-Al could be formed into a useful shape by a simple air pressure operation like glass blowing. Since then, there was a great upsurge in the field of superplasticity. Subsequently, Lee281 reported a tensile ductility of 2100% in Mg-Al eutectic alloy, and Ishikawa et al.282 obtained an elongation of 2900% in Zn-Al eutectoid alloy. Higashi and coworkers283 found a maximum tensile elongation of 5500% in Al-bronze alloy. Honda et al.284 reported superplasticity in Cu38Zn-1.9Sn-1.9Pb alloy in the temperature range of 773 to 873 K and at initial strain rates of 8.3 ¥ 10-4 s-1 which showed shape memory effect after heat-treating at 1073 K, forming thermoelastic martensite. Ma and Langdon285 found a tensile elongation up to 7550% in Pb-62Sn eutectic alloy during testing at 413 K. Grant286 showed a fine grain size of ~0.3 mm in PM (rapidly solidified) Ni-Cr stainless steel (containing 12% B) with an elongation of 300 to 400% at 10-2 s-1 and 600% at 10-1 s-1 at 1000°C.287 The current world record of superplastic ductility is 8000%, obtained in a thermomechanically processed commercial bronze alloy based on the composition Cu-10Al-4.5Fe-6Ni-2Mn.288 Currently, superplasticity is extensively studied to understand the fundamental plastic flow and failure mechanisms and for its technological significance in superplastic forming (SPF) operation at a suitable temperature, called the SPF temperature, and within a given strain rate. There are now more than 100 alloy systems which have been shown to exhibit superplasticity.289,290 Among them, the main commercial alloys are: (a) the 7xxx series (e.g., 7475 and 7075 Al alloys); (b) 5xxx (e.g., 5083 Al alloy); (c) Al-Cu-Mg;291 (d) Al-6Cu-0.15Zr (Supral); (e) Al-Li alloys (e.g., 2090, weldalite, and 8090 alloys); (f) Al-5Ca-5Zn alloy; (g) Zn-22Al alloy; (h) Ti and Ti-based alloys (e.g., Ti-6Al4V);292 (h) nickel-based superalloy (such as IN100 and IN718) (prepared by conventional or powder metallurgy processing);276 (i) ultrahigh-carbon steels; (j) stainless steels;293 and so forth. A number of commercial SP alloys are listed in Table 15.11 together with the SPF temperature, total elongation, and strain rate sensitivity m.277,292,294 Among the Al alloys, Al-Cu-Mg, Supral, Al-Li, and the 7xxx series are the high-strength heat-treatable alloys. However, the lower operating strain rate in the 7xxx series leads to increased forming times and costs and probably limits its application to the aerospace industry. The eutectic 5083 Al and Al-Ca alloys are non-heat-treatable alloys and may be cast and heavy rolled to produce excellent superplastic ductility. Al-6Cu-0.15Zr (Supral) is the only alloy that can be superplastically deformed in the severely cold-worked state. In this case, the deformation is associated with the strain-induced continuous (dynamic) recrystallization process, where grain refinement slowly increases with strain. This results in a fine-grained structure without the usual nucleation and growth. Supral 210 and 220 (another modification of

TABLE 15.11 Superplastic Alloys and Their Associated Properties254,294 Alloy/grain size (mm)/grade

15.80

Ti-6Al-4V Ti-5.5Al-1Fe Ti-6Al-2Mo-4Zr-2Sn (6242) Ti-6Al-4Mo-5Zr-1Cu-0.25Si (IMI 700) IMI 834 Al-20Cu (strip made by casting/rolling) Al-33Cu Modified 2004 Al-6Cu-0.5Zr (Supral) Supral 100 (T6) Supral 210† (T6) Supral 220† (T6) Al-Ca Alcan 08050 (Al-4.5Zn-4.5Ca) Al-5Ca-5Si 7050 Al [+(0.08–0.12)Zr, (0.17–0.33)Sc] 7475 Al (fine-grain) 7475 Al (DR alloy) (6 mm) 7075 Al (T6) 7091 Al (PM) (3 mm) Al-4Cu-3Li-0.5Zr Al-4Cu-2Li-0.15Zr (or Al-3Cu-2Li-1Mg-0.1Zr) Al-1.2Cu-2.4Li-0.6Mg-0.1Zr (8090) Al-10Mg-0.1Zr (1.9 mm) Al-10Mg-0.6Zr (0.7 mm) Al-4.7Mg-1.6Mn (modified 5083) Al-4.2Mg-0.7Mn (commercial 5083) 5083 (PM) Al-4Mg-3Zn (Formall 570, ~7 mm) Al-16Si-5Fe (PM) Sn-38Pb Zn-22Al

Superplastic temperature, °C

Elongation, %

927 827 900 800 990 520 380–520

1000–2000 >400 538 300 300 354 1150

450 450 440 450 550 565 600 477 516 530 510 300 450 450

2000 600–1000 1350 1100 1500 500 900 744–1108 600–1200 ~2000 600 450 900

510 300 300 550 550 550 500 520 140 250

1000 1100 600 500 >400 465 300–800 ~400 4850 2000

Strain rate, s-1 -4

2 ¥ 10 1 ¥ 10-3 2 ¥ 10-4 2 ¥ 10-4 10-4 4.2 ¥ 10-3

10-3 10-3 10-3 10-3 10-3 1.2 ¥ 10-1 1 ¥ 10-2 2 ¥ 10-4 2.8 ¥ 10-3 10-4 8 ¥ 10-3

2 ¥ 10-4 2 ¥ 10-3 8 ¥ 10-3 4.2 ¥ 10-3 5 ¥ 10-4–1 ¥ 10-3 3 ¥ 10-5 2 ¥ 10-4–2 ¥ 10-3 1.38 ¥ 10-1 10-2

m 0.8 0.5 0.7 0.7 >0.6 0.32 0.9 0.5 0.38

0.3 0.63 0.75 0.67 ~0.5 0.38

Flow stress, MPa 10

10

1 2 — 3 4 5

9

2.8 18.7–20 2

6 7 8 9

0.6

10 10 11 12 — 13 14

0.5 0.53–0.61 0.6 ~0.61 0.5

Ref.

10

Zn-0.3Al Cu-10Al-4.5Fe-5.98Ni-1.6Mn (Al-bronze) Mg-Al eutectic Mg-AZ31 alloy (extruded condition) Mg-ZK 60 alloy Mg-ZK 61 alloy (P/M) Fe-1.6C-1.5Cr Fe-1.51C-3.38Al (DETWAD, 1 to 2 mm) Fe-1.4C-1.5Cr-6.7Al (1 to 3 mm) Fe-26Cr-6.5Ni (IN744) Fe-23Cr-5.6Ni-1.3Mo-0.12N (duplex) Fe-23.5Cr-5.7Ni-1.4Mo (duplex) Fe-24Cr-7Ni-3Mo (duplex) SAF 2304 (2 to 3 mm) (duplex) IN-100 (P/M) IN-718 †

R. T. 800 375 325 270 200 650 750 900 900 900 850 970 1010 954

1400 5500 2100 608 1700 659 1200 783 ~800 1000 2240 2240 750 450 1000 350–750

2 ¥ 10-4 1 ¥ 10-4 1 ¥ 10-3 10-4 8 ¥ 10-3 5 ¥ 10-5 2 ¥ 10-4 3 ¥ 10-3 10-4 -3

1.3 ¥ 10 –3.3 ¥ 10

15 16

~0.64 0.8 0.5 0.52

-5

17

0.46

45

0.76 0.76

21.2 21.2

0.4–0.75 0.5 0.28–0.83

~30 35 72 - 1000%). Note that this alloy is not prone to cavitation during superplastic deformation.302a

15.92

CHAPTER FIFTEEN

Supral ranges of Al alloys have been used in the aerospace, automotive, rail, and other industries, whereas Al-Li alloys (such as 8090, 2090, and 2095) are used mostly in the aerospace industry. 15.10.2.5 Superplasticity in Inconel Alloy. There is significant need for complexshaped airframe and engine components for commercial and military aircraft applications requiring high Ni-base alloys to withstand a combination of high strength, high temperature, and hot gas corrosion. Fine-grained (1 to 10 mm) SP Inconel 718 alloy [with varying proportions of the g and d (Ni3Nb) phases] can be used in these applications, which is reported to exhibit, at 954°C (1750°F), (1) 350% elongation at an initial strain rate of 1.3 ¥ 10-3 s-1 with flow stress of 2200% at 900°C and 2 ¥ 10-4 s-1 due to its optimum chemical composition and a specific manufacturing route (see Table 15.11).341 This SP duplex stainless steel is capable of complementary wrap forming (CWF) to produce auxiliary power unit (APU) components with properties comparable to those of Ti-6Al-2Sn-4Zr-2Mo SP alloy.341 The fine-grained (2- to 3-mm) Fe-22Cr-5Ni-3Mo-0.3N duplex stainless steel is found to exhibit maximum elongation of ~800% at 950°C and 10-3 s-1. Its elongation of ~300% at 900 to 1050°C and its high strain rates of about 5 ¥ 10-2 s-1 allow hot-forming operations such as deep drawing, blow forming, and die forging of complex parts at high deformation rates and low stresses.342

15.10.3 Recent Advances in Superplasticity This section describes new developments in a few specific areas where noteworthy progress has been made. These areas include superplasticity in nanocrystalline materials, intermetallics, and ceramics; high-strain-rate superplasticity (HSRSP); and low-temperature superplasticity (LTSP). The HSRSP and LTSP may be achieved by reducing the grain size below 1 mm and can be applied in the aerospace and automotive industries to increase productivity and reduce production cost.

THERMOMECHANICAL TREATMENT

15.93

TABLE 15.14 Superplasticity Observations in Nanocrystalline Materials344

Alloy Ti-6Al-3.2Mo Zn-22Al Ni3Al alloy Ni3Al alloy Electrodeposited nickel Electrodeposited nickel Al-Mg-Li-Zr alloy (1420) Al-Mg-Li-Zr alloy (1420) Al-Cu alloy (2124) Ti-6Al-4V

Grain size,† nm

Stress, MPa

.

-1

e, s

-4

60 80 50

2 ¥ 10 10-2 1 ¥ 10-3

20

1 ¥ 10-3

100

1 ¥ 10-1

100 70

1 ¥ 10-3 1 ¥ 10-3

T, °C

e = 0.1

Max.

Elongation, %

575 120 650 725 350 420 250 300 350 575

150 18 400 270 108 25 106 33 6 165

— — 1530 790 360 67 154 146 50 —

1200 250 380 560 300 900 330 850 405 215

Ref. 1,2 3 4 5 6 7 7

†

Describes the starting average grain size. Reprinted by permission of the Metallurgical Society, Warrendale, Pa. 1. G. A. Salischev et al., Materials Science Forum, vols. 170–172, 1994, p. 121. 2. G. A. Salischev et al., Materials Science Forum, vols. 243–245, 1997, p. 585. 3. R. S. Mishra, R. Z. Valiev, and A. K. Mukherjee, Nanostructural Materials, vol. 9, 1997, p. 473. 4. R. S. Mishra et al., Mater. Sc. Eng., vol. A252, 1998, p. 174. 5. Ref. 26 in Ref. 344. 6. Ref. 27 in Ref. 344. 7. Ref. 28 in Ref. 344.

15.10.3.1 Superplasticity in Nanocrystalline Materials. The nanocrystalline materials can be grouped into two types: (1) D-type produced from ingot or powder by severe plastic deformation and (2) S-type produced by sintering of nanocrystalline powders. Nanocrystalline materials have been extensively studied, which have increased our understanding of grain-size-dependent phenomenon to a much finer scale. The experimental results exhibit higher flow stresses for superplasticity in nanocrystalline materials. It is believed that slip accommodation plays a dominant role during superplasticity of nanocrystalline materials. The flow stress needed for slip accommodation in nanocrystalline materials is found to be higher than those for conventional SP materials.343 SP behavior has been observed in several nanocrystalline materials using miniature tensile specimens (Table 15.14),344 but the level of SP properties is lower than expected for such a small grain size. Formation of metastable states, disordering, and presence of supersaturated solid solution during severe plastic deformation may be possible factors of reduced SP properties in the nanocrystalline alloys investigated.345 15.10.3.2 Superplasticity in Intermetallics. Superplasticity has been observed in several intermetallics with fine microduplex structure (such as aluminides and silicides) or coarse-grained quasi-single-phase structures prior to deformation. (Their m values range, typically, from 0.32 to 1, and grain sizes are >10 mm.) In the former case, as in many SP alloys and ceramics, grain boundary sliding plays an important role, whereas in the latter, dynamic recrystallization or subgrain formation is needed for SP deformation.346 Lin and coworkers347 have reported superplasticity in Fe3Al and FeAl alloys with a grain size of 100 and 350 mm, respectively. They observed the SP behavior in Fe-28Al-2Ti (in at%) alloy exhibiting 620% elongation with m = 0.4 at an 850°C and an initial strain rate of 1.25 ¥ 10-3 s-1 and in Fe36.5Al-2Ti alloy exhibiting 297% elongation with m = 0.34 at 1000°C under an initial strain rate of 2.08 ¥ 10-2 s-1. The observed SP behavior is attributed to continuous

15.94

CHAPTER FIFTEEN

recovery and recrystallization. Lin and his coworkers348 also reported superplasticity in Fe3Al and FeAl alloys at 800 to 900°C and 2 ¥ 10-4 to 4 ¥ 10-3 s-1 and at 900 to 1000°C and 1.4 ¥ 10-4 to 2.78 ¥ 10-2 s-1, respectively. Ti aluminides such as TiAl (g), near-g, and Ti3Al (super-a2) alloys containing various alloying elements are promising candidates for applications as aerospace (and automotive) structural and engine components due to their attractive elevatedtemperature (as high as 650 to 800°C) specific strength, good oxidation resistance, and low density. The fine-grained Ti-Al (43 at%) (g alloy) is superplastic in the temperature range of 1000 to 1100°C with an elongation of 275%.349 The optimum SP deformation parameters for Ti-25Al-10Nb-3V-1Mo (a2 alloy) are ~800% elongation in the transverse rolling direction at 950°C and a strain rate of 8 ¥ 10-5 s-1.350 Two interesting areas have emerged in the development of superplasticity of intermetallics—low-temperature superplasticity in TiAl (g) and coarse-grained (~100-mm) superplasticity in Fe3Al- and FeAl-based alloys. For low-temperature superplasticity, a soft phase is intentionally introduced, by microstructural design, to accommodate effectively sliding strains at triple-grain junctions to retard and suppress cavitation and fracture. This approach is possibly applicable to many intermetallic alloy systems. For coarse-grained superplasticity in iron aluminides, although the actual SP deformation mechanisms are not established, scientifically and technologically important progress has been made. In the near future, SPF of intermetallics may possibly be employed only by the aerospace industry, mainly because of the high cost of intermetallics and relatively high SPF temperatures.290 This is of particular interest because intermetallics have very limited ductility at room temperature and cannot be processed by conventional means. 15.10.3.3 Superplasticity in Ceramics. SP ceramics cover a wide range of chemical composition, crystal structure, and physical properties, but all have the common feature of a stable, very fine-grained microstructure (150† 350

Ref.

1 2 2 3

4

†

Measurements are restricted by test conditions. 1. K. Matsuki, H. Matsumoto, M. Tokizawa, and Y. Murakami, Superplasticity in Advanced Materials, JSRS, Osaka, Japan, 1991, pp. 551–556. 2. K. Higashi, T. Okada, T. Mukai, and S. Tanimura, Superplasticity in Advanced Materials, JSRS, Osaka, Japan, 1991, pp. 551–556, 569–574. 3. H.-P. Pu and J. C. Huang, Superplasticity and Superplastic Forming, eds. A. K. Ghosh and T. R. Bieler, TMS, Warrendale, P., 1995, pp. 33–40. 4. Ref. 363.

current die and tooling methods.344 Lower forming temperature also offers the reduction in energy cost and surface oxidation.363 15.10.3.6 Quasi-Superplasticity. High-strain-rate sensitivity may be achieved in metallic materials by a mechanism involving no grain boundary sliding and fine grains. These materials are termed class I solid-solution alloys, in which the glide portion of the glide/climb dislocation creep process is rate-controlling due to solutedrag controlled dislocation motion (or creep). These alloys have modest m value (m = 0.33) and elongation of 200 to 400%. The extended elongation inherent in these solid-solution alloys would suggest that they may be categorized as quasisuperplastic materials (i.e., superplastic-like, or resembling superplasticity). Since these quasi-superplastic materials do not possess high strength at low temperatures, they are used mostly as secondary structural components rather than primary structural components (such as mechanically alloyed Al-Mg-Li alloys and UHC steel), which are truly superplastic at elevated temperature (made from ultrafine-grained materials with ultrafine second phases) with improved elastic stiffness, lightness, and very good tensile ductility at low temperature.299,363a

15.11 SUPERPLASTIC SHEET FORMING PROCESS SPF is a metalforming process that utilizes the SP material behavior to produce near-net shape components for the aerospace, transportation, and architectural applications. As SPF is a stretching process, the design engineer must take into account the amount of strain the component and forming method will induce into the sheet. This, in turn, will result in material selection and give an estimation of the final thickness of the component and thus the initial thickness. SPF of Al- and Tibased alloys has become a significant manufacturing method for aerospace engine

15.100

CHAPTER FIFTEEN

and fuselage components. Superplastic Inconel 718 can be superplastically formed at 954°C using flow stresses less than 72 MPa (10.4 ksi). Since SPF materials have low flow stresses, lower gas or air pressure [usually £500 psi (3.4 MPa)] is used to form the sheet in a reasonable time. For most superplastic sheet forming operations,the material flow stress should be 107 (shear) 700 (tensile)

Solid phase Solid phase

850 900

10 20

1 1

Solid phase

990

20

1

450 (tensile) 406 (shear) 798 (tensile) 488 (shear)

THERMOMECHANICAL TREATMENT

15.107

FIGURE 15.47 Three basic forms of Ti-6Al-4V structures fabricated by SPF/DB process.367,380,382 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

For the fabrication of integrally stiffened monolithic structure (Fig. 15.47), a pattern for bonding and forming is set up, and a stop-off compound (such as a mixture of yttria and boron nitride in a polymeric binder)375 is used selectively to one of the sheets of the mating surfaces, to prevent bonding in specific regions corresponding to the die cavity or cavities. Initially, external gas pressure is applied to the sheets at a suitable temperature (927°C) to produce selective diffusion bonding. This is followed by introduction of gas pressure between the sheets to superplastically form the unbonded (unmasked) regions into the tooling cavity or cavities. The boron nitride, used earlier as a stop-off compound, now serves as a lubricant. In some designs, additional details are put in the die cavities for post-form bonding and selective stiffening.306 SPF/DB sandwich structure is fabricated by utilizing three or more titanium sheets and selectively bonding and forming with a desired core pattern. Advantages of this structure include380,382 1. The outer configuration of the fabricated sandwich part depends on the depth and configuration of the tooling cavity. 2. The core configuration is related with the stop-off pattern, which can be varied and modified to a large extent without a change in tooling. 3. The process inherently renders greater flexibility in edge closure and core designs. Thus a conventional honeycomb sandwich structure or panel can be fabricated with a considerable cost reduction. Moreover, the truss members of the SPF/DB core and the edge member can become an integral unit by using the proper DB pattern. This leads to cost savings (through reduced production steps) and enhanced structural efficiency due to simplified shear ties. The F-15E Built-up Low Cost Advanced Titanium Structure Program gives a good example of the advantages of SPF and SPF/DB processes. In this program, the

15.108

CHAPTER FIFTEEN

FIGURE 15.48 A variety of structural parts fabricated from Ti-6Al-4V sheet representing three forms of SPF/DB structures.382 (Reprinted by permission of The Metallurgical Society, Warrendale, Pa.)

number of aft fuselage components was reduced from 772 to 46, and 10,000 fasteners were removed. The load factor capability of the aircraft improved, weight reduction was great, and there was an increase of an additional 10 cubic feet of equipment space.383

15.11.8 Aerospace Applications of SPF/DB Ti-6Al-4V Structures Several thousand titanium DB/SPF structural components have been used in service, e.g., as access panels on Airbus aircraft and civil and military applications. Figure 15.48 illustrates a variety of structural parts fabricated from Ti-6Al-4V sheet representing the three forms of SPF/DB structures.380 Applications of reinforced titanium sheet structures include the nacelle frame from B-1 aircraft (to withstand exposure to high temperature), monolithic helicopter fire-wall, and hemispherical propellant tank. Examples of integrally stiffened SPF/DB sheet structures are hollow structures, B-1 auxiliary power unit (APU) door, F-15 aft nozzle fairing, and toroidal tank. Important applications of SPF/DB sandwich structures are the B-1 windshield hot-air blast nozzle assembly; the 279 cm (length) ¥ 152 cm (width) ¥ 3.8 cm (height) ¥ 3.8 cm (deep compound curvature) (110 in. ¥ 60 in. ¥ 1.5 in. ¥ 1.5 in.) B-1 engine access door (which is a complex monolithic structure); T-38 main landing gear strut door (which is a flight component of the operating aircraft); F-4 horizontal stabilator inboard trailing edge; F-14A wing glove vane (which is a flight testing aircraft component); 454-kg (1000-lb) wing and fuselage sections of an advanced fighter;

THERMOMECHANICAL TREATMENT

15.109

circular sandwich; four-sheet sandwich for Tornado aircraft (for British Aerospace).255,382 Other applications of the SPF/DB process include hollow fan blades; F-100 aft fan augmenter duct segment; F-100 augmenter seal; and gas-turbine stator vanes. It is concluded that Ti-alloy SPF/DB structures can replace highly expensive and complex aluminum alloy and stainless steel structures. From a commercial standpoint, there is currently a possibility of more cost-effective designs for an advanced supersonic transport where higher corrosion resistance and strength and longer life are required.372

15.11.9 DB/SPF Al Alloys Previously, aluminum alloy structures have been limited to SPF parts because diffusion bonding of aluminum alloy was not successful306,375 due to insufficient hot peel strength of Al alloy bonds and the existence of a tenacious Al oxide layer on the surface. Recently, DB/SPF 7475 Al sheet has been successfully produced with ductile, oxide-free bonds at considerably lower bonding pressure (200% and comparable processing cycle time (3 kHz. These new transformers have low power losses and very low inductance, and enjoy loadmatching flexibility.49 16.2.2.9 Steel Grades for Induction Hardening. The induction hardening process is carried out on plain carbon and low-alloy steels containing sufficiently high carbon content, usually in the range of 0.3 to 0.5%, in order to transform the surface layers to martensite after quenching from the austenitizing temperature, which will give rise to surface hardness in the range of 50 to 60 Rc. The upper value is achieved with high carbon content, which may run the risk of quenching or hardening cracks. However, if well-controlled heat treatment is carried out, higher carbon contents are permissible for rolls made of steel containing 0.8% C, 1.8% Cr,

† This rating term for capacitors identifies the amount of reactive volt-amperes that the capacitor can supply to the circuit when run at a particular voltage and frequency.18

SURFACE HARDENING TREATMENTS

16.27

and 0.25% Mo.8 Other materials that can be induction hardened are cast irons and sintered metals. Plain carbon steels are usually water-quenched, whereas the alloy steels are quenched in air, quenching oils, oil emulsion media, or polymer quenchants (such as PAG-based solution). Before induction hardening, steels should be in either the normalized or the hardened and tempered condition. Steels containing more than 0.5% carbon in the fully spheroidized state should not be chosen for induction hardening due to their poor response to induction hardening; higher temperatures and/or longer heating-up times are needed for dissolution of spheroidal carbides compared to those needed for carbides formed in normalized or hardened and tempered steels. Such conditions may produce coarse austenite grain size at the surface, which will result in coarse martensite structure combined with a high percentage of retained austenite; this may reduce fatigue resistance and promote seizing or galling.12 16.2.2.10 Pre- and Post-Heat Treatments. Prior to application of higherfrequency induction heat for rapid austenitization, preheating is carried out using a low-to-medium frequency until the desired temperature (⬃400°C) is reached. Preheating makes the attainment of heating and hardening patterns on irregular shapes easier; it increases the hardened depth, and perhaps it provides better size and residual stress control. Preheating in contour gear hardening ensures a reasonable heated depth at the roots of the gear, the attainment of the required metallurgical results, and a decrease in distortion; reduces the amount of energy needed in the final heat; and permits the high-powered RF current applied during the final stage to follow the contour of the gear.18 Tempering of the induction hardened surface may be carried out separately in a conventional furnace or by the use of a second induction coil at lower power densities. The latter process, called induction tempering, is a short-time high-temperature treatment rather than a conventional long-time low-temperature treatment. (See Sec. 14.9.1 also for more details.) Induction tempering, like induction hardening, has become a viable commercial process, replacing the conventional furnace operations in many high-production applications such as oil well pipes, railroad rails, tubes, ball screw, shaft, and bar.12,16 It may be pointed out that time and temperature are both critical in induction tempering. Hence, it is necessary to establish equivalent time and temperature, using an extension of the Hollomon-Jaffe correlation with appropriate tempering parameter P when induction tempering lines are set up initially or subsequently modified. (See Sec. 14.9.1 also for more details.) This extended relation takes into account rapid heating at fixed temperatures as well as continuous heating and continuous cooling. This is done by measuring an effective tempering time t* for an isothermal temperature heating cycle which corresponds to the continuous cycle. For continuous heating from room temperature to typical induction tempering temperatures, an increment in t (that is, Dti, usually 0.005 to 0.01 times ttotal, where ttotal is the total heating time) provides sufficient calculation accuracy.16 Induction tempering is a valuable tool for manufacturing cells. The main advantages of induction tempering are52 1. Reduced energy cost, because the energy induced into the part is often confined to the hardened region 2. Precise control of power, monitoring of the final temperature of each component, and improved working environment

16.28

CHAPTER SIXTEEN

3. The shortness of the time/temperature relationship to accomplish tempering compared to furnace tempering, thereby minimizing the number of parts in process to just a few. 16.2.2.11 Computer Simulation of Induction Hardening Processes. Computer simulation is an effective tool for induction heat-treating processes and improved coil design. There is no universal software program available that can accurately simulate all features of the induction heat-treating process. Hence the best approach consists of electromagnetic and thermal field simulation; and if desired, the thermal fields can then be exported to a structural transformation program.53,54 There are a number of different software programs available for simulating the induction heating process, but only a few of them, notably Flux 2D (a 2-D coupled electromagnetic plus thermal program) and ELTA (Electro Thermal Analysis) (1-D coupled electromagnetic plus thermal program), are adequate for induction coil optimization. Figure 16.14a and b shows the temperature distribution in an induction scanning application. Figure 16.14a is a map of temperature and isolines at specified temperatures. The white background represents the full austenitization temperature needed for complete martensite formation. In this process, induction was used to harden a 1.8-in.-O.D. shaft for the automotive industry. The frequency for this application was 3 kHz, and the scanning speed was 0.3 in./s. Using ELTA, it was possible to make a good 1-D approximation of this 2-D system.54 Some researchers have suggested a hierarchical approach (or a rule of pyramid), whereby the simpler software packages are employed as a foundation for the more complicated simulations, if they are required. This approach has been used in practice and found to significantly reduce design time and process development costs. The main source of error in simulation of induction heat-treating processes lies in the accuracy of the material property and heat-transfer coefficient data. More experimental study is greatly needed in this area to create a database for accurate simulation of induction heat-treating processes.54 16.2.2.12 Applications. This process is now employed as an effective and economical method of hardening a very wide range of components in the general engineering and automotive industries. Components which are usually induction heat-treated range from bolts to crankshafts.32 Induction heating may also prove to be an ideal technique for annealing, stress relieving, and through-hardening and tempering, and hot forming processes such as forging, extrusion, rolling, swaging, and bending.17 Crankshafts, camshafts, splined shafts, transmission shafts, driveshafts, tube-type shafts,55 universal joints, various gears, valve seats, rocker arms, cylinder heads, wheel spindles, and ball studs are a few of the components chosen for selective surface hardening in the automotive power train industry. Other examples of applications include selective crown hardening of railroad rails and torque hub assembly.56 The frequencies commonly used range from 10 to 450 kHz for the smaller articles in motor cars and from 1 to 10 kHz in the heavy-vehicle industry.57 The production of induction-hardened cast pearlitic malleable iron transmission gears has now been widely used in the automotive industry because of various advantages, which include:36 (1) cheaper casting than an alloy steel forging (material saving); (2) ease of machining cast iron (machinery saving); (3) use of less energy in induction heat treatment; (4) large decrease in noise level; (5) increased wearability due to malleable cast; (6) increased dimensional and distortion control; and (7) cleaner and cooler working conditions.

SURFACE HARDENING TREATMENTS

16.29

(a)

(b)

FIGURE 16.14 (a) Simulation of induction scanning using ELTA program.54 (b) Temperature graphs at different radii versus time for the induction scanning process using ELTA program.54 (Courtesy of Centre for Induction Technology, Inc., Michigan.)

Surface hardening in the machine tool field includes lathe blade, machine column, and transmission gear and shafts. Other examples include contour gear hardening by the dual-frequency method,28 tooth-by-tooth hardening of large fraction and other gears up to more than 1422 mm in diameter, and surface hardening of hydraulic press tool holders weighing up to 5000 kg. This clearly demonstrates that size is no problem for the induction hardening process. However, in these cases,

16.30

CHAPTER SIXTEEN

specially designed inductors should be used to ensure closer tolerance of localized heating patterns.57 Applications in the metalworking and hand-tool fields include rolling mill rolls, pliers, hammers, chisels, screw drivers, axe and hatchet products, cutters, and hacksaw blades. Through-hardening applications include (1) oil country tubular products; (2) structural members; (3) spring steels; (4) chain links; (5) commercial airframe components;58 (6) head hardening of rails; (7) austempering of lawn mower blades by induction austenitizing and quenching in a salt bath at 650°F to produce the desired bainitic microstructure without distortion; and (8) hardening of sharp edges of axe and hatchet product using channel-type inductor coil, 20- to 50-kW system, and in-line quench product at a rate of 1000 parts/day. 16.2.2.13 Safety Practices of Induction Hardening Equipment. For better performance of the induction hardening equipment, the following safety practices should be adopted:17 1. Regular visual inspection should be made of inverter for evidence of water leakage or cracked hose, arcing or overheating of component and electrical connections. 2. Inverter doors must be closed at all times to avoid the accumulation of contaminants within the enclosure. 3. Operation of all protective circuits—door switches, pressure switches, and temperature switches—should be properly carried out. 4. Air (or coupling) gaps between the coil and workpiece must be constantly maintained once they have been established, because deviations cause under- or overheating. For induction heating of fillet areas, as on axle shaft and for scan inductors, more coupling gap should be provided due to the likelihood of distortion of the workpiece.59 5. Where an inductor has been grounded, it is necessary that grounding is not dispensed with or disconnected, for the safety of the operator. 6. All inductors and work coils must have adequate flow,† volume, pressure, and temperature of deionized cooling water supplies without undue concentration of solids in the form of salts, carbonates, and atmospheric source pollutants, in order to lessen failure of inductors. A water flow switch should be incorporated if the primary pump fails. 7. Preventive and protective measures must be considered for all dangerous and hazardous factors specific to induction heating systems, such as electric shock at high frequency, magnetic fields, arcing, and melt metal, in order to meet the safety standards.22

16.2.3 High-Frequency Resistance Hardening Recently, a new method of selective surface hardening using high-frequency contact resistance has been developed in which a small area of the component becomes part † Adequate water flow for proper cooling of copper coil is given by gpm = PK1K2/K3DT, where gpm = gallons per minute; P is total coil power, kW; K1 is a tubing coefficient (for most high-frequency heattreating applications), usually 0.5; K2 = 3415 is a conversion constant that is derived from Btu/kWh; K3 is a conversion constant denoting the heat capacity of water, typically 500; and DT is the allowable temperature increase in the cooling water, usually 40°F or less.

SURFACE HARDENING TREATMENTS

16.31

FIGURE 16.15 Basic principles of highfrequency resistance hardening method.60 (Courtesy of Wolfson Heat Treating Center, England.)

of the coil.60 Figure 16.15 shows the basic principles of this method. High-frequency current in the range of 300 to 500 kHz is supplied directly to the component through two small contacts at the ends of the area to be hardened. The current then flows through a proximity conductor placed near the surface to be heated. This produces a high current density and very rapid (usually less than 0.5 s) and highly localized heating of the component surface beneath the conductor. The heated area is selfquenched very rapidly by the adjacent large mass of cold material when the current flow ceases.60 In conclusion, when compared to conventional heat treatment, both the energy consumption and distortion are very low. In addition, the process is very fast; i.e., a high production rate of parts in the automotive or appliance industries is achieved. It requires a high-frequency power supply and can be employed with a variety of interchangeable contact systems for hardening a number of different components.60

16.2.4

Laser Surface Hardening

Laser surface hardening (LSH), being a promising alternative to conventional thermal surface hardening, has been investigated for more than two decades.61 In LSH, thermal energy is produced by absorption of the laser radiation at the surface. The increase of temperature in the interior of the workpiece is due to conduction only. In laser (beam) surface hardening (LSH) treatment, unlike the flame and induction hardening processes, the large mass of unheated iron-base alloys serves as the quenchant. Figure 16.16 shows the principle of laser beam hardening treatment (LBHT) by shaping and integration techniques. An LBHT machine comprises the following separate elements:62–69

CHAPTER SIXTEEN

16.32

Oscillating Flat Mirror

Focusing Mirror

Lissajous Pattern

Laser Source

LB

Oscillating Flat Mirror

Orthogonal Oscillation Technique

Beam Shaping Technique

Multifaceted Concave Beam Integrator Beam integrated image

Beam Integration Technique

FIGURE 16.16 Principle of laser (beam) hardening treatment by shaping and integrating techniques.62 (Courtesy of The Institute of Metals, England.)

1. The CO2 laser consists of an optical resonant cavity where a gaseous mixture of CO2, N2, and helium flows. An electric discharge excites the gas mixture, emitting photons in the form of a high-power-density coherent beam from the end of the cavity. The Nd:YAG laser consists of a crystal in a gold-plated cavity, excited by flash lamps and emitting high-power-density coherent beams. 2. The laser beam (intense heat source) is brought to the working table by focusing optics, either mirrors or lenses, or by optical fiber (for Nd:YAG only). The focusing optics can shape the beam into a standard pattern such as a 0.25-in. (6.3-mm) square or specialized configurations such as a horseshoe, depending on the process profile.63 Optical fibers are available for Nd:YAG lasers having shorter wavelength of 1.06 mm; however, power transmission fibers for the CO2 laser with longer wavelength of 10.6 mm have not yet been developed. Alternatively, hollow waveguide sapphire tubes or shiny tube or square section waveguides may be used.64 3. The beam can be focused by manipulation of lenses and mirrors on the surface to be heat-treated. This can be achieved in two ways: (a) by the beam oscillating method, where the last two mirrors are oscillated with a certain amplitude and phase shift in two perpendicular directions, so that Lissajous curves can be obtained to cover the surface to be heat-treated; and (b) by the beam integration methods (for

SURFACE HARDENING TREATMENTS

16.33

spots >3 mm). These methods include scanning or rastering the near focused beam at sufficiently high speed to avoid melting; passing the beam through a reflective tube of the required shape, called a kaleidoscope, and imaging the beam at the exit from the kaleidoscope with lenses; and concentrating or focusing each part of the beam on the same region by an integrating mirror, i.e., segmented polished mirror. Another method uses integration by specially shaped lenses or even holographically etched mirrors, called kinoforms.64 4. To capture the energy of the beam and to hold the heat in the component to be treated, the component surfaces should be covered with an energy-absorbing black paint, colloidal graphite, spray paint (flat black), india ink, or a coating of zinc-, potassium-, or manganese phosphate [(Mn,Fe)5(H2PO4)4·4H2O], and a few proprietary black oxide coatings. They may also be etched or sand-blasted. Mn phosphate is more popular, but black spray paint and colloidal graphite are usually the most efficient and economical coating. In all cases, the reflectivity of the surface must be reduced to 0.5 mm) at higher carburizing temperatures (925 to 955°C, or 1700 to 1750°F) followed by direct transferring of the workpieces to a neutral salt bath at 845°C (1550°F) for stabilization, reduced distortion, minimal retained austenite, and finally direct quenching in marquenching oil at 175 to 260°C (347 to 500°F), depending on the alloy content and the hardness desired.114 When alloy steels are quenched directly, the case hardness is found to be low, and so they are given a doublequenching treatment. The first quench of this process involves reheating followed by quenching in oil or salt bath and a second requenching treatment at a somewhat lower temperature.8 Application. Liquid carburizing generally is used for small- and medium-sized parts requiring case depth less than 0.5 mm. Greater economy is obtained with smaller case depth requirement due to very high heating rates attained in this process. This is highly recommended (compared to gas carburizing) for highly stressed components in critical applications such as bearing steels for ball and socket joints, racing machines,116 transmission shaft with integral gear, and layshaft gear in order to obtain superior fatigue strength. Liquid carburizing is not recommended for parts containing small holes, threads, or recessed areas due to difficulty in cleaning.114 Noncyanide Liquid Carburizing. Noncyanide liquid bath contains mainly special grade carbon and carbonates.12,114 In this bath, mechanical stirring is done by means of one or more simple propeller stirrers to disperse the carbon particles in the molten salt. The chemical reaction between these ingredients may cause the generation of CO and the adsorption of CO on the carbon particles which are presumed to react with steel surfaces in much the same manner as in pack or gas carburizing.12 The noncyanide liquid carburizing bath is characterized by a higher rate of graphite consumption than a cyanide bath. Normally, replenishment is necessary every hour to maintain a proper bath activity.114

16.54

CHAPTER SIXTEEN

This type of bath usually is operated at temperatures higher than those of conventional liquid carburizing (i.e., cyanide-type) baths. Temperatures in the range of 900 to 955°C (1650 to 1750°F) are usually recommended while temperatures below 870°C (1600°F) are avoided, possibly due to the resulting decarburization of the steel. Advantages of this noncyanide carburizing include the following: 1. Parts that are cooled slowly following noncyanide carburization are easily machinable compared to parts slowly cooled following cyanide carburizing, because of the absence of nitrogen in the former case. 2. As a result of reason 1, parts that are quenched after noncyanide carburizing contain less retained austenite than parts quenched after cyanide carburizing. Disadvantages of noncyanide carburization are as follows: 1. Control of carbon potential and consistency of carburizing have proved to be difficult. 2. Evenly dispersed solid suspensions are difficult to maintain in large salt baths.116 Another variation of a cyanide-free regenerator has been developed recently by Degussa and is suited ideally for automation with molten salt bath quenching. Other advantages of this patented CECONSTANT salt bath carburizing using a low percentage (10%) of cyanide in the salt bath reaction are (1) minimum bailout and waste disposal, (2) safer quenching in molten nitrate salt bath, and (3) attainment of closely controlled carbon potential of 0.5, 0.8, and 1.1% possible with minimum bailout and waste salt disposal.116 16.3.1.5 Gas Carburizing. Gas carburizing has now become the most widely used method of case hardening of large volumes of production, its growth increasing every year at the expense of pack or salt bath carburizing. Therefore this process is often referred to as case carburizing. This method imparts high fatigue strength, high wear resistance, and retention of a tough and ductile core in the complex parts.117 Shallow-case carburizing with case depth 1 mm (0.04 in.) is recommended for differential gears, axles, and steering system components. Carburized parts vary in weight from about 45 g (0.1 lb) for planet gear shafts to 3.5 kg (8 lb) for output shafts. Pinion, sun, and ring gears weigh about 180 g (0.4 lb), 450 g (1 lb), and 1100 g (2.5 lb), respectively.118 Controlled carburizing atmospheres are produced by blending a carrier gas with an enriching gas, which acts as the source of carbon. The usual carrier, endothermic gas (i.e., endogas), not only is a diluent, but also serves as the accelerator of the carburizing reaction at the component surface. The amount of enriching gas required for gas carburizing depends mainly on the carbon demand, i.e., the rate of absorption of carbon by the workload. Endogas, consisting of CO, H2, CO2, and N2 with smaller amounts of CH4 and H2O, is produced by reacting a hydrocarbon gas such as natural gas (mainly methane), propane, or butane with air. For endogas produced in an endogas generator from methane, the air/methane ratio is about 2.5; from pure propane, the air/propane ratio is about 7.5; both of them will produce an O/C ratio of about 1.05 in the endogas. The air/fuel ratio, however, will vary with the composition of the hydrocarbon feed gases and the amount of water vapor in the ambient

SURFACE HARDENING TREATMENTS

16.55

TABLE 16.9 Specific Gravity and Composition of Natural Gas in Three Locations of the United States119 Composition, vol% State New York Illinois California

Specific gravity

CH4

CH3CH3

N2

CO2

0.58–0.59 0.57–0.61 0.60–0.63

94.1–96.3 89–97.5 92–98.8

1.8–2.0 1.6–4.4 3.9–5

0.3–1.8 0.31–5.7 1.2–1.24

0.83–0.96 0.39–0.75 0.76–3.0

CH4, methane; CH3CH3, ethane; N2, nitrogen; CO2, carbon dioxide. Source: American Gas Association. Reprinted by permission of ASM International, Ohio.

air. Table 16.9 lists typical compositions of natural gas.119 Sometimes a purified exothermic gas enriched with 5 to 10% hydrocarbon gas (e.g., natural gas or certain propanes) is used as the carrier gas for gas carburizing. Liquid hydrocarbons are also used as sources of carburizing gas. These liquids are mostly proprietary compounds that range in composition from pure hydrocarbons such as terpenes, dipentene, or benzene to oxygenated hydrocarbons such as alcohols, glycols, or ketones. In general, the liquid is fed in droplet form to a target plate in the furnace, where it volatilizes immediately. The vapors dissociate thermally to provide a carburizing atmosphere comprising carbon monoxide, carbon dioxide, methane, and water vapor. Forced fan circulation causes uniform temperature and even distribution of the atmosphere within the furnace. The control of liquid flow is effected either manually or automatically to ensure the desired carbon potential. Figure 16.27 is the microstructure of a 1018 steel gas-carburized at 927°C (1700°F) for 12 hr, followed initially by furnace cooling and finally by air cooling to room temperature. This microstructure illustrates a surface carbide network outlining the prior austenite grain boundaries in the pearlite matrix.120 Furnaces. Gas carburizing furnaces are classified into two groups: (1) continuous-type furnaces, e.g., mesh belt type, pusher type, shaker hearth, rotary hearth, and integrated pusher/rotary hearth furnace line, and (2) batch-type furnaces, e.g., pit and horizontal sealed quench furnaces. The selection of furnace type depends on the size, shape, quantity, and production run of workpieces and variety of case depth, fixtures, and space requirements.117 For example, where economical mass production (reduction of energy consumption by 50% or more) with reproducibility and high production rate of similar components of case depths between 0.015 and 0.12 in. (0.4 and 3 mm) (such as in automobile gears and parts) is desired, a continuous-type furnace is the best choice. In these furnaces, the well-separated workpieces, with or without fixtures, enter at one end and pass through preheating, purging, soaking and carburizing, and diffusion zones followed by either cooling or transformation zone for two-stage hardening treatment or oil-quenching zone for single-stage direct hardening treatment. The tempering zone is located last in the continuous line from which workpieces come out in the fully heat-treated condition. If an oil quench is used, parts must be washed prior to tempering to remove oil. Batch-type furnaces are preferred for small lots with varying case depths, such as large industrial-duty gears, components of machine tools, and material handling equipment. In horizontal batch furnaces, components in small batches are loaded

16.56

CHAPTER SIXTEEN

FIGURE 16.27 Microstructure of a 1018 steel (gas) carburized and diffused at 927°C (1700°F) for 12 hr, furnace-cooled at 538°C (1000°F) for 2 hr 10 min, and then air-cooled to room temperature. This microstructure shows a high surface carbon content (⬃1.1%) and a carbide network outlining the austenite grain boundaries in the pearlite matrix.120 Etched in 1% nital. (Reprinted by permission of ASM International, Materials Park, Ohio.)

on the heat-resistant fixtures in the furnaces, preheated, soaked and carburized, and diffused at the required temperature and time. They are then furnace-cooled in the same or different cooling chamber or quenched directly in oil in directly quenched batches. In pit-type furnaces, contact of the hot load with air occurs prior to quenching while in horizontal sealed quench furnaces, air ingress does not occur because quenching is done here under protective atmosphere.121 Furnace Atmosphere Parameters. For uniformity of carburizing, the furnace should be equipped with internal fans for good circulation of the atmosphere through the workload. The individual parts must be well separated to allow atmosphere to penetrate the load. Critical parts should be put on fixtures. The furnace should be operated at a positive pressure of 12 to 37 Pa (0.09 to 0.28 torr, or 0.015 to 0.15 in. column of water). It is a good practice to purge air entering the furnace during charging of the parts by using high flow rates of carrier gas. Alternatively, an automatic control system can be used to increase the flow of hydrocarbon enriching gas, compensating for the air entering during door openings.119 Gas Carburizing Atmospheres. The carburizing atmospheres are produced by combustion of methane or propane present in natural gas or other hydrocarbon gas in endothermic (or exothermic) gas generators. Endogas is a very complex mixture of CO, N2, H2, CO2, H2O, and CH4, and its free and rapid circulation and composition control are very important. In this furnace atmosphere, CO and CH4 are the sources of carbon; N2 is inert and acts as a diluent. Many constituents of this atmosphere react with the steel at high temperature in the austenite range, and several reactions take place simultaneously. The most important reversible reaction is 2CO ∫ CO2 + C (in solution in austenite)

(16.21)

SURFACE HARDENING TREATMENTS

16.57

After the equilibrium composition of the gas is determined, its carbon potential (or carbon in solution of austenite) at any temperature is obtained by the equilibrium constant at a given pressure Kp, which is written as Kp =

pCO2 ac 2 pCO

(16.22)

where pCO2 and pCO are the partial pressures of CO2 and CO, respectively, and ac is the activity of carbon in austenite. The equation can be represented in the form Kp =

pCO2 fc ( wt% C) 2 pCO

(16.23)

where fc is the activity coefficient of carbon. Rearranging Eq. (16.23), we get wt% C =

2 K p pCO fc pCO2

(16.24)

The sum of partial pressures is equal to the total pressure, so pCO + pCO2 + pinert = 1, where pinert is the partial pressure of inert gas, e.g., nitrogen present. Usually parts are carburized by two-step cycles. During the first step, known as a carburizing step, CO content of the atmosphere is greater than the partial pressure required to maintain a required carbon content; the reaction in Eq. (16.21) will proceed to the right direction, and carburizing will take place until a new equilibrium or very high carbon concentration, say 1.2%, is attained. During the second step, pCO2 >> pCO; the reaction will proceed to the left direction, and the steel surface will lose carbon, i.e., decarburization will occur so as to decrease the surface carbon to a lower level, say 0.9%.122 Thus in the former step, a high-potential atmosphere results in a rapid carbon penetration and a high surface carbon concentration, say, 1.2% carbon. In the latter step, often called a diffusion step, carbon potential is adjusted and maintained in order to produce desired surface carbon concentration in the finished part and diffuse the initially high surface carbon to deeper levels of the case.123 Equation (16.24) requires a knowledge of fc, which varies as a function of temperature and composition of the austenite.124 In addition to the carburizing reactions with CO and CO2, many other reactions may occur as follows: CH 4

∫ 2H 2 + C (in solution in austenite) CH4 + CO2

∫ 2CO + 2H2 CH4 + H2O ∫ CO + 3H2 CO + H2O ∫ CO2 + H2 CO + H 2 ∫ C (in solution in austenite) + H 2O

(16.25) (16.26) (16.27) (16.28) (16.29)

Equation (16.25) represents the principal carburizing reaction for either (1) an atmosphere composed of hydrocarbon gas diluted with nitrogen or (2) a methane atmosphere used in vacuum carburizing. The reaction coefficient for Eq. (16.29) is quite high, suggesting that this reaction controls the carburizing mechanism of the overall process for the gaseous atmosphere containing substantial amounts of hydrogen and carbon monoxide. Note that CO2 and water vapor present in the reac-

16.58

CHAPTER SIXTEEN

tion product are potent decarburizing agents; therefore, these gases must be removed quickly for the carburizing reaction to proceed. The CO2 and H2O content which can be tolerated without causing decarburizing may be calculated from equilibrium data. In gas carburizing, it is primarily the chemical balance among competing Eq. (16.21) and Eqs. (16.25) through (16.29) that determines the chemical potential of the atmosphere, which, in turn, determines the carbon content at the surface of the component. The optimum control of carburizing process can be achieved by using computer dynamic control technology, which can accurately control the quality of the case. Atmosphere Reactions during Gas Carburizing. Carbon transfer from the atmosphere to the workpieces under constant carbon potential can be described by the following scenario:125 1. Reactions (16.21), (16.25), and (16.29) are “fast” in both directions. All other reactions are slower. 2. As carbon is transferred to the workpieces, reactions (16.21) and (16.29) consume CO and H2, respectively, to form CO2 and H2. 3. To maintain a constant carbon potential, CH4 must react with CO2 and H2 to restore the gas reactions characteristics of the carbon potential [e.g., Eq. (16.24)]. 4. Since reactions (16.26) and (16.27) are “slow,” the amount of CH4 required must be many times the equilibrium amount in order for the “slow” reactions to proceed as quickly as the “fast” reactions. 5. The net result of Eqs. (16.21), (16.26), (16.27), and (16.29) is just Eq. (16.25). Therefore, as carburizing proceeds (at constant carbon potential), CH4 is consumed and H2 is generated. 6. An important reason to maintain a flow of carrier gas is to remove the H2 generated by the carburizing reaction. If the carrier gas flow is too low for the carbon demand, H2 will build up in the atmosphere and atmosphere CO content will significantly decrease. (A decrease of a few percent is normal for most operations; however, a decrease of 5% is too much.) 7. However, as the carrier gas flow rate increases, the dwell (or residence) time of the atmosphere within the furnace decreases. As a result, the amount of CH4 needed to drive Eqs. (16.26) and Eq. (16.27) to the right becomes greater. 8. Generation of soot at cold spots within the furnace tends to increase the carbon demand. If sooting becomes so common that brickwork and metal surfaces are covered, all reactions become sluggish because they depend on having catalytically active surfaces available. In a sooted furnace one observes that the atmosphere carbon potential does not respond to changes in enriching gas flows. In situ Carburizing Atmospheres. Recently, industry’s demands for highly flexible heat treatment installations to produce high quality and remain energy-efficient are best met by fully automated batch furnaces running with in situ produced alternative gas carburizing atmospheres. The two atmospheres for which there is the best documentation are as follows:125 1. A blend of air and a hydrocarbon gas (such as propane and methane) is introduced directly into the carburizing furnace. If the atmosphere flow rates are relatively low and the furnace temperature is relatively high (e.g., 925°C), the furnace gas composition for a specific carbon potential is the same as when a

SURFACE HARDENING TREATMENTS

16.59

traditional endothermic carrier gas is employed. At higher flow rates and/or lower temperatures, the percent of CH4 in the atmosphere at a given carbon potential will be greater. 2. A blend of nitrogen and methanol, in a 1 : 2 molar ratio, is introduced into the furnace to provide the carrier gas. 2N2 + CH3OH + heat Æ CO + 2H2 + 2N2 Methane addition serves as an enriching gas to satisfy the carbon demand. When properly set up, either of these in situ processes can produce results equivalent to conventional gas carburizing with an endothermic carrier gas. The second method is useful primarily when inexpensive sources of hydrocarbon gas are not available. Advantages claimed in the use of alternative nitrogen-methanol gas atmosphere over the conventional endogas atmosphere are as follows: 1. It has prompt availability. 2. It is economical: (a) It takes only a few minutes to start up and introduce this atmosphere in the furnace and to achieve the right chemistry, whereas several hours are wasted in the case of endogas generation; (b) It takes less than 10 min to purge and shut down compared to 1 or 2 hr in the case of endogas generation. 3. There is no sooting problem; i.e., it largely contributes cleaner surfaces. 4. There is much lower rate of rework because the results obtained are more reproducible. 5. The system is flexible. 6. There is a substantial decrease in maintenance cost. 7. Given all these savings, it is found that the high cost of the nitrogen-methanol atmosphere is offset to the point that the cost of using this atmosphere matches with that of the endogas atmosphere.126 Carburizing Process Variables. Control of three main process variables—temperature, time, and atmosphere composition as carbon potential of the atmosphere—determines the successful operation of the gas carburizing methods. These parameters affect the carbon profile.127 Other variables that influence the amount of carbon transferred/absorbed to the parts are the alloy content of the parts and the extent of atmosphere circulation.119 effect of temperature. Lower temperatures are often used for shallow case depths. The temperature most commonly employed for carburizing is 925°C (1700°F). Sometimes carburizing temperatures are raised up to 980°C (1800°F). Figure 16.28 shows the effect of increasing carburizing temperature from 925 to 1065°C (1700 to 1950°F) for a 3-hr treatment period on case depth and carbon content in AISI 1018 steel. The control of carbon potential for this atmosphere— endothermic gas enriched with natural gas—was done automatically throughout each cycle by the dew point method to produce a surface carbon content of 0.90% and 0.95%.12 High-temperature carburizing is defined as carburizing in the temperature range of 982 to 1093°C (1800 to 2000°F), i.e., above the traditional carburizing temperature of 870 to 925°C (1600 to 1700°F). High-temperature carburizing may be recommended where deeper case depths (i.e., over 0.05 in., or 1.3 mm) are required. This increases the production rate by reducing the cycle time, produces more

CHAPTER SIXTEEN

16.60

Depth of case, 0.001 in. 1.0

0

20

40

80

100 1018

Carburizing Dew temperature, °C point, °C 1065 –22 to –21 1040 –19 to–18 1010 –17 to –16 980 –14 to –13 955 –12 to –11 925 –10 to –9

0.8

Carbon, %

60

0.6

0.4

0.2 3 h at carburizing temperature 0

0.5

1.0

1.5

2.0

2.5

Depth of case, mm FIGURE 16.28 Effect of increasing carburizing temperature from 925 to 1065°C (1700 to 1950°F) for a 3-hr treatment period on case depth and carbon content in 1018 steel. Natural gas-enriched endothermic gas atmosphere was used, and the carbon potential was automatically controlled by the dew point method to produce a surface carbon of 0.90 to 0.95%.12,129 (Reprinted by permission of ASM International, Materials Park, Ohio.)

gradual carbon gradient between case and core (Fig. 16.29),127 and cuts down on equipment, energy consumption, and floor space (i.e., furnace size). Thus it offers economic advantages over carburizing at 925°C (1700°F).107,108 Higher operating temperatures permit higher carbon potential to be employed with less sooting. Moreover, the carbon diffusion reaction is less likely to take place at high carburizing temperatures.128 Higher temperatures are hard on furnace fixtures—parts and fixtures are more prone to distortion because they lose strength as the temperature increases.125 effect of time. Harris12 has developed a formula relating the effect of time and temperature on case depth for normal carburizing of plain carbon and alloy steels which can be given as d = 660e -8287 T t

(16.30)

where d is the case depth in millimeters, T is the temperature in kelvins, and t is the time in hours. For a particular temperature the relationship reduces to d(case depth ) = k t

(16.31)

where k is the diffusion coefficient for a given operating temperature and is a function of process temperature, becomes 0.635, 0.533, and 0.457 for 925, 900, and 870°C, respectively, when the case depth is expressed in mm (in.) and the time is expressed in hr. The case depth/time relationship calculated by Harris is shown in Table 16.10

Case Depth, in. (mm) 0.175 (4.5) 0.150 (3.8) 0.125 (3.2)

1850°F (1010°C)

0.100 (2.5) 0.075 (1.9)

1700°F (925°C)

0.050 (1.3) 0.025 (0.6) 1600°F (870°C) 0 0 2 Time, h

4

6

8

10

12

(a)

Carbon Content, % 1.0

1850°F (1010°C) 0.8 1700°F (925°C) 0.6

0.4 1600°F (870°C) 0.2 Base carbon level

0 0

0.020 (0.5)

0.040 (1.0)

0.060 (1.5)

0.080 (2.0)

Case Depth, in. (mm) (b)

16.61

0.100 (2.5)

FIGURE 16.29 Effect of higher carburizing temperature showing (a) reduced cycle time for a given case depth in AISI 8620 steel and (b) a more gradual carbon gradient between case and core (due to heavier cases) than that obtained by carburizing at lower temperatures.127 (Reprinted by permission of ASM International, Materials Park, Ohio.)

CHAPTER SIXTEEN

16.62

TABLE 16.10 Values of Case Depth versus Time Calculated by Harris12 Case depth† after carburizing at: 870°C (1600°F)

900°C (1650°F)

925°C (1700°F)

Time t, hr

mm

in.

mm

in.

mm

in.

2 4 8 12 16 20 24 30 36

0.64 0.89 1.27 1.55 1.80 2.01 2.18 2.46 2.74

0.025 0.035 0.050 0.061 0.071 0.079 0.086 0.097 0.108

0.76 1.07 1.52 1.85 2.13 2.39 2.62 2.95 3.20

0.030 0.042 0.060 0.073 0.084 0.094 0.103 0.116 0.126

0.89 1.27 1.80 2.21 2.54 2.84 3.10 3.48 3.81

0.035 0.050 0.071 0.087 0.100 0.112 0.122 0.137 0.150

† Case depth, mm = 0.635÷t (case depth, in. = 0.025÷t ) for 925°C (1700°F); 0.533÷t (0.021÷t ) for 900°C (1650°F); 0.457÷t (0.018÷t ) for 870°C (1600°F). For normal carburizing (saturated austenite at the steel surface while at temperature). Reprinted by permisssion of ASM International, Metals Park, Ohio.

FIGURE 16.30 Carbon profiles produced in single-stage carburizing at 925°C in times ranging from 1 to 48 hr.12,129 (Reprinted by permission of Pergamon Press, Plc; after Still and Child.)

for three common carburizing temperatures. Figure 16.30 shows the carbon profiles produced in single-stage carburizing at 925°C in times ranging from 1 to 48 hr.129 This demonstrates that the depth required to obtain a certain carbon level at a certain temperature is directly proportional to the square root of time. When carburizing is deliberately controlled to produce carbon content of the surface somewhat less than saturated austenite, the calculated case depth will be slightly lower than Eq. (16.31) (or Table 16.10) shows.

SURFACE HARDENING TREATMENTS

16.63

In addition to the time at carburizing temperature, several hours may be required to bring large components to operating temperature. When these components are quenched directly from the carburizer, the cycle may be further stretched to permit time for the components to cool from the carburizing temperature to a quenching temperature of about 845°C (1550°F). This period may be treated as a moderate diffusion period; during this period, surface carbon concentration is decreased by maintaining an atmosphere of low carbon potential which is in contact with the component surface. Harris12 has also developed a procedure for calculating the carburizing time and diffusion time to produce a given case depth and surface carbon concentration, which is Carburizing time tc = total time t ¥ or

(C - C0 )2 C s - C0

Diffusion time td = total time t - carburizing time tc

(16.32) (16.33)

where total time t (in hours) is calculated from the equation in Table 16.10, C is the final required surface carbon concentration, C0 is the original (or core) carbon concentration, and Cs is the surface carbon concentration at the end of the carburizing cycle. This method is most suited for batch-type furnaces.12 alloy effects. The various alloying elements found in carburizing steels affect the activity of carbon dissolved in austenite. Cr seems to decrease the activity coefficient, and Si and Ni seem to increase it. Consequently, the Cr-bearing steel parts equilibrated with a certain furnace atmosphere will take on more carbon than pure iron, while Ni-bearing steels will take on less carbon. In reality, carbides form at lower carbon potentials in Cr-bearing steels than in carbon steels. carbon concentration gradient and carbon potential. The carbon concentration gradient of the carburized parts is a function of the carburizing time, temperature, type of cycle (different combinations of carburizing and diffusion times), carbon potential of the carburizing atmosphere, and composition of steel. The carbon potential of a furnace atmosphere at a certain temperature is defined as the percentage of carbon dissolved in iron which is in thermodynamic equilibrium with the furnace atmosphere at that temperature. It is the driving force for the carburizing reaction. The influence of carbon potential of the atmosphere on the carbon concentration gradient, at any given temperature, is shown in Fig. 16.31.12,129 Carbon-potential control during carburizing is achieved by changing the flow rate of the hydrocarbon enrichment gas and maintaining a steady flow of endothermic carrier gas. Close control of carbon potential must be obtained because it affects the case depth directly. Too much enriching gas leads to sooting. To control the carbon potential of hydrocarbon enrichment gas, the concentration of some constituents such as CO2, water vapor content, and O2 present in the carburizing atmosphere must be determined; these four basic methods are described below: 1. Dew point method. Measurement of average dew point at a given temperature is made by determining the contained water, because the amount of water vapor in the atmosphere is directly related to carbon potential based on the reaction C + H2O ∫ H2 + CO

(16.34)

CHAPTER SIXTEEN

16.64

Carbon (%)

0.8

0

20

40

60

1022 steel carburized at 918°C Distance below surface (0.001 in.) 80 0 20 40 60 80 0 20

0.5%C potential

0.6

40

0.75%C potential

60

80

1.1%C potential

0.4 0.2 0

2hr 0.5

4 1.0

8

16

2hr

4

8 16

1.5 2.0 0 0.5 1.0 1.5 2.0 0 Distance below surface (mm)

16 2hr 0.5

4 1.0

8 1.5 2.0

FIGURE 16.31 Carbon concentration gradients in 1022 steel carburized at 920°C (1685°F) with 20% CO, 40% H2 gas containing sufficient H2O to produce the carbon potentials shown, namely, 0.50, 0.75, and 1.10% carbon.12 (Reprinted by permission of ASM International, Materials Park, Ohio.)

The main disadvantage of this method is registry of inaccurate results due to either the condensation or presence of hygroscopic materials in the gas sampling system. The two widely used measurement methods are aluminum oxide capacitor and the chilled mirror; both have the drawback of contamination risks from soot and complex vaporized hydrocarbons usually observed in carburizing atmospheres. Consequently, this automatic dew point measurement technique is not reliable for carburizing applications. 2. Infrared method. CO2 present in the atmosphere is measured by an infrared (IR) gas analyzer, based on the reaction C + CO2 ∫ 2CO. In this method the absorption of infrared radiation of a gas atmosphere sample is measured by using an infrared gas analyzer. This method can detect changes of CO2 concentration measuring as low as 0.001%.130 Since the CO2 level is very low at high carburizing temperatures, an infrared dew point (dual) system is utilized for greater reliability and in diagnosing conditions within the furnace.124 IR analysis provides better uniformity in case depth and helps keep equipment free of soot. IR analysis is especially important when nitrogen-methanol atmospheres are in use,131 because monitoring of CO indicates the N2/CH3OH ratio. 3. Wire method. This method involves the measurement of the electric resistance of a wire of iron-nickel alloy (0.003 in. thick and 2 in. long) surrounded by the furnace atmosphere, which is based on the carbon content in the surface. 4. Oxygen probe method. Unlike the conventional dew point and infrared analyzers which require a sample of the furnace atmosphere, the oxygen probe (i.e., solid electrolyte oxygen cells) can be located in the furnace chamber in a manner similar to that of a thermocouple132,133 for measuring the oxygen content of the atmosphere; the CO/CO2 ratio can be known by using the equilibrium reaction124 CO + 12 O2 = CO2 . The most common problems with oxygen probes are as follows: (1) They are dependent on assuming a constant CO, which might not be true. (2) Formation of carbon soot on the probe distorts the bulk reading and causes the automatic control

SURFACE HARDENING TREATMENTS

16.65

instrument to reduce enriching gas flows, thereby leading to shallow case and low carbon in the parts. (To avoid this situation, an oxygen probe burn-out device is used to periodically oxidize soot away.) (3) There is an unwanted catalytic reaction between probe electrodes and the enriching gas, carrier atmosphere, and hydrocarbon (e.g., leading to the decomposition of CH4 into CO and H2 around it). This leads to the same scenario as those for soot contamination. (4) Reference problems are due to (a) the contamination of the reference air side of the electrode with the furnace atmosphere gas and (b) leakage of the reference air of the probe near the measuring electrode (due to cracks or other flaws in the ceramic materials used in the probe construction—this results in the decrease in millivolt output). (5) Electrode failure occurs over time. (It can sometimes be diagnosed with response time and impedance tests ahead of time, but not with 100% confidence.134,135) At present the best method for analysis includes a combination of separate oxygen probe and infrared system to compute carbon potential independently; however, it is not cost-competitive.135,136 conventional-controlled versus computer-controlled carbon profile. The main purpose of a carburizing cycle is to achieve a certain carbon profile, which, in turn, gives rise to a certain hardness profile after quenching. In conventional gas carburizing practice, the carbon control is accomplished by (1) analyzing one of the furnace atmosphere constituents, (2) determining the carbon potential, and (3) maintaining this carbon potential at one or two selected set points for a predefined time period (Fig. 16.32).137 In contrast, the computer-controlled technique developed by J. Wunning138 utilizes the microprocessor and a specialized computer program to regulate continuously a desired carbon profile of the workpiece at any carburizing time during different phases of the carburizing cycle by continuously measuring the carbon potential of the atmosphere and temperature. In this case the carbon potential is not regulated at a fixed value for a set time period; instead, it is varied by the microprocessors throughout the processing cycle without the soot or carbide formation (Fig. 16.33). Advantages claimed by computer-controlled processes include (1) optimization of the carburizing process, thereby reducing the furnace idle time137 and carburizing cycle time (thereby producing cost savings); (2) excellent reproducibility, better quality, and uniformity of carburizing; and (3) greater flexibility based on the choice of different cycle programs.137,139,140 16.3.1.6 Vacuum Carburizing. Vacuum carburizing was developed and applied in the United States nearly three decades ago. Vacuum carburizing is a hightemperature, nonequilibrium, boost-diffusion type carburizing process where the steel part being treated is austenitized at 900 to 1040°C (1650 to 1900°F) in a rough vacuum or partial pressure of hydrogen; carburized in a partial pressure of hydrocarbon gas, a mixture of hydrocarbon gases, or hydrocarbon/nitrogen mixtures; diffused in a rough vacuum; and then quenched in either oil or gas. Figure 16.34 describes a typical thermal processing cycle as a function of time.141 This shows pulsing rather than a constant flow. Note that some systems have constant pressure and gas flow while others use pulse and evacuation cycles to circulate the gas. Table 16.11 provides a comparison of vacuum, gas, and plasma carburizing processes.142 Both batch and continuous vacuum carburizing equipment have been used throughout the industry.143 The furnace atmosphere usually consists of natural gas, pure methane, propane, acetylene, and/or propylene as an enriched gas, and nitrogen as a diluent and inert gas. After loading, the furnace is evacuated to a partial pressure of 13 to 40 Pa (0.1 to 0.3 torr) of hydrogen in a graphite-lined heating chamber and a partial pressure

16.66

CHAPTER SIXTEEN

FIGURE 16.32 Carbon profile accomplished in conventional controlled gas carburizing for predefined time periods.137 (Courtesy of Wolfson Heat Treatment Center, England.)

of 40 to 67 Pa (0.3 to 0.5 torr) of hydrogen in a ceramic-lined heating chamber.141 During the carburizing step, the furnace pressures are kept in the range of 1.3 to 6.6 kPa (10 to 50 torr) in furnaces of graphite construction and 13 to 25 kPa (100 to 200 torr) in furnaces of ceramic construction. Partial pressure exceeding 40 kPa (300 torr) is usually not recommended due to excessive carbon deposition within the furnace associated with higher partial pressure.141 In this process, only one gas reaction predominates. When methane or propane gas is used, the dissociation reaction at the steel surface is:144 CH 4 + Fe Æ Fe(C) + 2H 2 or C3 H8 + 3Fe Æ 3Fe(C) + 4H 2

(16.35)

After the appropriate carburizing time tc, a diffusion time td follows during which the carburizing gas is evacuated, and diffusion is allowed to occur in rough vacuum of 67 to 135 Pa (0.5 to 1.0 torr) at the same temperature employed for carburizing prior to cooling to the hardening temperature and subsequent quenching in either oil or gas. The relation between these times at 871°C (1600°F) is given by the equation td = (2/3)tc.144 If carbon potential control was employed during the carburizing (boost) step, the diffusion step might be shortened, or eliminated.141

SURFACE HARDENING TREATMENTS

16.67

FIGURE 16.33 Computer-controlled technique for desired carbon profile of the workpiece at any carburizing time.137 (Courtesy of Wolfson Heat Treatment Center, England.)

Unlike gas carburizing, where carbon potential is a function of the gas atmosphere, the carbon potential in a vacuum furnace is determined by the carbon saturation of steel and its carburizing time at a given temperature (Fig. 16.35a). However, the carbon potential of a propane atmosphere is higher than that of methane due to the availability of more carbon during the (above) cracking reactions. At or below 871°C (1600°F), carburizing or carbonitriding containing CH4 atmosphere is not practiced due to its poor cracking efficiency compared to propane.144 Note that large furnace pressure in the carburizing step produces too much carbon soot on the parts, which is not desirable. Hence the recent practice is to pulsate the propane or methane gas in the hot chamber and maintain the vacuum pressure between 20 and 30 torr.145 Generally, graphite/carbon fibers are used as the heating materials and fixtures (which are not susceptible to attack by carbon) instead of metals in furnace construction. The use of a higher-temperature carburizing does not (1) create any undue increase in wear or in maintenance costs,146 (2) cause excessive distortion or warpage, and (3) result in reduction in mechanical properties. The resultant carbon concentration is known by the ratio of carburizing period to diffusion (or boost/

FIGURE 16.34 Plot of temperature and pressure versus time for a typical vacuum carburizing process with a reheat cycle.141 (Courtesy of ASM International, Materials Park, Ohio.)

PROPANE

% CARBON

2.00

METHANE

1.60 1.20 0.80

0

10

20

30 40 50 60 TOTAL TIME (MIN)

Carbon (wt%)

(a)

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

Plasma Vacuum Gas

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Depth (mm) (b)

FIGURE 16.35 (a) Percent carbon versus total time at 871°C (1600°F) after vacuum carburizing of steel part using propane and methane.144 (b) Carbon profile for plasma, vacuum, and gas after 15 min.159 [(a) Reprinted by permission of ASM International, Materials Park, Ohio; (b) reprinted by permission of Fairchild Publications, New York.]

16.68

TABLE 16.11 Comparison of Plasma, Vacuum, and Gas Carburizing Processes142

16.69

Plasma

Low pressure

Gas

Gas mixtures

CH4, H2, Ar

CH4, C2H6

Species for carburizing Pressure range Gas consumption Gas disposal Availability of plant Thermal emission Control/regulation of the process Carburizing depth (end)

Activated by plasma 2–30 mbar 10 bar

Thermodynamical dissociation 10–600 mbar ~1 m3/hr (Burn off)/pumping Always None None Small up to large end’s

N2/methanol, endothermic atm., direct gassing Thermodynamical dissociation 1 bar 5 m3/hr–10 m3/hr Burn oil After forming Yes O2-probe/dew point Middle up to large end’s

Partial carburizing Carburizing speed Surface oxidation Geometry of parts Type of load Quenching media

Courtesy of ASM International, Materials Park, Ohio.

Expensive protection with paste Small case depth quicker than vacuum or gas carb., also diff. laws None Similar to gas carburizing Ordered load High-pressure gas > 10 bar

Expensive, protection with paste f(Cr, 0) diffusion laws Yes Shadow effects, carbon deposition (e.g., blind holes) (Bulk load) ordered load Oil

16.70

CHAPTER SIXTEEN

diffusion) period, whereas the required case depth depends on total carburizing time according to the following reactions. The characteristic difference between endothermic carburizing and vacuum carburizing lies in the fact that in the former the diffusion of carbon occurs intergranularly at the surface while in the latter the diffusion occurs transgranularly at the surface. The advantages of this method over the conventional gas carburizing include: 1. Preheating and post-carburizing treatment may be accomplished under vacuum, which causes exceptionally clean parts (i.e., bright surface quality). 2. There is elimination of (endothermic) atmosphere generator, which also causes elimination of simultaneous reactions involving CO, CO2, H2, and CH4. 3. There is faster carburizing (i.e., reduced cycle times) and higher effective case depths >0.9 to 1.0 mm (0.035 to 0.040 in.), mainly due to boost-diffusion cycle, greater carbon potential, and use of higher carburizing temperatures, typically 980 to 1040°C (1800 to 1900°F). 4. Carburizing gas consumption is lower. The gas consumption is only about 10% of that in a gas carburizing process.142 5. Mechanical properties (such as high fatigue strength, hardenability, etc.) are improved because of no formation of intergranular oxidation products on treated parts.142 6. Heating up and shutting down of equipment are done in a few minutes.145 7. There is an exceptionally uniform gradation of carbon from the surface in; exact, uniform, repeatable, and predictable surface carbon contents, case depths, and metallurgical properties due to high degree of process control variables, possible with vacuum furnaces.142,146,147 8. Use of vacuum media permits the selection of precise levels of carbon availability by varying or constant gas flow, gas pressure, and hydrocarbon concentration in the gaseous mixture.142,147 9. There is the ability to avoid external effects of nonequilibrium gas states, surface abnormalities, and temperature differentials.142,147 Disadvantages are that 1. Equipment cost is high. 2. A delicate balance exists in vacuum carburizing where the process conditions must be tuned to achieve the best compromise among case uniformity, carburizing rate, and risk of sooting.148 Problems of both lack of case depth uniformity and the production of soot in the furnace have been recently reported to be solved by special gas injection techniques and lower gas pressure (usually below 10 mbar).149 Advantages of small addition of acetylene in carburizing atmosphere at a level of medium vacuum below 0.13 kPa (1 torr) are (1) considerable reduction in the sooting problem, making it possible to minimize the maintenance cost; (2) high uniformity of case depth even in the complex-shaped parts on loads with high component density and even on bulk loads without increasing the supply amount of the gas; (3) very small thermal losses due to convection;150 and (4) extremely high productivity due to high load density and the short carburizing times due to high carbon transfer.

SURFACE HARDENING TREATMENTS

16.71

TABLE 16.12 Typical Carburizing Constants and Boost/ Diffusion Ratio to Obtain a 0.8 to 0.9% Surface Carbon Content in a Low-Alloy, Low-Carbon Steel141 Temperature °C 840 870 900 925 950 980 1010 1040

Carburizing constant

°F

k†

k‡

Boost/diffusion ratio r

1550 1600 1650 1700 1750 1800 1850 1900

0.25 0.33 0.41 0.51 0.64 0.76 0.89 1.02

0.010 0.013 0.016 0.020 0.025 0.030 0.035 0.040

0.75 0.65 0.55 0.50 0.45 0.40 0.35 0.30

† To obtain effective case depth (50 HRC hardness) D, in millimeters, when D = k ÷t and t is in hours. ‡ To obtain effective case depth (50 HRC hardness) D, in inches, when D = k÷t and t is in hours. Courtesy of ASM International, Materials Park, Ohio.

Carbon Gradient and Case Depth Prediction. The carburized case depth d in vacuum (or gas) carburizing can be simply and accurately predicted from the following equations: d=k t

(16.36)

tc = rt

(16.37)

t = tc + td

(16.38)

where k is a temperature-dependent carburizing constant, t is the total carburizing time, r is the boost/diffusion ratio, tc is the carburizing time, and td is the diffusion time. Table 16.12 gives the typical carburizing constants and boost/diffusion ratios to obtain a 0.8 to 0.9% surface carbon content in a low-alloy, low-carbon steel.141 Effect of Alloying Elements. Alloying elements affect the rate of carbon absorption; for example, Si and Mn reduce the carbon potential whereas Cr, Mn, and Mo increase the effective carbon absorption and form more stable carbides. Incorrect boost/diffusion ratio may cause the formation of carbide network in these materials. Special higher-temperature cycles must be used on some materials such as stainless steels to depassify the surface before carburizing. The use of high temperature results in dissolution of carbide formers, thereby making carburization effective at high temperatures. Gas composition tends to have a greater effect on case depth and uniformity than does gas partial pressure.151 Additional advantages of vacuum carburizing with high-pressure gas quenching are (1) the provision of clean, dry hardened parts; (2) vacuum as protective atmosphere; (3) reduction in servicing, maintenance, and repair costs; (4) ease of hightemperature carburizing; and (5) control of quench intensity, no Leidenfront phenomenon, uniform quenching, and small scatter of distortions.152 16.3.1.7 Plasma Carburizing. Plasma carburizing is a nonequilibrium, boostdiffuse method of carburizing where a constant carbon potential is not maintained;

16.72

CHAPTER SIXTEEN

rather, the steel is heated to carbon saturation for the selected temperature. The steel components to be treated are introduced [with a spacing of ⬃6.4 mm (0.25 in.) in between] into the hot temperature zone, which constitutes the cathode with respect to the furnace wall (anode), where a dc voltage (in the range to 350 V to 1 kV) is applied in an oxygen-free, low-pressure [in the range of 130 Pa to 3.3 kPa (1 to 25 torr) or, preferably, 130 to 670 Pa (1 to 5 torr)] carburizing gas, i.e., a mixture of hydrocarbon gas such as methane (or propane), hydrogen, and argon (or nitrogen). When the flow rate becomes very low at a few cubic feet per minute and pressure is in the range of 10 Pa to 3 kPa, the glow-discharge plasma is produced between the furnace wall (anode) and the workpieces (cathode), which cracks or dissociates methane into carbon and hydrogen, excites the ionized gas atoms (e.g., carbon) to react with the surface of the components,153 and heats up the component to the processing temperature of 850 to 1040°C (1562 to 1904°F) or usually at about 927°C (1700°F). Thus, all the carbon required is deposited on the surface during this very short carburizing stage. The thickness of the cathode visible glow or luminescence depends on the pressure, gas composition, and temperature. The key factor in determining the necessary operating pressure is the visual observation to ensure that the plasma covers the load and that no hollow cathode effects are obtained. The carburizing gas mixture and the glow discharge supply are shut off, and the workpiece temperature is allowed to fall—typically to 849°C (1560°F); this represents the longer (3 times) carbon diffusion period. (A higher temperature is used to obtain more rapid diffusion rates. This is not favored due to the increased potential for distortion.) Finally, the components are transferred from the hot zone and subsequently oil-quenched to obtain the desired martensitic hard case.154 Best results are achieved with an integrated oil-quenching facility with component transfer under vacuum.155 Note that when workloads are heated by a conventional heating process, as used in vacuum furnaces, it reduces the heating-up time during plasma carburizing.153 The carbon transfer in propane is very high (Fig. 16.36a);156 as a result, the carbon saturation at the workload surface is reached within 10 to 15 min. This phenomenon can be exploited to run a multistage cycle with several boost and diffuse periods, as shown in Fig. 16.36b.157 Parts in a load can be mechanically masked by not allowing the glow discharge into contact with the area not to be carburized, e.g., by stacking or proper fixturing. Alternatively, copper plating of selected areas is effective for masking.148 The increased carburizing rates during plasma carburizing are attributed to the omission of several reaction steps in the dissociation process of hydrocarbon, i.e., (CH4) gas to produce active soluble carbon owing to the ionizing effect of the plasma. The effective carbon potential is governed by the gas mixture and by its flow rate, gas pressure, plasma power, and temperature. All these parameters may be accurately controlled electrically using experimental calibration from an extensive database.146 The advantages of plasma carburizing methods over gas carburizing methods are as follows:153,155–160 1. Like plasma nitriding, shorter (door-to-door) processing times are consistently obtained when compared with traditional gas or vacuum carburizing at similar temperatures (e.g., by 20% over gas carburizing and by 5% over vacuum carburizing). This is attributed to the very rapid carbon mass transfer into the component surface (Fig. 16.35b)159 (Table 16.11).142 2. There is predictable surface hardness and case depth, with improved case uniformity because of accurate automatic control of carburizing mechanism. [Note:

SURFACE HARDENING TREATMENTS

16.73

Surface carbon content (wt.-%)

1.0

0.8 0.6 900°C (1650°F) Propane based

0.4

Pure methane

0.2

0

10

20

30

40

Time (min)

Diffuse

Carburize

Diffuse

Carburize

Diffuse

Sputter

Carburize

Heat-up

Temperature

(a)

Duration (b)

FIGURE 16.36 (a) Effect of type of gas on plasma carburizing rate.156 (b) Schematic of a multistage cycle employed for plasma carburizing in propanebase gases.157 (Courtesy of Wolfson Heat Treating Center, England.)

Required uniformity of carburizing can be obtained, provided a minimum carbon mass flow (of 3 ¥ 10-4 g of carbon/min/cm2 of surface area) into plasma is obtained.148] 3. There is improved blind-hole penetration or “downhole” carburizing up to an L/D (length/diameter) ratio of about 12 with respect to L/D ratio up to 9 for atmosphere carburizing. 4. Gas and energy consumption is lower (i.e., very large gas and energy savings) because it is a low-pressure process with very low gas flow. Moreover, there is an absence of furnace idling time and very short shut-down and start-up times.155

16.74

CHAPTER SIXTEEN

5. Better fatigue and wear resistance and minimal distortion are observed. 6. Periodic maintenance cost is low due to minimum deterioration of furnace elements and the hearth. 7. Parts are bright and clean (free from soot and internal oxidation)155 with reproducible and high-quality metallurgical structure (due to uniform penetration of carbon into the surface and no formation of carbides). 8. Plasma in ion carburizing increases the carbon potential at a much reduced absolute pressure. 9. Elimination of intergranular oxides (due to carburizing and oil-quench hardening in a vacuum) improves fatigue properties in the unground state.148 10. Like vacuum carburizing, high-temperature plasma carburizing reduces cycle times with higher effective case depths. However, plasma carburizing eliminates two major problems associated with vacuum carburizing, i.e., risk of sooting and carburizing rate. 11. No post-treatment machining is required. 12. There is freedom from pollution.153 13. Masking methods are simple for selective carburizing.158 14. There is no need for furnace conditioning.158 15. Applicability to high-alloy steels such as stainless steels is an advantage.158 16. They are integrable into production lines.158 The main disadvantages of plasma carburizing methods are (1) they are not applicable to bulk loads (individual loading of components is required), (2) carburizing of entire surface of workpieces is not feasible (due to cathode contact requirement), and (3) capital costs are higher (counterbalanced by low running costs). Plasma carburizing using integral high-pressure (10- and 20-bar) gas quenching with nitrogen or helium is applied to different core hardening steel parts in the automotive industry. Process parameters have been successfully established for gear parts and for injection nozzles. Other benefits of this integrated system include greatly improved environment-friendliness, no need of cleaning oil residues and disposal of its effluents, and considerable decrease in heat treating costs. The purchase and operation of washing stations to remove oil residues and the cost of disposing of these effluents are no longer necessary. The process cuts heat-treating costs considerably.161 Plasma carburizing is used for applications such as constant-velocity joints, gears, hydraulic valve components, diesel engine components, and clutch components.158 16.3.1.8 Pulsed Plasma Carburizing. The newly developed pulse plasma carburizing leads to a controllable and reproducible carbon mass flow into the workload, where the pulse length is on the order of microseconds. By varying the plasma parameter, very high mass flow can be achieved, which after just a few minutes causes the saturation of the surface with carbon. A linear increase in the mass flow occurs with increasing pressure. Thus, with a high mass flow at the initiation of the process, a high surface carbon content to just below the saturation point can be achieved by selecting suitable parameter combinations, in order to obtain a steep carbon gradient into the material. Then the mass flow is reduced to such a level that the surface carbon content remains just below the saturation point, but simultaneously avoids the carbide formation in the surface zone. By restricting the mass flow density to the required level, no pauses without plasma for diffusion have to be

SURFACE HARDENING TREATMENTS

16.75

made during the treatment, a steep carbon gradient is maintained, and process times are as short as possible.161 16.3.1.9 Fluidized-Bed Carburizing. Over the last two decades, design innovations have led to the use of (direct and indirect) fluidized-bed furnaces as a practical tool for carburizing, carbonitriding, nitriding, and nitrocarburizing processes. Mostly direct fluidized beds have been used where the workload is heated by immersion directly in the fluidized bed of small (80-mesh or 180-mesh) dry aluminum oxide particles, which behaves as a liquid and produces a fluidizing effect. This is achieved by feeding a supporting gas up through the bed of particles, which also provides the protective atmosphere for the heat treatment of the workload immersed in the bed. When external gas burners or electric elements are used as the heat source, the bed provides a faster heat-transfer medium. This is integrated with quench and tempering furnaces. The indirect fluidized-bed furnace comprises a direct fluidized bed which is heated by either internal gas combustion or immersed electrodes where one or a number of radiation walls, typically tubes, are positioned either horizontally or vertically through the bed, separating the workpieces to be treated from the fluidized bed. The normal atmospheres employed in a conventional fluidized-bed carburizing are (1) endogas atmospheres plus hydrocarbon additions, (2) N/methanol system plus hydrocarbon additions, and (3) direct injection of hydrocarbon and air systems. It is apparent that the use of fluidized beds has become more widespread with the understanding of the particular advantages of the technique by industry.162 The advantages of this process include the following: 1. Uniformity of heating as well as high rates of heating and flow cause the utilization of higher treatment temperatures which, in turn, provide rapid carburizing 2. Close control on temperature uniformity, typically ±5°C, coupled with low capital cost and flexibility are ensured 3. A fluid-bed furnace is very tight (with the upward pressure of the gases, it is difficult for air to leak in) 4. This process produces parts with very uniform finish 5. Elimination of internal oxidation 6. Control of microstructure to improve the wear resistance of the carburized case 7. Development of the uniform carbide dispersion in carburized microstructures/ tempered martensitic structure163 8. The rate of atmosphere conditioning and change of atmosphere composition, typically 2 minutes 9. The use of metallic retorts as against ceramic-lined furnaces providing greater atmosphere integrity163 16.3.2 Carbonitriding Carbonitriding is grouped under austenitic thermochemical surface hardening treatments. This is one of the most commonly employed case hardening treatments in which both carbon and nitrogen are absorbed simultaneously into the surface of steel held at an elevated temperature in the austenite phase field by diffusion

CHAPTER SIXTEEN

16.76

TABLE 16.13 Composition and Properties of Sodium Cyanide Mixtures114

Mixture grade designation 96–98† 75‡ 45‡ 30‡

Composition, wt%

Melting point

Specific gravity

NaCN

Na2CO3

NaCl

°C

°F

25°C (75°F)

860°C (1580°F)

97 75 45.3 30.0

2.3 3.5 37.0 40.0

Trace 21.5 17.7 30.0

560 590 570 625

1040 1095 1060 1155

1.50 1.60 1.80 2.09

1.10 1.25 1.40 1.54

† Appearance: white crystalline solid. This grade also contains 0.5% sodium cyanate (NaNCO) and 0.2% sodium hydroxide (NaOH); sodium sulfide (Na2S) content, nil. ‡ Appearance: white granular mixture. Courtesy of ASM International.

process, using a liquid salt bath or gaseous atmosphere, and are then quenched in water, oil, or gas. 16.3.2.1 Cyaniding (or Liquid Carbonitriding). In cyaniding or salt bath carbonitriding, the steel parts are heated above the transformation temperature Ac1, usually at 843°C (1550°F), in a suitable molten bath containing alkali cyanide to serve as active salt mixed with other salts such as sodium chloride (NaCl) and sodium carbonate (Na2CO3) to provide adequate fluidity and to regulate the melting points of salt mixtures (Table 16.13). The cyanide bath decomposes and releases less carbon and more nitrogen (than those from the nonactivated liquid carburizing bath), which absorb into the part surface. On quenching in water, brine, or oil, a hard case develops. In these baths, mostly NaCN is used instead of KCN due to its lower cost and higher efficiency. The cyanide concentration is varied, depending on the specific application. However, a normally active cyanide bath contains 30% NaCN, 40% Na2CO3, and 30% NaCl. The chemical reactions that occur during cyaniding are essentially similar to those of nonactivated salt bath carburizing [Eqs. (16.11) through (16.14)]. The active hardening agents of these baths are CO and N2, which liberate from the decomposition of sodium cyanate. The rate of cyanate decomposition, being a measure of carbonitriding activity of the bath, increases with the higher cyanate concentration and bath temperature, which produce larger case depths with increased carbon concentration. A fresh cyaniding requires aging for about 12 hr in a molten state to furnish a sufficient amount of cyanate for efficient carbonitriding activity. During aging, any carbon scum formed on the surface must be removed for effective performance. To remove scum, it is necessary to lower the cyanide concentration of the bath to a 25 to 30% level by means of addition of inert salts (sodium chloride and sodium carbonate). The rate of decomposition of the bath at the aging temperature [normally 700°C (1290°F)] is low.114 Case depth is thus a function of NaCN concentration but is limited to 0.254 mm (0.01 in.). This process has some drawbacks, as mentioned above for cyanide-type salt bath carburizing. 16.3.2.2 Gaseous Carbonitriding. In this process, a relatively low percentage (2.5 to 8%) of NH3 is introduced in the carburizing, that of atmosphere [e.g.,

SURFACE HARDENING TREATMENTS

16.77

FIGURE 16.37 Comparison of endquench hardenability curves for 1020 steel carbonitrided at 900°C (1650°F) and carburized at 925°C (1700°F). Hardness was measured along the surface of the asquenched hardenability specimen. Ammonia and methane contents of the inlet carbonitriding atmosphere were 5%, and the balance was carrier gas.164 (After G. W. Powell, M. B. Bever, and C. F. Floe, Trans. ASM, vol. 46, 1954, pp. 1359–1371.)

carbon-rich gas (or vaporized liquid hydrocarbon) mixture] to diffuse nitrogen into the steel together with carbon. This operation is carried out at a lower temperature and for a shorter time than that of gas carburizing, which results in a relatively shallower case depth, usually from 0.075 to 0.75 mm (0.003 to 0.03 in.). This process may be treated as a modified gas carburizing process, rather than a form of nitriding. The main advantages of carbonitriding over carburizing are as follows:113 1. The carbonitrided case has a greater hardenability than a carburizing case (Fig. 16.37).164 (Nitrogen increases the hardenability of steel, stabilizes the austenite, and induces retained austenite, especially in alloy steels.) This favors attainment of a hard case and allows the case of the part to be hardened by oil-quenching or even gas-quenching at reduced expense.113,164 2. Since absorption of nitrogen by the surface layers of steel decreases the critical cooling rate to form martensite, compared to that in carburizing treatment, less distortion will occur. 3. The resistance of carbonitrided layers to softening during tempering is higher than that of the carburized layer, which is attributed to the precipitation of g ¢ nitrides. This resistance also increases with an increase in nitrogen content at tempering temperatures of 573 K and above. This resistance is further enhanced by shot-peening, the effect of which is more profound after carbonitriding than after carburizing, as observed in Fe-0.22C-0.12Si-0.78Mn-0.01P-0.018S-1.1Cr steel automatic transmission gear.165 4. This gives increased wear resistance as well as a hard and uniform case, when compared to ordinary carburizing. 5. Better fatigue properties than those of equivalent carburized components are obtained, owing to different residual stress pattern in the case. 6. For many applications, carbonitriding of the less expensive steels will furnish properties similar to those obtained in gas carburizing of alloy steels.164 7. Sometimes carburizing and carbonitriding are combined to achieve deeper case depths and superior service performance than are feasible by carbonitriding alone.164 Disadvantages of carbonitriding are as follows:

16.78

CHAPTER SIXTEEN

1. Compared to carburizing treatment, more time is required to produce a case depth greater than 0.635 mm (0.025 in.) because of lower carbonitriding temperature. 2. Carbonitriding is restricted mostly to case depths up to about 0.75 mm (0.03 in.) or less, whereas no such restriction is applicable to carburizing. Case Composition. The composition of a carbonitrided case is governed by the time, temperature, atmospheric composition, and type of steel subjected to this treatment. Carbon addition is favored at high temperature. More nitrogen absorption is favored at the lower temperatures, which results in a “compound layer” consisting of Fe-C-N compounds at the surface. This type of case structure is preferred in some wear-resistant applications. For this layer of compounds, a considerable large percentage of NH3 in the gas mixture is required, and control of furnace atmosphere becomes more critical. Figure 16.38 is the microstructure of carbonitrided and oil-quenched AISI 1020 steel showing an outer white layer of case (left) as cementite followed by retained austenite mixed with martensite and interior martensite matrix (right).120,164 Recently, it was reported that the 1 to 2% NH3 addition at the end of the carburizing cycle is most effective in minimizing the amount of surface nonmartensitic transformation products.166 Case Depth. As in the case of carburized parts, carbonitrided parts are usually measured for total case depth or effective case depth.12 Suitable case depth is governed by service applications and by core hardness. Figure 16.39 shows the effects of time and temperature on effective case depths which are based on a survey of industrial practice.105,164 Medium-carbon steels with core hardness of 40 to 45 Rc usually require less case depth than steels with core hardness of 20 Rc or below.

FIGURE 16.38 Microstructure of a 1020 steel carbonitrided (with high carbon potential) and oil-quenched showing an outer white layer of case (left) as cementite followed by retained austenite mixed with martensite and interior martensite matrix (right). Nital etch.120 (Reprinted by permission of ASM International, Materials Park, Ohio.)

SURFACE HARDENING TREATMENTS

16.79

FIGURE 16.39 Effects of time and temperature on case depths based on the results of a survey of industrial practice.164 (Reprinted by permission of ASM International, Materials Park, Ohio.)

Medium-carbon low-alloy steels, i.e., those employed in automotive transmission gears, are often provided the minimum case depth of 0.2 mm (0.008 in.).164 Furnaces. Almost any furnace suitable for carburizing treatment can be used for carbonitriding. The furnace must be equipped with a fan to circulate the atmosphere and with protective atmosphere vestibules to the quenching area. The atmospheres generally consist of a mixture of carrier gas, enriching gas, and NH3. Basically, any atmosphere composition used for gas carburizing can be used for carbonitriding with an additional 2 to 12% anhydrous NH3 of 99.9+% purity.12,164 Temperature Selection. Choice of carbonitriding temperature is based on steel composition, dimensional control, fatigue and wear properties, microstructural constituents, hardness, cost, and equipment. Lower temperature near 704°C (1300°F) causes an explosion hazard as well as superficial, high-nitrogen, brittle cases with low core hardness which are not suitable for most applications. For this reason, most carbonitriding operations are carried out at 790°C (1450°F) or above. Carbonitriding at 900°C (1650°F) appears to be advantageous from the standpoint of favorable combination of hardness, microstructure, wear, and economy. However, other factors, such as distortion and quench cracking, must also be accounted for.167 The usual compromise is made at about 843°C (1550°F) to achieve both ends. Void Formation. Case structure may contain subsurface voids or porosity, if the processing conditions are inappropriately adjusted. This problem is often related to excessive NH3 additions. Table 16.14 summarizes the effect of material and processing variables on the possibility of void formation.164 Control of Retained Austenite. Minimum retained g in the carbonitrided case can be formed by (1) increasing the furnace temperature, (2) reducing the NH3 flow, (3) maintaining the carbon potential and surface carbon concentration of 0.70 to 0.85%, and (4) restricting the NH3 content to about 5%. This can also be drastically reduced by cooling the quenched parts to -40 to -100°C (-40 to -150°F). Since the amount of retained g is usually maximum near the steel surface, it can also be removed from symmetrical parts by grinding. However, these latter methods are expensive.12 Quenching Media. Selection of quenching media such as water, oil, or gas is based on steel composition, size, and shape of the part; allowable distortion; desired (case and core) hardness level; and type of furnace equipment employed. Tempering. Carbonitrided parts developed primarily for wear resistance such as dowel pins, brackets, and washers do not need tempering. Low-carbon steel parts are usually tempered at 135 to 175°C (275 to 350°F) to stabilize austenite and reduce dimensional variations. Most carbonitrided gears are tempered at 190 to 205°C (375

CHAPTER SIXTEEN

16.80

TABLE 16.14 Effect of Material and Variables on the Possibility of Void Formation in Carbonitrided Cases164 Material and processing variables† Temperature increase Longer cycles Higher case nitrogen levels Higher case carbon levels Aluminum-killed steel Severe prior cold working of material Increase in alloy content of steel NH3 addition during heat-up cycle

Possibility of void formation Increased Increased Increased Increased Increased Increased Decreased Increased

†

All other variables remained constant. Reprinted by permission of ASM International, Materials Park, Ohio.

to 400°F) to reduce surface brittleness while maintaining a minimum case hardness of 58 Rc. Alloy steel parts requiring surface grinding are tempered to reduce grinding cracks. Tapping screws made of AISI 1020 steel are tempered at 260 to 425°C (500 to 795°F) to minimize breakage in tapping holes in sheet metal. The parts requiring repeated shock loading are frequently tempered at 425°C (795°F) to improve impact (i.e., notch toughness) and fatigue strength.12 Applications. Gas carburizing is widely used on steels such as 1000, 1100, 1200, 1300, 1500, 4000, 4100, 4600, 5100, 6100, 8600, and 8700 series with up to 0.25% carbon content. Also, in many cases, these steel series with a medium carbon (0.30 to 0.50%) content (such as 4140, 4340, 5130, 5140, and 8640) are carbonitrided at 845°C (1550°F) to case depths up to ⬃0.3 mm (0.010 in.) to obtain a combination of a hard, more wear-resistant surface, and a reasonably tough, through-hardened core (e.g., shafts and transmission gears, heavy-duty gears).164 Carbonitriding of P/M Parts. This process is extensively used in treating ferrous P/M parts with (or without) copper infiltration and with sintered densities of 6.5 g/cm3 (0.23 lb/in.3) minimum. Carbonitriding is accomplished at 790 to 815°C (1450 to 1500°F), which overcomes several problems associated with carburizing of P/M parts made from electrolytic iron powders.164 The carbonitrided parts are usually tempered at temperatures slightly higher than the temperatures employed for carbonitrided wrought steel parts.164 16.3.2.3 Plasma Carbonitriding. As in plasma carburizing, plasma carbonitriding has been successfully used to produce carbonitrided cases. These cases possess greater hardenability and resistance to tempering and are, therefore, preferred to carburized cases for smaller, less massive parts. In plasma carbonitriding, both carbon and nitrogen are added to the steel surface by using methane/nitrogen or methane/nitrogen/hydrogen glow discharge plasma.168 16.3.2.4 Fluidized-Bed Carbonitriding. The carbonitriding in fluidized beds was studied by Jesinski et al.169 on AISI 1022 steel with bed material of quartz sand with 0.2- to 0.3-mm particle size and fluidizing gaseous mixture of propane-air-NH3 or a mixture of N2-H2-CO-NH3 type ammonia-base atmosphere170 in a suitable proportion to produce an endothermic-type atmosphere in the temperature range of

SURFACE HARDENING TREATMENTS

16.81

820 to 940°C (1508 to 1724°F) for 0.5 to 5 hr. Their results were comparable to those for gaseous carbonitrided parts. Their findings clearly demonstrated the attainment of higher-quality hard cases and increased durability of the (fluidized-bed carbonitrided) parts.169

16.4 FERRITIC THERMOCHEMICAL SURFACE HARDENING TREATMENTS The carburizing and carbonitriding treatments discussed in the previous section are austenitic thermochemical treatments, because they involve the addition of alloying elements into the austenite phase and rely on the subsequent transformation of austenite into martensite to produce a high surface hardness. Ferritic thermochemical treatments, on the other hand, involve the diffusional addition of nonmetallic elements into the surface of ferrous parts at temperatures below the eutectoid temperature. Subsequently, the parts are quenched or cooled in the processing medium.171 In this section, nitriding and nitrocarburizing are considered in detail.

16.4.1 Nitriding The nitriding process was first used as a commercial heat treatment process during the late 1920s, and since then it has continuously grown to worldwide application. The nitriding process is the result of interactions of the substitutional alloying elements in iron with nitrogen in interstitial solid solution.172 That is, it involves the diffusion of atomic nitrogen into the surface of steel in the ferritic phase by holding the material at temperatures below 590°C [the eutectoid temperature of Fe-N system (Fig. 16.40)173] and usually between 500°C and 590°C, and consequently no phase transformation takes place on cooling to room temperature. The thin nitrided layer (usually 25 mm thick) so developed is usually subdivided into a compound or white layer near the surface and a diffusion zone beneath the compound layer, which are composed of nitrided phases e-Fe2–3N and g ¢-Fe4N, respectively. On prolonged nitriding, porosity can be observed in iron nitrides due to their metastability with respect to the molecular nitrogen gas and pure iron at the usually applied temperature (723 to 863 K) and total pressure (⬃1 atm).174 The extremely hard, wear-resistant iron-alloy nitrogen compounds which are formed on nitriding introduce beneficial compressive stresses at the surface and are resistant to some kinds of corrosion and softening when heated; therefore, quenching or any other treatment is no longer required. Nitriding allows the reduction or even elimination of the need for a subsequent expensive machining operation. The components must be given proper heat treatment prior to nitriding to develop the right kind of structure of the core material and, hence, to obtain better response to nitriding. Since the nitriding process is the final heat treatment and produces growth of material, proper allowance for growth should be taken into account in the final shape. Sharp corners must be avoided; otherwise, growth will occur at the corners, producing projections which are brittle and tend to chip off. The growth is dependent upon temperature and extent of nitriding. The great hardening produced on nitriding is due to a combination of (1) small (50- to 150-Å), closely spaced particles, (2) a very high dislocation density (>1010/cm2), and (3) a high concentration of N associated with dislocation and with Fe-AlN interfaces.

16.82

FIGURE 16.40 Ohio.)

CHAPTER SIXTEEN

Fe-N phase diagram.173 (Courtesy of ASM International, Materials Park,

SURFACE HARDENING TREATMENTS

16.83

16.4.1.1 Nitridable Steels. In principle, all steels can be nitrided. However, since the iron nitrides are unstable, the majority of steels which are nitrided are alloy steels containing a small amount of Al (from 0.85 to 1.5%), Cr, Mo, and V, which will produce stable and hard nitride needles in the outer skin of the steel at the nitriding temperature; these are commonly known in the United States as Nitralloy steels. They may also contain other alloying elements such as TI, Nb, W, Mn, or other reasonably strong nitride-forming elements in order to obtain high hardness. Unalloyed carbon steels are not desired for gas nitriding due to the formation of extremely brittle case that spalls easily and the small increment of hardness in the diffusion zone. Alloy steels used for nitriding can be classified into the following groups:175 1. Aluminum-containing low-alloy (Nitralloy) steels (e.g., G, 135M, N, and EZ types), which are generally employed where very high surface hardness and excellent wear resistance are essential (but they provide lower ductility). 2. Medium-carbon, chromium-containing low-alloy steels such as AISI 4100, 4300, 5100, 6100, 8600, 8700, 9300, and 9800 series. These steels produce considerably greater ductility as well as good antigalling and wear resistance properties, but with lower hardness. 3. Hot-work die steels having 5% Cr, for example, H11, H12, and H13. These steels, such as H11 and D2, render substantially high case hardness with exceptionally high core strength. 4. Low-carbon, chromium-containing low-alloy steels such as 3300, 8600, and 9300 series. 5. Air-hardening tool steels such as A2, A6, D2, D3, and S7. 6. High-speed steels such as M2 and M4. 7. Nitronic stainless steels such as 30, 40, 50, and 60. 8. Ferritic and martensitic stainless steels such as AISI 400 series. 9. Austenitic stainless steels such as AISI 200 and 300 series. 10. Precipitation-hardening stainless steels, such as 13-8 PH, 15-5 PH, 17-4 PH, 177 PH, A-286, AM 350, and AM 355. Prior to nitriding, these alloy steel parts are usually austenitized, quenched, and tempered at a high temperature, usually at or above 575°C, to guarantee a structural stability at the nitriding temperature. The rate of nitriding of the alloy depends on the strength of the interaction of alloying elements with nitrogen, the nitrogen potential of the gas mixture, the composition of the alloying elements, the ease with which precipitates can nucleate and grow, and the nitriding temperature.172 Figure 16.41 shows a comparison of nitriding characteristics of two conventional nitriding steels and the newly developed nitriding Imanite steel.176 The advantages of nitriding over other surface hardening methods are as follows:177,178 1. Because low treatment temperature is involved and no quenching is required, distortion can be kept to a minimum, even though some dimensional growth does occur. 2. Increased high surface hardness, higher wear and fatigue resistance, improved corrosion resistance (except for stainless steels) and antigalling properties, good

16.84

CHAPTER SIXTEEN

FIGURE 16.41 Comparison of the nitriding characteristics of steels BS 722M24 (3% Cr), BS 709M40 (SAE 4140), and Imanite treated under the same conditions.176 (Courtesy of Wolfson Heat Treatment Centre, England.)

resistance to softening during tempering,175 and high-temperature hardness of components are obtained. 3. It can be employed safely at a reasonably high temperature, e.g., up to 650°C (1200°F) for short periods and up to 538°C (1000°F) for long periods. 4. Surface contamination encountered in ordinary heat treatment is avoided. Hence components can be machined to final size and hardened by nitriding without any further operation. 5. Selected areas and irregular shapes can be nitrided. Among the disadvantages of the nitriding process are the following: 1. The process is slow (i.e., requires very long process time, typically 24 to 72 hr). 2. This process requires the use of special steels containing Al and/or Cr. 3. The formation of a brittle and thin white layer (sometimes referred to as a compound zone) consists of a mixture of iron nitride phases on treated parts, which can be detrimental on bearing surfaces because it tends to spall (and crack) in service. Removal of this layer by expensive and time-consuming mechanical grinding or shot blasting/peening with fine glass beads at suitable pressure or chemical treatment is necessary179 before the parts can be put into service. It has been found that complete removal of the white layer is possible by a prolonged hot soaking in a cyanide solution (consisting of 1 lb of NaCN in 1 gal of water). The solution is heated to between 70 and 90°C, and components are immersed in this solution for short periods (which causes the white layer to be friable), followed by blast cleaning (220-mesh grits and 80 psi of pressure). 4. The presence of NH3 fumes (for gas nitriding) requires the provision of an adequate ventilation system. 5. Allowances must be made for a small amount of growth resulting from an increase in volume at the surface. 6. There is also the danger of inducing temper brittleness; hence steels containing Ni or Mo are considered best, because these elements resist the onset of embrit-

SURFACE HARDENING TREATMENTS

16.85

tlement. Parts must be so designed as to avoid sharp corners (by chamfering), which may spall, and dissymmetry, which may cause distortion. 16.4.1.2 Procedure. Nitriding can be carried out on steel by using a gaseous or liquid nitrogenous medium. The former is called gas nitriding, and the latter is called salt bath (or liquid) nitriding. Gas nitriding is the more widely used process. Other developments in gas nitriding, such as Floe process, pressure nitriding, bright nitriding, and Nitreg nitriding as well as the more recent ones such as plasma nitriding and fluidized-bed nitriding, are also available. 16.4.1.3 Salt Bath or Liquid Nitriding. A liquid nitriding bath is a salt bath such as fused cyanide-cyanate salts which contains mostly alkali cyanide (60 to 70 wt% NaCN and 30 to 40 wt% KCN)—an active ingredient. The process involves melting of salt bath, aging of the molten salt bath, and immersion of the steel parts in the temperature range of 560 to 570°C (1040 to 1060°F) for 1 to 2 hr. During melting of dry salt, the retort is covered or the equipment is completely hooded and vented to guard against explosion or sputtering of the salt. The molten salts should be aged by being held at 565 to 595°C (1050 to 1100°F) for at least 12 hr. Aging produces a decrease of cyanide content of the bath and the formation of a small amount of carbonates (Na2CO3) and cyanates (NaCNO).When a level of 5% NaCNO is reached after aging, the bath can be safely used.The NaCN contents for high-speed steels and hot-working tool steels and low-carbon and alloy steels are 15% min. and 20% min., respectively. The alloy steel parts for nitriding must be either in the quenched and tempered or in the stress-relieved condition to develop the required core properties.12 During nitriding, the NaCN is partially oxidized to form cyanate according to the equation 4NaCN + 2O2 Æ 4NaCNO

(16.39)

Sodium cyanate is an unstable compound and is readily decomposed to liberate nascent nitrogen, which is transferred to the surface of the steel part in the following manner: 4NaCNO Æ Na2CO3 + 2NaCN + CO + 2N

(16.40)

The evolution of CO in Eq. (16.40) may produce iron carbide by the following reactions: 2CO Æ CO2 + C

(16.41a)

3Fe + C Æ Fe3C

(16.41b)

Following nitriding, the parts should be quenched in water, polymer solution, oil, soluble oil solution, or air, depending on the steel compositions. For satisfactory performance, all parts should be thoroughly cleaned and free of surface oxide. They are preheated prior to immersion in the bath to drive off surface moisture. The bath should be analyzed once or twice per week, and necessary additions should be made to maintain compositions within closer limits in order to achieve consistent nitriding rates. All the contaminants and oxidation products must be removed from the bath. Overheating above 600°C should be avoided. Salts should be completely changed every 3 to 4 months of operation. Usually a titaniumlined or -plated furnace container is recommended for the best result. It is a general practice to cover the bath when not in use. The normal operating temperature for salt bath nitriding is 550 to 750°C (1022 to 1382°F).

16.86

CHAPTER SIXTEEN

The amount of cyanate may be controlled analytically, and its addition can be kept to the desired level in the bath. The optimum limit will vary, depending on the final surface hardness required. Advantages of salt bath nitriding are as follows:180,181 1. Salt baths provide a more uniform nitrogen potential and reduced distortion by supporting the components in the bath. 2. Bath composition varies from the high-speed types containing 90 to 95% cyanide (used in the steel-hardening operation) to those with 20 to 30% cyanide salts for general applications.180 3. This treatment might prove cheaper for mass production of components. 4. Post-nitriding treatment improves corrosion resistance and decreases friction coefficient. Steam treatment following nitriding further extends the life of the cutting tools. 5. It involves rapid heating and processing. 6. Good and reproducible nitrided layers can be obtained with ease on low-carbon and low-alloy steels. Disadvantages of salt bath nitriding include (1) toxicity of salt bath, (2) waste disposal problem, (3) need for thorough washing to remove salt residues to avoid corrosion, (4) lack of in-process control, and (5) limitation to those steels which can be heated to higher temperatures without sacrificing core hardness.182 Applications. Salt bath nitriding is applied to a wide variety of carbon and lowalloy steels (e.g., crankshafts, camshafts, cylinder liners, tappet guides, light-duty gears, connecting rods, and clutch plates), tool steels, high-speed tool steels (e.g., end mill drills, reamers, side and face cutters, and form tools), hot-working tool steels (e.g., press forging dies, extrusion dies, and mandrels made from H13 steel), stainless steels, and cast irons.183 This method offers superior properties to those of other case hardening methods, particularly in the production of automotive parts such as thrust washers (from AISI 1010 steel), shaft, seat bracket (from AISI 1020 steel), and rocker arm shaft (from SAE 1010 steel). However, it is not suitable for many applications requiring deep cases and hardened cores.181 Other Salt Bath Compositions. Another typical nitriding salt bath for tool-steel applications has the following composition:12 30% (max.) NaCN, 25% (max.) Na2CO3 or K2CO3, 4% (max.); other active ingredients, 2% (max.) moisture, and balance KCl. A proprietary nitriding salt bath has the following composition: 60 to 61% NaCN, 15 to 15.5% K2CO3, and 23 to 24% KCl.12 Special Liquid Nitriding Processes. Although there are several commercial proprietary liquid nitriding processes, Table 16.15 represents the basic types.181 In these special liquid nitriding salt bath processes, proprietary additions, either gaseous or solid, are made to serve several roles such as accelerating the chemical activity of the bath, expanding its applicability to a wide variety of steels that can be processed, and improving the properties obtained after nitriding.12 Cyanide-free liquid nitriding salt compositions are also available given their wide acceptance, which contain a small amount of cyanides, usually up to 5% in the active bath. The following are the two main processes: 1. Liquid pressure nitriding. In this proprietary process, anhydrous NH3 is introduced through the bottom of the sealed retort into a cyanide-cyanate bath and maintained at a pressure of 7 to 205 kPa (1 to 30 psi) to speed up the nitriding reaction of the bath. The percentage of nascent nitrogen in the bath is controlled by

TABLE 16.15 Liquid Nitriding Processes181

Process identification

Suggested posttreatment

Operating temperature

16.87

°C

°F

U.S. patent number

Water or oil quench; nitrogen cool

570

1060

3,208,885

Water or oil quench

510–650

950–1200

Sodium cyanide (NaCN), sodium cyanate Strongly reducing (NaCNO)

Air cool

525–565

975–1050

Type A: Potassium cyanate (KCNO), potassium carbonate (K2CO3) Type B: Potassium cyanate (KCNO), potassium carbonate (K2CO3), 1–10 ppm, sulfur (S)

Water, oil, or salt quench Water, oil quench, or salt quench

580

1075

4,019,928

540–575

1000–1070

4,006,643

Operating range composition

Chemical nature

Aerated cyanide-cyanate

Sodium cyanide (NaCN), potassium cyanide (KCN) and potassium cyanate (KCNO), sodium cyanate (NaCNO)

Strongly reducing

Casing salt

Potassium cyanide (KCN) or sodium cyanide (NaCN), sodium cyanate (NaCNO) or potassium cyanate (KCNO), or mixtures

Strongly reducing

Pressure nitriding Regenerated cyanate-carbonate

Courtesy of ASM International, Materials Park, Ohio.

Mildly oxidizing Mildly oxidizing

CHAPTER SIXTEEN

16.88

maintaining the NH3 flow rate at 0.6 to 1 m3/hr (20 to 40 ft3/hr), which causes the NH3 dissociation of 15 to 30%.181 The selection of appropriate pressure is a function of the retort volume, part geometry, surface area to be treated, and process temperature. Maintenance of this bath at an operating temperature of 525 to 565°C (975 to 1050°F) is greatly simplified, because it does not require aging and may be placed into immediate operation using the recommended cyanide/cyanate ratio of 30 to 35% cyanide and 15 to 20% cyanate.181 2. Aerated bath nitriding. In this process, a measured amount of air is forced through the molten bath to provide agitation and stimulate chemical activity, thereby increasing the rate of nitriding. The bath contains 50 to 60% NaCN, 32 to 38% cyanate, 10 to 30% (usually ⬃18%) elemental potassium content as cyanide and/or cyanate, and the remainder sodium carbonate. Note that aerated bath nitriding is suited to the plain-carbon (nonalloyed) steels whereas conventional bath nitriding is well suited to only Cr-, Ti-, and Al-alloyed steels.181 In aerated low-cyanide nitriding, the base is provided with a cyanide-free mixture of KCNO and a combination of Na2CO3 and K2CO3, or NaCl and KCl. In the case of nitriding with heavy loading, there is likelihood of the formation of small percentages of cyanide in the bath. This problem can be overcome by quenching in an oxidizing quenching salt bath that destroys the cyanide and cyanate compounds, thereby solving the pollution problem and producing less distortion compared to water quenching. However, in another proprietary cyanide-free salt bath mixture, a very small amount of sulfur (1 to 10 ppm) and lithium carbonate are added to the base salt to keep the cyanide formation below 1.0%.181 In aerated cyanide-cyanate nitriding, a high-cyanide, high-cyanate fused salt bath containing 45 to 50% cyanide calculated as KCN and 42 to 50% cyanate calculated as KCNO is used. It is applied to treat carbon and low-alloy steels and stainless steels. This salt composition is also referred to as the Tufftride, according to AMS 2755C-1985.181 16.4.1.4 Gas Nitriding. In the conventional gas nitriding process, first the quenched and tempered parts should be thoroughly cleaned, vapor-degreased, and conditioned (by abrasive cleaning with aluminum oxide grit or other abrasives such as garnet or silicon carbide, or by applying a light phosphate coating) immediately prior to nitriding. After loading and sealing the furnace at the start of the nitriding cycle, purging with nitrogen or anhydrous NH3 is used to expel air from the furnace before the furnace is heated above 150°C (300°F). This prevents oxidation of the parts and the furnace components and, if NH3 is used as a purging atmosphere, avoids the formation of potentially explosive mixture.12,175 Subsequently, moisturefree anhydrous NH3 is allowed to flow into the nitriding furnace over the parts in such a way that all surfaces come in contact with the gas, usually at uniform temperature of 525°C (975°F). However, gas nitriding can be accomplished from 500 to 600°C (930 to 1100°F). The NH3 gas, when in contact with the hot steel surfaces, dissociates to produce atomic nitrogen, which reacts with the alloying elements in the steel surfaces to produce nitrides according to the equation NH3

∫ 3 H 2 + N (dissolved in Fe) 2

(16.42)

Nitriding is accomplished in an electric furnace with a device for precise temperature control. Most heat treaters employ batch-type furnaces with essential features. They must be equipped to (1) seal the components from air, (2) provide uniform temperature, (3) maintain and circulate the controlled atmosphere

SURFACE HARDENING TREATMENTS

16.89

throughout the mass of the parts to be nitrided, and (4) brush the nitriding container (i.e., inside of the retort) at intervals in order to clean and remove any deposits such as sulfides which may inhibit the nitriding treatment. Among several types of equipment which can be employed are the vertical retort furnace and movable bell-and-box type furnaces.180 Retorts, fixtures, and furnace accessories which are in contact with NH3 atmosphere are normally constructed of heat-resisting 25Cr-20Ni steels, although nickel, Inconel, Incoloy, and similar alloys are ideal. They frequently require cleaning and care in order to prevent any reaction with the atmosphere. The composition of the exit gas is maintained and measured regularly with the help of a dissociation or absorption pipette; for the first 4 to 10 hr, the dissociation rate of gaseous NH3 is kept at 15 to 30%, depending on the duration of the total cycle. Usually, a constant degree of dissociation, which is used to control the nitriding process, is maintained in a single-stage technique. However, in two-stage process (Floe process), the temperature and degree of dissociation are varied.184 On completion of nitriding, the box is taken out of the furnace without interrupting the gas flow. When the charge has cooled to 150°C (300°F), the gas supply is stopped and the gases remaining in the box are expelled with ease with compressed air prior to opening of the box. The nitriding parts now generally exhibit the characteristic matte gray color. Sometimes the color has the shades of blue, yellow, or purple, which are derived from the presence of O2 in the system, originating from some leak in the box or in the supply tubing or from incomplete drying of the gas. Treatment times are normally between 40 and 60 hr, but for relatively deeper cases up to 90 hr is used.185 Note that emergency purging is also required when the supply of NH3 is cut off or a break occurs in the supply line during the nitriding or cooling cycle.11 Advantages of conventional nitriding are (1) simple control methods; (2) low temperature with respect to carburizing; (3) increased surface hardness, wear resistance, fatigue resistance, and corrosion resistance; (4) low distortion; (5) hot hardness; and (6) controlled growth. Disadvantages include (1) inadequate NH3 dissociation rate for control of layer properties, (2) possible requirement for removal of brittle white layer, in many instances, (3) a need for copper plating or painting with protective paste for masking, (4) requirement of special activation techniques for stainless steels, (5) requirement of a nitride former in nitridable alloys, and (6) potentially long process time. Developments in Gas Nitriding. Gas nitriding may be carried out using either a single-stage or double-stage process. In the single stage, which has just been described above, a temperature range of about 500 to 525°C (930 to 975°F) with dissociation rate of 15 to 30% (i.e., an atmosphere containing 70 to 85% NH3) is used. This process yields a brittle, nitrogen-rich layer, called the white nitride layer, at the surface of the nitrided case. The double-stage process, also called the Floe process, has the benefit of reduced thickness of the white nitrided layer. The first stage of the double-stage process is a duplication of the single-stage process, except for time. The second stage uses a temperature of 550 to 565°C (1025 to 1050°F) and a dissociation rate of 65 to 85% (preferably, 75 to 80%), which has been found to minimize or eliminate the white layer of iron nitride (produced in the single stage) and increase the surface hardness. Figure 16.42 illustrates the microstructures observed when AISI 4140 steel is given single- as well as double-stage nitriding treatment. The microstructure in the single-stage process consists of an outer white surface layer of Fe2N followed by iron nitride and a matrix of tempered martensite (Fig. 16.42a), whereas the microstructure in the double-stage process consists of only a diffused nitride layer and tempered martensite (Fig. 16.42b).120

CHAPTER SIXTEEN

16.90

(a)

(b)

FIGURE 16.42 Microstructure of 4140 steel (a) single-stage nitrided at 525°C (975°F) for 24 hr with 20 to 30% dissociation showing 0.005- to 0.0075-mm (0.0002- to 0.0003-in.) white surface layer of Fe2–3N followed by iron nitride and tempered martensite and (b) double-stage nitrided—first-stage nitriding at 525°C (975°F) for 5 hr with 20 to 30% dissociation, followed by second-stage at 565°C (1050°F) for 20 hr with 75 to 80% dissociation. The microstructure shows absence of white layer, diffused nitride layer, and tempered martensite. Note: The prenitriding condition involves austenitizing at 845°C (1550°F) followed by oil-quenching, tempering at 620°C (1150°F) for 2 hr, and surface activation with manganese phosphate. 2% nital etch.120 (Reprinted by permission of ASM International, Materials Park, Ohio.)

The total case depth, which depends on the nitriding time and temperature, can be extended to 0.5 mm (0.019 in.) below the surface. The total case depth constitutes several zones. In the light microscope, the outermost white or compound zone, when etched in 2% nital, appears white. Below this zone, a nitrided zone, called the diffusion zone, appears in which nitrogen is assumed to combine with substitutional solutes in the ferrite to form a very fine dispersion of alloy nitrides at the prior grain boundaries perpendicular to the direction of nitriding. Beyond the nitrided (or diffusion) zone, there is a region which is revealed by the black appearance in the microstructure when etched by Oberhoffer’s reagent. The extent of hardening within the zone is small.186 Pressure Nitriding. This process differs from the conventional gas nitriding in that it demands the use of a sealed retort which will withstand higher pressures than atmospheric. The surfaces to be nitrided are first cleaned and then introduced into the carbon steel retort, which has been evacuated of air and filled with NH3 to a predetermined pressure. The selection of pressure is a function of the total surface area of the parts to be nitrided and the volume of the retort. The retort is then heated in any furnace where temperature can be controlled for the desired time cycle, after which the retort can be air-cooled, vented, and opened.187 The advantages claimed for this nitriding process are (1) the controlled thickness of white layer, (2) production of high surface hardness together with considerable toughness, (3) rapid formation of case during the first few hours, and (4) convenience in nitriding complex-shaped parts that are difficult to handle by other methods. It also has some drawbacks:

SURFACE HARDENING TREATMENTS

16.91

1. It is inconvenient to seal the retort. 2. The dangerous pressure might develop during filling of the retort with gas if enough NH3 gas is allowed to condense. However, this hazard can be prevented by employing a safety disk or by keeping the retort at a temperature higher than that of the NH3 supply container.188 3. After 45 hr of operation, the NH3 content expands by about 50%, and further development of the case proceeds at a very slow rate. 4. To limit the depth of white layer to 0.00025 to 0.00050 mm (9.8 to 20 min.), case depth must not exceed 0.50 to 0.63 mm (0.020 to 0.025 in.). Application. Gears, pinions, shafts, bushings, seals, cylinder barrels, clutches and piston rings,180 door sectors (used in car windows), spiral springs, and exhaust valves are hardened by this technique. Gas nitriding is widely employed on steels, e.g., AISI 4130, 4135 modified (with 0.15% V), 4140, 4340, H11, stainless types 302 and 430, and various Nitralloy grades.177 In gas nitriding of stainless steels (type 300, austenitic), surface oxide film which acts as a barrier to nitrogen penetration must be removed by sand or vapor blasting followed by acid pickling.189 However, hard (70 Rc) and superior wear-resistant cases accompanied by a reduced corrosion resistance property are produced after nitriding austenitic stainless steels.185 The 18% Ni maraging steels in the finish-machined and ground condition are also suitable for nitriding between 420 and 450°C (800 and 850°F) for 20 hr or more, which produces increased wear resistance and surface hardness of the case (67 Rc) due to the combined effect of both nitriding and aging treatment in a single operation.188 Bright Nitriding. This is a modified version of gas nitriding using NH3 and H2 gases. Atmosphere gas is continually withdrawn from the nitriding furnace and passed through a temperature-controlled scrubber with a water solution of NaOH. Trace amounts of HCN formed in the nitriding furnaces are eliminated in the scrubber, thereby increasing the rate of nitriding. The scrubber also sets up a desired moisture content in the nitriding atmosphere, decreasing the rate of cyanide formation and hindering the cracking of NH3 to molecular nitrogen and hydrogen. By this process, control over the nitrogen activity of the furnace atmosphere is improved, and nitrided parts are produced containing small or no white layer at the surface. If present, the white layer will consist of only the ductile Fe4N (g ¢) phase.175 Controlled Gas (or Nitreg) Nitriding. In traditional nitriding, the controlled parameters that include the atmosphere flow rate and the resultant dissociation rate of NH3 are insufficient because none of these parameters is directly related to the properties of the nitrided layer.182 Recently, a controlled gas nitriding, called Nitreg nitriding, has been developed based on constant monitoring and maintenance of the nitriding potential by a computer-controlled and fully automated gas nitriding system, according to the modified Lehrer phase diagram, which is a plot of nitriding potential versus temperature (Fig. 16.43).182,190 The nitriding potential is considered as the nitriding capacity of the atmosphere, and its value strictly corresponds to the equilibrium concentration of nitrogen in the steel surface for a given temperature. Mathematically, nitriding power or nitriding potential KN is given by: KN =

pNH3 ( pH2 )3 2

(16.43)

CHAPTER SIXTEEN

16.92

700

Temperature, °C 550 500 450

600

400 31.6

1.5

–1/2

0.5

10.0 e

5.897

g¢

3.2

5.89

1.8

5.87%N

0

1.0

5.85 5.82

10

5.78 5.73

–0.5

–1.0

0.0

7 0.0 05 0. 3%N 0.0

g

1.1

1.2

0.3 5

0 0.0

a

1

0.1

0 0.0

–1.5 1.0

5.6

–1/2

1.0

logrN, atm

17.8

%N 10.00 9.75 9.50 9.25 9.00 8.75 9.50 8.25

rN, atm

Fe-N

1.3 1/K, ¥103

1.4

1.5

0.03 1.6

rN ≤ KN < Nitriding potential FIGURE 16.43 Lehrer KN versus T phase diagram, modified by L. Maldzinski.182,190 (Courtesy of L. Maldzinski.)

where pNH3 and pH2 are partial pressures of NH3 and H2 in the outgoing atmosphere. To obtain a superficial layer with load-bearing capacity, it is necessary that the combined carbon and nitrogen content be £8.5% because higher concentrations are prone to brittle spalling of the layer.182 Alternatively, the use of very low nitriding potential (usually in the final stage of the nitriding process) provides a case without a compound (white) layer due to the nitrogen concentration below its maximum solubility in ferrite. Thus, control of nitrided layer properties implies control of surface nitrogen and carbon concentration and, consequently, control of phase composition.191 Nitriding processing is performed in the extended temperature range of 460 to 600°C (860 to 1112°F). Advantages of this process include192 (1) ease of operation, (2) direct relation of controlling parameter (such as nitriding potential) with nitrogen concentration and properties, (3) predictable compound-layer thickness and phase composition, (4) excellent uniformity of layer, irrespective of the part geometry, and (5) no requirement of finish-grinding.182 Like other gas nitriding processes, the disadvantages of this process are182 (1) need for copper plating or protective paste to mask the surface, (2) need of special activation technique for stainless steels, (3) inability to treat satisfactorily sintered P/M components, and (4) frequent occurrence of embrittlement.176 Examples of applications of Nitreg process are door sectors (used in car windows), spiral springs, automotive exhaust valve, clutch hubs, rocker arm, cast iron

SURFACE HARDENING TREATMENTS

16.93

FIGURE 16.44 Paschen curve showing the relationship between voltage and current for nitrogenhydrogen mixture.

differential housing, hydraulic components of dump truck lifting mechanisms, and crankshafts.192 16.4.1.5 Plasma Nitriding. There are many terminologies used for plasma nitriding, e.g., plasma-ion nitriding, plasma-assisted nitriding, ion nitriding, ionitriding, and glow discharge nitriding. Plasma nitriding is a nitriding process that was developed by Bernhard Berghaus in the 1930s, but the process has gained worldwide popularity in recent years, with about 400 industrial furnaces in operation in Europe and about 150 in North America. Plasma Generation. When a voltage is applied between two electrodes positioned within a nitrogen-hydrogen mixture at some suitable partial pressure (1 to 10 torr), a glow discharge occurs in the abnormal glow discharge range (Fig. 16.44), where total work surfaces conforming to its geometrical form will be completely covered with a uniform purple glow. This glow discharge produces nascent nitrogen in a stable manner by accelerating electrons through an electrical field to give them sufficient kinetic energy to cause the reaction e- Æ N2 = N+ + N + 2e-

(16.44)

That is, nitrogen is dissociated, ionized, and accelerated toward the workpiece (cathode), and electrically conductive gas is formed within a short distance of the workpiece. The positively charged nitrogen ions acquire an electron from the cathode (workpiece) and thus emit a photon. This photon emission during the return to the atomic state causes the visible glow discharge that is characteristic of plasma techniques.193

16.94

CHAPTER SIXTEEN

Usually the heating of the workpieces occurs by the plasma-driven impact of the nitrogen ions, and no extra heat source is needed. However, in newer system designs, convective heating is employed to shorten the cycle times. The thickness of the glow envelope can be varied by temperature, time, atmosphere pressure, gas mix, atmosphere composition, dc voltage, and current. Typically, a large or thick glow envelope is created with lower pressure, higher temperature, high hydrogen concentration in the gas mix, and higher dc voltage and current. There is a desirable glow discharge thickness of about 6 mm (0.25 in.), unless parts with holes or slots require a thinner glow envelope.194 Equipment. There are two types of widely used plasma nitriding systems: coldwall (conventional) technology (Fig. 16.45) and hot-wall (pulse-type) technology (Fig. 16.46). Currently dc pulse is becoming widely accepted as a plasma power control system. RF gas discharge has not yet gained any significant acceptance by industry.194a Other types of plasma such as arc plasma, electron cyclotron resonance (ECR) plasma, high-density plasma focusing (PF) with ion energy up to MeV, and laser plasma are used rarely in metallurgical and industrial heat treatment processes due to shortcomings, typical of each type of plasma source.195 A typical plasma nitriding installation thus comprises (1) a vacuum chamber, (2) the workload, (3) the gas supply, and (4) a cold-wall continuous dc plasma power supply or hot-wall dc high-frequency pulse plasma power supply (Fig. 16.45) and automatic control of the entire process variables, e.g., power voltage, current, current density, temperature of the workpiece and chamber, time, pressure, and gas mixture.196–198 The nitriding process consists of five steps: (1) vessel evacuation with a roughing pump or pump-blower combination to a pressure of 0.05 to 0.1 torr to remove air and contaminants, (2) heating to nitriding temperature,† (3) introduction of a mixture of nitrogen and hydrogen at a pressure, typically 2.7 to 4 mbar or 2 to 3 torr, and composition after stabilization of load at a uniform temperature, (4) glow discharge processing at nitriding temperature, and (5) cooling. In this process, the parts to be nitrided are vigorously cleaned (hydrogen cleaned or sputtered, typically, at 1.3 to 4 mbar or 1 to 3 torr for several minutes) after reaching the full-load temperature, or the passivated part surfaces are activated and placed in a vacuum-tight chamber. For the best results (i.e., shorter processing time), auxiliary heating (e.g., radiant heating or convective heating) is used to heat the parts to the ion nitriding temperature, usually in the range of 400 to 565°C (750 to 1050°F). Note that lower ion nitriding temperature below 350°C greatly decreases the diffusion rate of nitrogen, and temperatures above 580°C are not employed because of a structural change that takes place at 592°C, the eutectoid temperature of the binary iron-nitrogen system (Fig. 16.40). Figure 16.45 shows an improved design incorporating convective heating and cooling.199 During heating and cooling, nitrogen gas at a pressure of 500 torr is used as the convective heat-transfer medium. This unit contains an electrically powered heating element‡ and a (water-cooled) heat exchanger which is moved into the work chamber only during the cooling stage. An outstanding advantage of this design is that the heating elements are used for radiant heating during the low-pressure plasma nitriding stages. This provides better temperature uniformity, irrespective of the variation of the power density of the † At this point usually H2 is added for both convective heating and sputter cleaning. In a conventional dc system, N2 + H2 is introduced on the process commencement, and the heating of the part is by continuous dc power. Heating is effected by a combination of both plasma and conductive means.194a † This design has been reported to be problematical due to nitriding of elements and its susceptibility to early failure by mechanical damage.194a

SURFACE HARDENING TREATMENTS

16.95

(a)

(b)

FIGURE 16.45 Schematic of a coldwall plasma nitriding furnace.199 Gas circulation during (a) heating and (b) cooling. (Courtesy of Ipsen Industries International, GmbH.)

16.96

CHAPTER SIXTEEN

FIGURE 16.46 Schematic of a typical layout of a hot-wall plasma-ion nitriding furnace. (Courtesy of SECO/WARWICK Corporation.)

plasma.197 This allows time-saving (compared to plasma heating), efficient, and simultaneous heating of the large workloads with mixed sizes (Fig. 16.47b). After the load has stabilized at a uniform temperature, the pressure is reduced to 1 to 10 torr (typically, 2.7 to 4 mbar or 2 to 3 torr), and the chamber is backfilled with a mixture of 5 to 50% N2 and 50 to 95% H2 gas (and at times a few percent of CH4).† An electric current at a critical negative dc voltage of 700 to 1000 V‡ is passed between the cathodic workload and the anodic furnace walls, so that a uniform and stable plasma is produced around the workload. That is, nitrogen is dissociated, ionized, and accelerated toward the workpiece (cathode), and electrically conductive gas is formed within a short distance of the workpiece. Upon impact with the † CH4 is used to promote the epsilon phase on low-carbon steels. Medium- to high-carbon steels will promote epsilon phase without the addition of CH4. It is also used in the ferritic nitrocarburizing process.194a ‡ This voltage range is typical for continuous dc system. In the case of pulsed dc, the range is usually 400 to 700 V because the plasma voltage is only being used to create the plasma and probably contribute a small amount of heating in contrast to continuous dc, in which plasma is the primary heating source. The auxiliary heating element method is an attempt to reduce the process voltage. Of course, pressure and part geometry will influence the necessary voltage.194a

SURFACE HARDENING TREATMENTS

16.97

1200

Hardness, HV

1000

High speed steel

800

Cr/Mo nitriding steel

600

Hot work die steel

400

200

8 wt.% Cr stainless steel 0

0.2

0.4 0.6 Depth, mm

0.8

10

(a)

Nitriding 3 hr

600

Nitriding 3 hr

vection

100

a sm Pla ng

by

Accelerated cooling

ati

200

Accelerated cooling

He

300

0

Vacuum cooling

by Con

400

Heating

Temperature °C

500

Cleanup

0

5

10

15

Time, hr (b)

FIGURE 16.47 (a) The hardness versus case depth profile for several steels treated by plasma nitriding. [Courtesy of K. T. Stevens and A. Davies, in Constitution and Properties of Steels (vol. 7: Materials Science and Technology, F. B. Pickering, vol. ed.), VCH, Weinheim, 1992, p. 786.] (b) Comparison of overall process cycle times for a 3-hr nitriding treatment during resistance heating and fan-forced cooling.196 (Reprinted by permission of Fairchild Publications, New York.)

workpiece, the kinetic energy of nitrogen atoms is also converted into heat, which can totally (or in combination with an auxiliary heating source) bring the load to nitriding temperature. This results in an excellent penetration or absorption of nitrogen atoms into, and a gradual heating up of, the workload.196,197,199,200

16.98

CHAPTER SIXTEEN

FIGURE 16.48 Optical micrographs of (a) ion-nitrided Nitralloy 135 three-point bend test specimen at 510°C for 6 hr in a mixture of 33% N2 and 67% H2 at 2-torr pressure and (b) microalloyed steel (containing 0.12 C, 0.58 Si, 1.43 Mn, 0.2 Cr, 0.11 Mn, 0.001 V) at 530°C for 5 hr in a mixture of 20% N2 and 80% H2 at 3-torr pressure.196 (Courtesy of M. H. Jacobs.)

The plasma current, which ionizes the process gas to produce nascent nitrogen for nitriding, offers a substantial fraction of heat needed to maintain the heated load at temperature, i.e., to compensate for heat losses. However, the auxiliary heating system, i.e., gas burners† or resistance heaters, usually supplies the major fraction of the heat needed to bring the load and furnace internals to temperature quickly. After ion nitriding, power is shut off so that the voltage and process gas flow are terminated. Inert gas circulation at 500 torr may be used to accelerate cooling of the load (by using heat exchanger for heat extraction), or slow vacuum cooling is accomplished (to minimize distortion). The furnace is then ready to be unloaded. A high alloy content raises the hardness of the nitrided region but reduces the case depth and provides a very abrupt transition in the hardness profile from the case to the core (Fig. 16.47a). Figure 16.47b196 shows a comparison of two process cycles for a 3-hr nitriding treatment, illustrating time savings during convective heating of workload between room and operating temperatures.197,200,201 Figure 16.48a and b shows the ion-nitrided microstructures, respectively, of the Nitralloy 135 and microalloyed steel (containing 0.12 C, 0.58 Si, 1.43 Mn, 0.2 Cr, 0.11 Mo, and 0.001 V) obtained at different process parameters. Note that ion nitriding, like other nitriding processes, produces several distinct structural zones: (1) thin white layers containing mostly e-iron nitride (Fe2–3N) with some g ¢-Fe4N phase;‡ (2) a diffusion zone of fine iron-alloy nitrides;§ and (3) a gradient zone of interstitial nitrogen.190 The advantages of this process over conventional (gas) nitriding are as follows: 1. Processing time for a given case depth is reduced (as small as one-third to one-half of the time required for traditional gas nitriding). Saturation cycles are short. †

There is only one gas-fired unit built which has not been a good commercial venture.194a The proportion of constituent phases depends on the carbon content of the steel.194a § The nitrides formed in the diffusion zone are dependent on the alloy content of the steel. For example, Nitralloy will form nitrides of Cr, Mo, and V. Thus, fine alloy nitrides will form only when low-alloy and plain carbon steels are used.194a ‡

SURFACE HARDENING TREATMENTS

16.99

2. High-quality product in terms of surface structure and minimal distortion is produced. 3. There is precise control of thickness and composition of the compound zone or white layer.201 In fact, the white layer produced by ion nitriding is very adherent and coherent, and its microstructure is dense and microcrystalline, unlike the coarsegrained, porous white layers usually formed both by gas and salt bath nitriding.196 Thus, ion nitriding gives a less brittle case with higher superficial hardness (800 to 1200 HV), low friction coefficient, greater high-cycle fatigue strength, and enhanced wear resistance. White layer or compound zone development in ion nitriding remains unchanged at 0.0025 to 0.005 mm (0.0001 to 0.0002 in.). At 400°C (750°F) the case depth does not vary appreciably with time.198 More precise control of the nitrogen supply at the surface of the workpiece and the ability to choose either an e- or g ¢-monophase layer or to avoid white layer formation completely are benefits.194 4. Case depth is very uniform, even on complex-shaped parts; i.e., it achieves a far greater control over product properties than those of the conventional methods. Extreme care is needed when nitriding blind holes. 5. The process can be conducted at comparatively lower temperatures (as low as 375°C, or 700°F) due to plasma activation (which is not present in gas nitriding), enabling high core-strength retention, high surface hardness, low distortion, and improvement in dimensional stability in tools and precision parts.194,202 Low-temperature nitriding at 300°C, with a very thin (24-mm) e-Fe2–3N layer and nitrogen solid solution increases the hardness of and doubles the lifetime of the 60MD8 (AFNOR: Fe-0.6C-1.8Si-0.7Mn-0.3Cr-0.5Mo-0.2V) steel cutting knife tools in the beech wood rotary peeling process. Significantly, the friction coefficient of nitrided steel against beech wood is slightly higher than that of non-treated steel. It seems that from the point of view of a wet wood peeling process with a nitrided knife, the knife hardness is more important than the friction coefficient. This is a low-alloy steel commonly used for manufacture of cutting tools for rotary peelers of wet wood (at the speed of about 1 to 2 m/s).203 6. Plasma nitriding at 480°C improved the surface life of H13 and D2 steel roll by up to seven-fold by changing substrate material from AISI H13 to D2 steel and a change in surface geometry from a nonconformal (V) to a conformal (U) profile. Shorter nitriding times developed the e-Fe2–3N phase, and longer nitriding times developed g ¢-Fe4N with a release of molecular nitrogen, which may be partially responsible for observed increases in porosity.204 This can be modified by an appropriate ratio of N2/H2. 7. Cleaning and depassivation of surface layers (such as oxide) occur by the sputtering action by high-kinetic-energy ion bombardment.205,206 8. Easy masking (or shielding) or selective hardening is an advantage. Steel sheets or sleeves are employed as mechanical masking devices for stopping off, thus eliminating the need for copper or tin plating. 194,202 Sometimes, stop-off paints are used.194a 9. There are no environmental pollution, hazard, or disposal problems, and the part usually does not need any other surface finishing operation.198 10. There is accurate reproducibility of the operation for batch-type processing. 11. Operating costs are relatively low [due to low energy consumption, high (70 to 90%) energy efficiency of this mechanism, and low gas utilization], floor space

16.100

CHAPTER SIXTEEN

and operating supervision are reduced, and there is the ease of operating the equipment.198,206 12. There is the ability to automate.194 13. There is the ability to treat routinely materials with passive layers such as austenitic stainless steels and titanium alloys.207 Limitations of this process include the following: 1. High capital cost (which is offset by lower operating costs and improved metallurgical properties)198 2. Need for precision fixtures with electric connections (to avoid localized overheating)194 3. Lack of feasibility of liquid quenching for plain carbon steels 4. Long processing times compared to nitrocarburizing processes11 5. Problems with temperature measurement and temperature nonuniformity, especially for parts with complex geometry (the temperature uniformity of the load can, however, be promoted by the selection of a suitable pressure and by design of the jigging system and shields, attached to the workpieces207) 6. Prone to overheating, if pressure, voltage, and current are not accurately controlled (not applicable with modern controls and pulse plasma) 7. Results sensitive to part geometry and arrangement in furnace retort 8. Need for highly skilled and experienced operator182 Application. This process can be applied to all steels including a wide variety of alloy steels such as AISI 4140, 4340, Nitralloys; AISI M, A, D, H, and T series of tool steels; 17-4 PH; austenitic and martensitic stainless steels; and cast irons.211 Examples include gears, crankshafts, cylinder liners, pistons, machinery (plastic extrusion, agricultural, food, etc.), tooling (dies, drills, molds, punches, etc.), and P/M parts (such as transmission gears). The increased lubricity of white layers, together with higher hardness and fatigue strength, has offered significant growth of the plasma tool and die industry. Hot-work dies, which often fail by thermal fatigue and sticking, have particularly benefited from plasma nitriding following quenching and tempering. Low-temperature (below 400°C) plasma nitriding of austenitic and duplex stainless steel can be accomplished in the commercial plasma nitriding units to provide improved wear resistance without appreciable drop in the corrosion resistance.208 However, depassivation of the surface is required prior to nitriding.198 A wide range of component sizes can be processed, ranging from ball-point pen balls to 10-ton rolls or gears. Pulse Plasma Nitriding. The hot-wall, pulsed plasma nitriding furnace in conjunction with partial pressure control provides separate power supplies for part heating and for enhancing plasma process control (Fig. 16.46). Comparison of plasma and pulse plasma nitrided 4140 steel shows no difference in steel microstructure after nitriding, and nearly the same surface roughness and surface microhardness values, whereas higher content of nitrogen in nitriding atmosphere produces a higher surface microhardness in the case of pulse plasma nitriding. Pulse plasma nitriding makes it possible to obtain the same nitriding depth in a shorter time compared to conventional plasma nitriding. Compared to normal hardened AISI 4140 steel, the plasma and pulse plasma nitriding in a nitrogen-poor atmosphere improves the tribological properties of the 4140 steel. The results show that surface

SURFACE HARDENING TREATMENTS

16.101

treatment has practically no effect on the coefficient of friction; on the other hand, plasma and pulse plasma nitriding in a nitrogen-poor atmosphere reduces the wear of AISI 4140 steel compared to the hardened steel.209 In general, the advantages of pulse plasma ion nitriding are the following:193,210 1. Independent control of process parameters, notably electric current density, ion concentration, surface temperature, and nitrogen diffusion rate. 2. No arc discharge problems linked to the parts not adequately cleaned. 3. Possibility of dense loading of the furnace with parts without any risk of localized overheating or temperature nonuniformity. 4. Desired properties at low operating temperatures and no need for fast cooling. 5. A uniform, nonspalling, reproducible nitrided layer with a specific thickness, surface hardness, and hardness profile can be achieved on complex parts, even comprising narrow slots and deep holes. 6. Desirable metallurgical properties can be consistently produced even on passivated surfaces. 7. Control of the composition and phases of the compound layer. In some cases, the compound layer can be eliminated, leaving only the diffusion zone. 8. It can run automatically without operator supervision or intervention. 9. The process can be stopped at any time during a cycle and cooled to room temperature with no risk or adverse side effect. 10. Electric power and cooling water demands are comparatively low. 11. The total processing time is comparatively shorter without the need for secondary cleaning operations. 16.4.1.6 Fluidized-Bed Nitriding. This process is similar to the standard gaseous nitriding process as far as temperature, precise control of case formation, and surface hardness obtained are concerned. However, advantages claimed include increased deep case (up to 0.07 in. thick) and reduced processing times due to shorter handling and faster heating rates. However, nitriding in a fluidized bed can lead to high NH3 consumption when used for long cycle times.211,212 These costs are offset by the energy savings by using shorter cycles and from higher load densities.213 During the operation of the furnace, the parts are held in baskets or racks which may be either rested on a load support frame or suspended about 2 in. (50 mm) above the retort bottom. When parts are loaded, nitrogen is the fluidizing gas. After reaching the proper temperature, dry anhydrous NH3 alone or diluted and mixed with other gases is introduced, through the diffusion plate at the bottom of the bed. The used or effluent atmosphere escapes through a burn-off vent in the furnace lid. At the end of the nitriding cycle, the furnace is purged with nitrogen for about 2 min, and the load is withdrawn from the furnace and put in the fluidized-bed cooler operating on nitrogen to avoid discoloration and maintain the surface characteristics.213 Fluidized-bed nitriding is mostly done between 510 and 565°C (950 and 1050°F). It has been shown that the dissociation rate of anhydrous NH3 in a fluidized bed operating at 534°C (1000°F) is ⬃50%.213 After nitriding, parts may be quenched rather than the usual slow-cooled. On some steels, this is advantageous in improving surface hardnesses. Fluidized-bed nitrocarburizing can be made possible by adding methane to NH3 in the fluidized-bed atmosphere. (However, it is not normally used except to promote the epsilon phase on low-carbon and low-alloy steels.)

16.102

CHAPTER SIXTEEN

The process might be suited to nitriding components with deep small holes or blind holes, which are usually a concern for uniform nitriding by conventional gas nitriding or ion nitriding. The fluid-pulse nitriding (performed in the same fluidized furnace, with an additional electrical pulse signal generator) can produce somewhat the same microhardness profile, case depth, and higher surface hardness, but offers a decrease in the fluid-bed gas consumption by up to 52%.214 16.4.2 Nitrocarburizing Ferritic nitrocarburizing is a thermochemical surface hardening treatment which is rapidly developing, particularly to upgrade the performance of low-carbon steel. This process involves diffusional addition of nitrogen and carbon to the surface of ferrous components (where the enrichment of nitrogen is predominant) at temperatures on the order of 570°C (1060°F), i.e., completely within the ferrite phase field for 1 to 3 hr.171,215 This produces a thin 0.00055- to 0.0008-in. (0.014- to 0.02-mm) ductile single-phase e-iron carbonitride (Fe2–3N,C),“white layer,” or compound layer formed at the component surface.216 It possesses increased hardness and excellent corrosion resistance and tribological properties (oil retention and high wear and scuffing resistance) after a relatively short treatment time, usually less than 3 hr. In addition, a “diffusion zone” beneath the compound layer provides a considerable increase in tensile strength, stiffness, fatigue resistance, and pitting resistance of treated materials. Fatigue properties can be further improved by quenching in oil or water from the treatment temperature.217–219 Nitrocarburizing low-carbon nonalloy steel followed by oil-quenching provides corrosion resistance under conditions of humidity and neutral salt spray, thus diminishing the need for more conventional electroplated nickel finishes.217 The part to be nitrocarburized should be quenched and tempered or stressrelieved at 593°C (1100°F) minimum. 16.4.2.1 Gas Nitrocarburizing. Currently, gas nitrocarburizing is, next to gas carburizing, the most widely employed thermochemical surface hardening process.220 The atmosphere used consists of (1) NH3 diluted with a carrier gas; (2) 50% NH3 and 50% endothermic gas (AGA type 302); (3) 35% NH3 and 65% refined exothermic gas (AGA type 201, nominally 97% nitrogen) enriched with hydrocarbon gas; (4) propane/NH3/O2 mixture; or (5) methane/NH3/O2 mixture.171 Any gas leakage in the furnace and around furnace doors must be minimized, and double pilots should be used. Processing steps are as follows: 1. Cleaning and degreasing the parts thoroughly. 2. Loading the furnace at room temperature, and purging all the air from the furnace with nitrogen. 3. Raising the furnace temperature to 570°C under reduced flow rate. 4. Maintenance of proper nitrocarburizing atmosphere for 3 hr after attaining this temperature. 5. Purging with nitrogen after shutdown. 6. Unloading the furnace and cooling the parts—either oil- or gas-quenching. Applications. This process finds a wide variety of applications such as on: (1) textile machinery gears, pump cylinder blocks, rocker-arm spacers, and jet nozzles

SURFACE HARDENING TREATMENTS

16.103

TABLE 16.16 Atmosphere Systems Used to Produce a Compound Layer Depth of 17 ± 1 mm at 570°C221 System Ammonia Ammonia/nitrogen Ammonia/endothermic gas Ammonia/nitrogen/CO2 Ammonia/nitrogen/CO Ammonia/nitrogen/CH4 Ammonia/nitrogen/air Ammonia/exothermic gas

Ratio of ingoing gases

Residual ammonia, %

— 1 : 0.33 1:1 1 : 1.38 : 0.23 1 : 1.33 : 0.18 1 : 1.43 : 0.08 1 : 1.25 : 0.25 1 : 1.5

55 55 42 21 20 22 24 20

Material: 0.15% C nonalloy steel. Nitriding time: 2 hr, followed by oil quench. Number of atmosphere volume changes: 7.5 per hour. Courtesy of Wolfson Heat Treatment Centre, England.

for wear resistance; and (2) crankshafts and driveshafts for improved fatigue properties. This can be successfully applied to wrought and sintered plain carbon and alloy steels, stainless steels, and cast irons. The most significant improvement is, however, found with low-carbon nonalloy steels in both antiscuffing bending and rotational fatigue strength. Proprietary Methods. Nitrotec, a development of the early 1980s in the gas nitrocarburizing process with the addition of a precooling oxidation sequence, is a trademark of Lucas industries. As the name indicates, this process consists of three steps: namely, nitriding, oxidizing, and protection (e.g., sealing).221,222 The salient feature of the process is to use a wide range of atmospheres of NH3 and a carrier gas but with a minimum residual of 55% NH3 level for NH3 or NH3/N2 atmosphere, 42% NH3 level for NH3/endothermic gas mixture, and 20% NH3 level for NH3/N2/CO atmosphere and 20% NH3 level for NH3/exothermic gas atmospheres, shown in Table 16.16.221 The selection of the atmosphere actually depends on the economics based on local conditions. The first step of this Nitrotec treatment on steel specimens produces an e-iron carbonitride nonmetallic compound layer to a depth of 25 to 40 mm (0.001 to 0.0015 in.) thickness. This compound layer consists of the formation of a substantial void or porosity and e-carbonitride compound. An increase in porosity level of the compound layer occurs toward the surface in proportion to an increase in the nitrogen content. A dimensional growth is just 2.5 to 8 mm (0.0001 to 0.0003 in.). A maximum surface nitrogen content of 8% approaches closely to a nonmetallic, hexagonal close-packed e-nitride phase. A maximum hardness level of 1100 HK (25 g) (or 66 to 67 Rc) is reached just below the extreme surface and represents the limit of the nonporous part of the compound layer.217 Below this layer lies the nitrogen-rich subsurface/diffusion zone. The improvements in tensile and fatigue strengths and toughness of the Nitrotectreated parts are attributed to the greater depth of nitrogen-rich substrate (diffusion zone) beneath the compound layer. Actually, the enhancement of these properties depends on (1) the temperature of the part prior to quenching, (2) treatment time, (3) cooling rate, and (4) thickness of the part. The combined effect of e-carbonitride and voids in the nitrocarburized parts gives rise to two important characteristics: (1) better wear resistance than that of either carburized or carbonitrided parts and (2) oil-retention and antiscuffing

16.104

CHAPTER SIXTEEN

properties due to the microporous iron carbonitride compound layer similar to the nonferrous sintered porous metal bearings. The second step (i.e., oxidation stage) consists of exposing to an oxidizing atmosphere at a suitable temperature for a short time (flash oxidation) so that a thin (500 hr against salt spray exposures). The combined effect of post-nitriding oxidation and quenching together with a specific compound layer and depth produces an attractive aesthetic appearance and wear resistance. Applications include windshield-wiper linkage assemblies (Fig. 16.49) (for motor vehicles to replace the stainless steel with mild steel),221 automotive fan motor, viscous slip differential, bumper armatures, seat sliders, steel bars and tubes up to 7.3 m (24 ft) long, piston rods in hydraulic cylinders for such applications as food processing equipment, dry-ice machines, power-shearing systems, and shipboard material-handling equipment. Other potential applications are papermaking and oil and gas drilling industries where both corrosion and wear resistance are in demand.222

FIGURE 16.49 Windshield wiper linkage system utilizing Nitrotec surface treatment and used in the automotive industry. It contains five component parts: (a) spindle link, (b) primary link, (c) black plate, (d) drive link, and (e) support plate.221 (Courtesy of Wolfson Heat Treatment Centre, England.)

SURFACE HARDENING TREATMENTS

16.105

Oxynitrocarburizing. Oxynitrocarburizing consists of two steps. The first is nitrocarburizing in which an iron carbonitride compound layer, about 10 to 50 mm thick, occurs. In the second oxidation step, 5 mm) and a surface hardness of ≥350 HV and (b) an inner diffusion zone composed mainly of a g ¢-Fe4N-type solid solution. This process allows a low-nitrogen and high-carbon addition to occur at the surface layers. As the carbon content in the gas mixture increases, the compound layer thickness increases to a maximum and then decreases. At a very high carbon content, the compound layer does not form at all; instead, deposition of amorphous carbon occurs on the component surface.233 Following nitrocarburizing, the workpieces are allowed to cool under controlled vacuum conditions. The advantages over other nitrocarburizing treatments are:234 1. Reduced processing time 2. Environmental safety (i.e., less exhaust gas containing NxOy compounds and greater energy efficiency) 3. Minimum use of treatment gases 4. Fast cooling that favors a preferable monophased e-[Fe2–3(N,C)] structure of compound layer at room temperature235 Applications include P/M chain gear wheels.236 16.4.2.4 Fluid Bed Nitrocarburizing. Although complete replacement of the atmosphere within a fluidized bed can take place within a few minutes, it is a better approach to “condition” the fluidized beds with the process atmosphere for considerably longer periods (5 to 35 min) prior to the introduction of workpieces, for consistent results.237 The fluid bed nitrocarburizing method uses a fluid bed furnace in the temperature range of 316 to 649°C (600 to 1200°F) and a 50% NH3, 50% natural gas atmosphere; 60% NH3, 23% natural gas, 17% N2; 50% NH3, 40% natural gas, 10% N2; 40% NH3, 40% natural gas, 20% N2; 10 to 14% NH3, N2, natural gas; or NH3/CH4 mixture in varying proportions, depending on the end use of the heat-treated components.237,238 Since fluidized-bed nitrocarburizing is characterized by very high nitriding potential, there is the likelihood of severe porosity with subsequent compound layer breakdown and/or occurrence of spalling. In that case, not more than 10 to 14% NH3 may be needed for fluidized-bed nitrocarburizing, employing NH3, natural gas, and nitrogen.237 The depth of the compound layer can be controlled by varying the atmosphere, time, and temperature. The process imparts a light blue/black lustrous oxide finish which enhances the sliding wear resistance, impact resistance, and in some cases, galling resistance. The fluidized-bed nitrocarburizing has several advantages over the conventional atmosphere/integral quench nitrocarburizing:

SURFACE HARDENING TREATMENTS

16.111

1. The furnace is very tight because of upward pressure of the gases. 2. It results in lower distortion and greater bending strength of the treated parts without fracture. 3. There is a possibility of stuffing the parts together in the basket. 4. Very even finishing is produced due to fluid action. 5. Unlike in salt bath processing, no cyanide process is used and not all the part holes are plugged up after heat treatment. 6. It provides competition for some surface treatments, e.g., titanium nitride coating. Applications include parts such as hot forged dies and punches; cold-formed sleeves; cutting tools; hobs; milling cutters, typically of D2 or M2 type; paper slitter blades; and finished drills.238 16.4.2.5 Austenitic Nitrocarburizing. Austenitic nitrocarburizing can be considered from the viewpoint of atmosphere as the low-temperature carbonitriding process with an option for a higher nitrogen potential and for oxidation. Typically the process is accomplished in the same kind of sealed-quenched furnace as carbonitriding with similar atmosphere except NH3 content which is increased to 20%. One proprietary austenitic nitrocarburizing atmosphere system is 20%NH3/40% cracked methanol/40% N.239 Standard austenitic nitrocarburizing is accomplished at 700°C for 2 hr followed by an oil quench. However, a 3-hr treatment provides an improved indentation resistance. After the 2-hr treatment, mild steel shows a 25- to 30-mm-thick e-compound layer and an underlying carbonitrided case to a depth of 125 to 150 mm. The case structure beneath the compound layer, after quenching, comprises two zones: an austenitic zone and a back-up martensitic zone. Austenitic transformation is achieved either by isothermal treatment to produce lower bainite or by deep freezing and subsequent tempering to form tempered martensite with a microhardness in the range of 700 to 900 HV.239 The advantage of this treatment is the increased hardened case below the compound layer which extends the range of its applications.236

16.5 SUPERCARBURIZING The supercarburizing or saturation carburizing process was developed by O. Cullen in 1957. In this process the carbon concentration exceeds its normal solubility limit (e.g., 1.2%) in steel to as much as 4%. However, the usual surface carbon concentration ranges are 1.8 to 2.2%, 2.0 to 2.4%, 2.4 to 2.8%, and over 3%. These carbon levels can be obtained with conventional carburizing methods such as pack, gas (hydrocarbon-enriched gas or methanol plus nitrogen), and so forth, both at atmospheric pressure and in vacuum.240 The supercarburized surfaces are characterized by high density of globular, granular, or spheroidal carbides with increased wear resistance.241 Figure 16.53 shows typical case carbon probes of supercarburized steels. The shapes of these curves can be varied by suitable adjustment in the heat treatment cycle. This process is suited for alloy steels containing substantial quantities of carbide formers, especially Cr, Mo, and W. These include medium-alloyed and more

16.112

CHAPTER SIXTEEN

FIGURE 16.53 Typical case carbon probes of supercarburized steels.240 (Courtesy of Roy F. Kern.)

highly alloyed case hardening steels such as AISI 4118, 5120, 8620, 8720, 8822, 9310, and 52100.240 However, similar treatment applied to 12% Cr steels has produced dispersion of chromium carbides (Cr7C3) of 0.5-mm mean diameter constituting 30% of the microstructure.242 The process consists of heating to a carburizing atmosphere at 927 to 982°C (1700 to 1800°F) for 2 to 5 hr, to ensure that carbides do not dissolve completely in austenite but remain adjacent to the grain boundaries as leaky carbide peripheries enclosing austenite grains. They dissolve more carbon and form more carbide at temperature. Upon cooling to a subcritical temperature, say, 677°C (1250°F) for ⬃1 hr, the carbide network thickens. This cycle of reheating to 927 to 982°C and subsequent cooling to 677°C is repeated 2 to 4 times, and finally the parts are oilquenched after holding at sufficiently high hardening temperature, for example, 816°C (1500°F), for 0.5 to 1 hr to avoid the network carbide, and are subsequently tempered at 177°C (350°F). Table 16.17 lists two typical supercarburizing heat treatment cycles A and B applied to ASTM A335-P3B steel (composition: 0.15% C, 0.48% Mn, 0.26% Si, 0.20% Ni, 2.08% Cr, 0.52% Mo, 0.008% S, 0.02% P, and 0.07% Cu). A typical phase analysis of a low-alloy steel sample after regular carburizing and hardening and after being supercarburized and hardened is as follows:

Normal carburizing Supercarburizing

Martensite, %

Austenite, %

Carbide, %

81 65

14 10

5 25

The final hardened structure consists of a large-volume fraction of carbides at the surface with a high surface hardness and a depletion of carbide formers such as Cr and Mo in the matrix. Figure 16.54 shows the surface microstructure of a 9310 steel sample, pack carburized [to a case depth of 0.305 in. (7.75 mm)] and hardened. The case carbon concentration at 0.005 in. (0.127 mm) from the surface was found to be 2.03%.

SURFACE HARDENING TREATMENTS

16.113

TABLE 16.17 Two Typical Supercarburizing Cycles A and B for ASTM-A335-P3B Steel240

Carburize at 1700°F for: Set control to 1250°F Work at 1250°F for: Set control to 1700°F Work at 1700°F for: Carburize at 1700°F for: Set control to 1250°F Work at 1250°F for: Set control to 1700°F Work at 1700°F for: Carburize at 1700°F for: Set control to 1250°F Work at 1250°F for: Set control to 1700°F Work at 1700°F for: Carburize at 1700°F for: Set control to 1500°F Work at 1500°F for: Hold at 1500°F for: Quench in 120°F oil for: Wash and temper at 350°F for: Case hardness: Surface case carbon content:

Cycle A

Cycle B

2 hr

2 hr

65 min

75 min

40 min 5 hr

75 min 5 hr

60 min

80 min

40 min 4 hr

60 min 5 hr

—

65 min

— —

60 min 51/2 hr

15 min 55 min 15 min 3 hr 65 Rc 2.75%

20 min 30 min 20 min 3 hr 67 Rc 3.26%

Furnace: Surface combustion all-case Atmosphere: Furnace purge 400 cfh RX-gen. gas 40 cfh natural gas Carburize 400 cfh RX gas 70 cfh natural gas Dew point 9–19°F 0.0% CO2 CO 17–18% 9.1–10.0% CH4 Cool and diffuse 400 cfh RX gas Reprinted by permission of Fairchild Publications, New York.

Advantages of supercarburizing over conventional carburizing, when properly treated, are:240 1. Improved abrasive wear resistance in diesel-engine injection pump parts 2. Increased compressive residual stress over -690 MPa (-100 ksi) and greater matrix toughness due to the formation of a large amount of lath martensite241 3. Increased long-life bending-fatigue strength by ⬃25% 4. Increased contact stress capability in parts such as gears and rolling bearings 5. Increased resistance to scoring, if NH3 is present in the hardening cycle Limitations when compared to conventional carburizing are as follows: During grinding, wheel surfaces are abraded away more rapidly. When harder grinding wheels are used, there is always a danger of burning.240

16.114

CHAPTER SIXTEEN

FIGURE 16.54 Surface microstructure of supercarburized and hardened 9310 steel. 4% picral etch.240 (Courtesy of Roy F. Kern.)

16.6 BORIDING (OR BORONIZING) Boriding, or boronizing, is a thermochemical surface hardening process that can be applied to a wide variety of ferrous, nonferrous, and cermet materials. The process involves heating well-cleaned material in the range of 700 to 1000°C (1292 to 1832°F), preferably for 1 to 12 hr, in contact with a boronaceous solid powder (boronizing compound), paste, liquid, or gaseous medium. Other developments in thermochemical boriding include plasma boriding, pulsed-plasma boriding, and fluidized-bed boriding. Currently, multicomponent boriding is also used. This section describes mainly the various media used for thermochemical boriding, their advantages, limitations, and applications.

16.6.1 Characteristic Features of Boride Layers During boriding, the diffusion and subsequent absorption of boron atoms into the metallic lattice of the component surface form the interstitial iron-boron compounds (or phases).243–247 Figure 16.55 is the Fe-B phase diagram showing various phases and their compositions and structures.248 The resulting layer may consist of either a single-phase boride or a polyphase boride layer. The morphology (Fig. 16.56),249 growth, and phase composition of the boride layer can be influenced by the alloying elements in the base material. The microhardness of the borided layer also depends strongly on the composition and structure of the boride layer and the composition of the base material (Table 16.18).245,250,251 16.6.1.1 Advantages. Boride layers possess a number of characteristic features with special advantages over conventional case hardened layers.

SURFACE HARDENING TREATMENTS

FIGURE 16.55

16.115

Fe-B phase diagram.248 (Courtesy of ASM International.)

1. Boride layers have extremely high hardness values (between 1450 and 5000 HV) with high melting points of the constituent phases (Table 16.18). The typical surface hardness values of borided steels compared with other treatments and other hard materials are listed in Table 16.19.245 This clearly illustrates that the hardness of boride layers produced on carbon steels is much greater than that produced by any other conventional surface (hardening) treatments; it exceeds that of the hardened tool steel, hard chrome electroplate, and is equivalent to that of tungsten carbide. 2. The combination of a high surface hardness and a low surface coefficient of friction of the borided layer also makes a significant contribution in combating the

FIGURE 16.56 Effect of steel composition on the morphology and thickness of the boride layer.249 (Courtesy of R. Chatterjee-Fischer.)

TABLE 16.18 Melting Point and Microhardness of Different Boride Phases Formed during Boriding of Different Substrate Materials245,250,251

Substrate

Constituent phases in boride layer

Microhardness of layer, HV (or kg/mm2)

FeB Fe2B CoB Co2B Co3B CoB Co2B Co3B (?) Ni4B3 Ni2B Ni3B — Mo2B MoB2 Mo2B5 W2B5 TiB TiB2 TiB TiB2 NbB2 NbB4 Ta2B TaB2 HfB2 ZrB2 ReB

1900–2100 1800–2000 1850 1500–1600 700–800 2200 (100 g)† ~1550 (100 g)† 700–800 1600 1500 900 1700 (200 g)†† 1660 2330 2400–2700 2660 2500 3370

Fe Co Co-27.5Cr Ni Inco 100 Mo W Ti Ti-6Al-4V Nb Ta Hf Zr Re †

100-g load.

††

Melting point °C

°F

1390 — — — — — — — — — —

2535 — — — — — — — — — —

3000 (100 g)† 2200

2000 ~2100 2100 2300 ~1900 2980 — — 3050

3630 ~3810 3810 4170 3450 5395 — — 5520

2500 2900 2250 2700–2900

3200–3500 3200 3250 3040 2100

5790–6330 5790 5880 5500 3810

200-g load.

16.116

SURFACE HARDENING TREATMENTS

16.117

TABLE 16.19 Typical Surface Hardness of Borided Steels Compared with Other Treatments and Hard Materials245 Material Borided mild steel Borided AISI H13 die steel Borided AISI A2 steel Quenched steel Hardened and tempered H13 die steel Hardened and tempered A2 die steel High-speed steel BM42 Nitrided steels Carburized low-alloy steels Hard chromium plating Cemented carbides, WC + Co Al2O3 + ZrO2 ceramic Al2O3 + TiC + ZrO2 ceramic Sialon ceramic TiN TiC SiC B4C Diamond

Microhardness, kg/mm2 or HV 1,600 1,800 1,900 900 540–600 630–700 900–910 650–1,700 650–950 1,000–1,200 1,160–1,820 (30 kg) 1,483 (30 kg) 1,738 (30 kg) 1,569 (30 kg) 2,000 3,500 4,000 5,000 >10,000

main wear mechanisms: adhesion, tribooxidation, abrasion, and surface fatigue.246,252 This fact has enabled the mold makers to substitute easier-to-machine steels for the base metal to still obtain wear resistance and antigalling properties superior to those of the original material.253 Figure 16.57 shows the effect of boriding on abrasive wear resistance of borided C45 steel, titanium, and tantalum as a function of number of revolutions (or stressing period) based on the Faville test.254 Figure 16.58 shows the influence of steel composition on abrasive wear resistance.249,252 3. Hardness of the boride layer can be retained at higher temperatures than, e.g., that of nitrided cases. 4. A wide variety of steels, including through-hardenable steels, are compatible with the processes.255 5. Boriding, which can considerably increase corrosion-erosion resistance of ferrous materials in nonoxidizing dilute acids (Fig. 16.59) and alkali media, is increasingly used to this advantage in many industrial applications.246 6. Borided surfaces have moderate oxidation resistance (up to 850°C, or 1550°F) and are quite resistant to attack by molten metals. 7. Borided parts have increased fatigue life and service performance under oxidizing and corrosive environments. 8. The plasma boronized thermally sprayed nickel stellite layer has been reported to cause a significant increase in the surface hardness, the resistance to corrosion and friction wear, and the adhesion of stellite to the substrate, thereby widening the application range of modern surface layers.257 9. Boronized steels have increased resistance to molten aluminum and zinc baths degradation at 630°C (1166°F) and 500°C (932°F), respectively, for 6 to 120 hr,

16.118

CHAPTER SIXTEEN

FIGURE 16.57 Effect of boriding on wear resistance (Faville test). (a) 0.45% C steel (C45) borided at 900°C (1650°F) for 3 hr. (b) Titanium borided at 1000°C (1830°F) for 24 hr. (c) Tantalum borided at 1000°C (1830°F) for 8 hr.254 (Courtesy of R. Chatterjee-Fischer and O. Schaaber.)

SURFACE HARDENING TREATMENTS

16.119

FIGURE 16.58 Effect of steel composition (nominal values in wt%) on wear resistance under abrasive wear (dv = thickness of the boride layer). Test conditions: DP-U grinding tester, SiC-paper 220, testing time 6 min.249,252

which could increase their applicability in various industries handling light molten metals.258 16.6.1.2 Disadvantages. Disadvantages of boronizing treatments are as follows: 1. The techniques are inflexible and rather labor-intensive, making the process less cost-effective than other thermochemical surface hardening treatments such as gas carburizing and plasma nitriding. Both gas carburizing and plasma nitriding have the advantage over boronizing because these two processes are flexible systems, offer reduced operating and maintenance costs, require shorter processing times, and are relatively easy to operate. It is, therefore, suited to engineering components that need high hardness and outstanding wear and corrosion resistance of the boride layers, and/or where cheaper labor is available.245

16.120

CHAPTER SIXTEEN

FIGURE 16.59 Corroding effect of mineral acids on boronized and nonboronized (a) 0.45% C (Ck 45) steel and (b) AISI 321 (X10CrNiTi-18-9) at a temperature of 56°C (130°F).246,256

2. The growth (i.e., the increase in volume) resulting from boronizing is 5 to 25% of the layer thickness (or case depth) (e.g., a 25-mm, or 1000-min., layer would have a growth of 1.25 to 6.25 mm, or 50 to 250 min.); its magnitude depends on the base material composition but remains consistent for a given combination of material and treatment cycle. However, it can be predicted for a given part geometry and

SURFACE HARDENING TREATMENTS

16.121

boronizing treatment. For treatment of precision parts, where little stock removal is permitted, an allowance of ⬃20 to 25% dimensional increase of the final boride layer thickness must be provided. 3. Partial removal of the boride layer for closer tolerance requirements is made possible only by a subsequent diamond lapping because conventional grinding causes fracture of the layer. Thus, precise boronizing is mostly practiced for components with a large cross-sectional area.245 4. Boriding of most steels provides a marginal increase, if any, in the bending fatigue endurance limit, although some improvement in the corrosion-fatigue strength has been noticed. 5. In general, the rolling contact fatigue properties of borided alloy steel parts are very poor compared to those of carburized and nitrided steels at high contact loads (2000 N, or 450 lbf). This is why boronizing treatments of gears are limited to those screw designs where transverse loading of gear teeth is minimized.245 6. There is frequently a need to harden and temper the tool after boriding,259 which requires a vacuum or inert atmosphere to preserve the integrity of the boride layer.

16.6.2 Boriding of Ferrous Materials Unlike the carburizing treatment on ferrous materials, where there is a gradual decrease in composition from the carbon-rich surface to the substrate, the boriding of ferrous materials results in the formation of either a single-phase or double-phase layer of borides with definite compositions. The single-phase boride layer consists of Fe2B, while the double-phase layer consists of an outer boron-rich, dark-etching phase of FeB and an inner boron-deficient light-etching phase of Fe2B. The formation of either a single or double phase depends on the availability of boron.260 16.6.2.1 Characteristics of FeB and Fe2B Layers. The formation of a single Fe2B phase (with a sawtooth morphology due to preferred diffusion direction) is more desirable than a double-phase layer with FeB. The boron-rich FeB phase is considered undesirable, in part, because FeB is more brittle than the iron subboride Fe2B layer. Also, because FeB and Fe2B are formed under tensile and compressive stresses, respectively, crack formation is often observed at or in the neighborhood of the FeB/Fe2B interface of a double-phase layer. These cracks may lead to flaking and spalling when a mechanical strain is applied247 or even separation (Fig. 16.60) when a component is undergoing a thermal and/or mechanical shock. Therefore, the boron-rich FeB phase should be avoided or minimized in the boride layer.247 If the formation of FeB is unavoidable, care should be taken to ensure that no closed FeB zones occur.261 It has also been reported that the tribological properties depend on the microstructure of the boride layer. The dual-phase, FeB-Fe2B layers are not inferior to those of monophase Fe2B layers, provided that the porous surface zone directly beneath the surface is removed.262 Alternatively, a thinner layer is favored because of less development of brittle and porous surface-zone formation and flaking. Typical properties of the FeB phase are: 1. Microhardness of ⬃19 to 21 GPa (2.7 ¥ 106 to 3.0 ¥ 106 psi) (hardness, 1900 to 2100 HV0.1) 2. Modulus of elasticity of 590 GPa (85 ¥ 106 psi)

16.122

CHAPTER SIXTEEN

FIGURE 16.60 Separation of two-phase boride layer on a low-carbon (St 37) steel (borided at 900°C, or 1650°F, for 4 hr) caused by grinding with cutting-off disk. 200X.256

3. Density of 6.75 g/cm3 (0.244 lb/in.3) 4. Thermal expansion coefficient of 23 ¥ 10-6/°C (13 ¥ 10-6/°F) between 200 and 600°C (400 and 1100°F)249,263 5. Composition with 16 to 16.2 wt% boron248 6. Orthorhombic crystal structure with four iron and four boron atoms per unit cell 7. Lattice parameters: a = 4.053 Å, b = 5.495 Å, and c = 2.946 Å 8. Residual stress on cooling: tensile Layers of Fe2B. The formation of single-phase Fe2B layers with a sawtooth morphology is desirable in the boriding of ferrous materials.264 A single Fe2B phase can be obtained from a double FeB-Fe2B phase by a subsequent vacuum or salt bath treatment for several hours above 800°C (1470°F), which may be followed by oilquenching to increase substrate properties.265 Typical properties of Fe2B phase are 1. Microhardness of about 18 to 20 GPa (2.6 ¥ 106 to 2.9 ¥ 106 psi) (hardness 1800 to 2000 HV0.1) 2. Modulus of elasticity of 285 to 295 GPa (41 ¥ 106 to 43 ¥ 106 psi) 3. Density of 7.43 g/cm3 (0.268 lb/in.3) 4. Thermal expansion coefficient of 7.65 ¥ 10-6/°C (4.25 ¥ 10-6/°F) and 9.2 ¥ 10-6/°C (5.1 ¥ 10-6/°F) in the range of 200 to 600°C (400 to 1100°F) and 100 to 800°C (200 to 1500°F), respectively249,263 5. Composition with 8.8 wt% boron248 6. Body-centered tetragonal structure with 12 atoms per unit cell 7. Lattice parameters: a = 5.078 Å and c = 4.249 Å 8. Residual stress on cooling: compressive The solubility of boron in ferrite and austenite is very small (0.3%) because the transgranular strength is high, and therefore a small amount of grain boundary sulfide precipitation is enough to induce intergranular failure.17 The phosphorus content has been regarded with the greatest concern in connection with burning. At constant phosphorus level, there is an increase in the overheating temperature with the increase of sulfur content, whereas the burning onset temperature decreases. Burning temperature is reduced with the increase in phosphorus content. At low sulfur contents, a wide gap between overheating and burning exists. For example, in the case of vacuum remelted steels, the temperature gap between the onset of overheating and burning is ~300 to 400°C (~570 to 750°F), and there is a remote possibility of burning occurring within the forging range, unless the overheating is severe.2 However, at high sulfur content, the gap becomes narrow. Temperature. To avoid overheating, care must be exercised in choosing a correct heating temperature so that uneven heating, flame impingement, and so forth do not occur.3 Cooling Rates. The cooling rate through the overheating range affects the size and dispersion of intergranular a-MnS particles. The intermediate cooling rate generally employed, 10 to 200°C/min (20 to 360°F/min), gives rise to maximum faceting as well as to the greatest loss in impact strength. However, slow and rapid cooling rates will suppress overheating. At very slow cooling rates, the sulfide particles become large, fewer in number, and more widely dispersed; and they have no more deleterious effects than the other inclusions already present. At rapid rates, the sulfide inclusions are too fine to produce any damaging effect.18 Methods of Manufacture. Electroslag remelted steels are less susceptible than vacuum remelted steels, presumably due to the difference in oxygen level. Similarly, nickel steels are more prone to overheating. Vacuum-remelted steels have a lower overheating temperature than some comparable air-melted steels.

17.2.4 Prevention of Overheating and Burning To prevent overheating of steels, a properly selected temperature should lie between a temperature low enough for the metal to be safe and one high

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.9

enough to be sufficiently plastic. The better the temperature control, the better the compromise. Severe overheating can be reduced to mild overheating by soaking the steel at 1200°C (2200°F); with care, it may be removed completely. Hot-working through the overheating range to a low finish temperature is also reported to remove the effects of overheating. The alloying additions with a greater sulfide-forming tendency, such as calcium,† zirconium, cerium (~0.03% of the melt), or mixed rare earth metal (in the form of misch metal containing 52% Ce, 25% La, and 12% Nd), have been shown to prevent overheating by increasing significantly both the overheating temperature and the mechanical properties of the steel (e.g., ductility and toughness). Provided that a high Ce/S ratio (>2) existed, a complete change in sulfide morphology occurred in low-alloy steels where the elongated MnS inclusion occurring in the untreated steel was totally replaced by small globular type-I rare earth sulfides and oxysulfides of high thermal stability even after austenitizing at 1400°C (2550°F).2 This treatment does not show intergranular faceting. Burning can also be avoided in the same way by treating with calcium, zirconium, cerium, or mixed rare earth addition to form refractory, less-soluble sulfides. Control of Cooling Rates. Control of cooling rates is not a practical method for large forgings because extremely slow cooling is prohibitively time-consuming and causes excessive scaling and decarburization, and rapid quenching from high temperatures produces cracking and distortion of the parts.2 17.2.5 Reclamation of Overheated Steel Severely overheated steels can often be completely restored by any of the following heat treatments: 1. Repeated normalizing (as many as six) starting at temperatures up to 50 to 100°C (90 to 180°F) higher than usual, followed by a standard normalizing treatment.2 2. Repeated oil-hardening and tempering treatments after prolonged soaking at 950 to 1150°C (1740 to 2100°F) in a noncarburizing (neutral) atmosphere. Rehardening more than three times is not advisable. 3. Soaking at 900 to 1150°C (1650 to 2100°F) for several hours. This causes growth of MnS particles by the Ostwald ripening process and results in an excessive scale formation and a loss of dimensional accuracy of the forgings. Alternatively, a large extent of hot reduction minimizes or reduces overheating.19

17.3 RESIDUAL STRESSES During heat treatment (while hot), thermal and transformation residual stresses can cause yielding (plastic flow), hence growth and/or distortion of a part. When the part is cold, the residual stresses should not exceed the yield strength of the material, so their effect on dimensions is elastic.20 Heat treatment often causes stress- and strain-related problems such as residual stress, quench cracks, and deformation and/or distortion. The residual stress may be †

Calcium-treatment of some machinable steels is for modification of the oxides.

17.10

CHAPTER SEVENTEEN

defined as the self-equilibrating internal or locked-in stress remaining within a body with no applied (external) force, external constraint, or temperature gradient.21,22 There are two types of residual stresses: 1. Macro- or long-range residual stress is a first-order stress that represents an average of body stresses over all the phases in polyphase materials. Macroresidual stresses act over large regions compared to the grain size of the material. Traditionally, engineers consider only this type of residual stress when designing mechanical parts. 2. Microresidual stress, also termed tesselated stress or short-range stress, is a second-order or texture stress which is associated with lattice defects (such as vacancies, dislocations, and pile-up of dislocations) and fine precipitates (e.g., martensite).23–25 Microresidual stress is the average stress across one grain or part of the grain of the material. This information is indispensable in studying the essential behavior of material deformation. These two types of residual stresses may also be classified further as a tensile or compressive stress located near the surface or in the body of a material. This section focuses on the effects, development, control, and measurement of long-range residual stresses.

17.3.1 Effects of Residual Stress The major effects of residual stress include dimensional changes and either an increased or decreased resistance to crack initiation and propagation, depending on whether the surface residual stresses are compressive or tensile.20 Dimensional changes occur when the residual stress (or a portion of it) in a body is eliminated. In terms of crack initiation, residual stresses can be either beneficial or detrimental, depending on whether the stress is tensile or compressive. Compressive Residual Stress. Because residual stresses are algebraically summed with applied stresses, residual compressive stresses in the surface layers are generally helpful because the built-in compressive stresses can reduce the effects of imposed tensile stresses that may produce cracking or failure. Compressive stresses therefore contribute to the improvement of fatigue strength, fatigue life, and resistance to stress corrosion cracking in a part and an increase in the bending strength of brittle ceramics and glass.25 Figure 17.3 shows that the endurance limit fatigue strength of selected steels increases with the surface residual compressive stress developed by specific heat treatment and surface processing. It is also apparent that, in the presence of high compressive stress, a poor microstructure in steel samples has a small influence on good endurance limit fatigue strength.26–28 These fatigue improvements are of great significance in components, particularly where stress raisers, such as notches, keyways, oil holes, and so forth, are highly desirable in the design of components (e.g., crankshafts, half-shafts, and so on).29 Many fabrication methods have been developed to exploit this phenomenon. Prestressed parts (including shrink-fits, prestressed concrete, interference fits, bolted parts, coined holes, wire-wound concrete pipe), mechanical surface working processes (such as conventional shotpeening, gravity peening, laser shock peening, roller burnishing, surface rolling, lapping, and so on) of hardened ferrous and nonferrous alloys, and surface harden-

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.11

FIGURE 17.3 Effect of surface residual stress on the endurance limit of selected steel. All samples were water-quenched except as shown, and all specimen dimensions are given in inches.26,27 (Reprinted by permission of John Wiley & Sons, New York.)

ing treatments are widely used to produce residual compressive stresses at the component surface. Shot-peening further improves the surface compressive residual stress (Fig. 17.4a and b)30–33 and skin hardness in carburized steel, which leads to substantial increases in bending fatigue performance and inducement of more random finish by eliminating directional machining marks.34 Figure 17.5a and b shows the residual stress distribution produced by air-blast shot-peening, gravity peening, and laser shock peening (LSP) in Ti-6Al-4V and Inconel 718 after final machining. It is clear from these figures that LSP residual stress distribution diminishes linearly with depth, without reducing the surface compression stress.35 In roller burnishing, a rolling force is applied to the surface, using either rollers or spherical bearings to locally strengthen the surface of the part (especially at fillets and in grooves) by producing compressive residual stress in a case hardened surface. It may also be applied to non-case-hardened parts (e.g., fillet rolling of some crankshafts).36 Tensile Residual Stress. Tensile residual stress at the surface of a part is usually undesirable because it adds to any applied tensile stresses and effectively increases the stress levels; it may cause unpredicted stress-corrosion cracking (due to the combined effect of stress and environment), quench cracking, and grinding checks at low external stresses, and tend to reduce fatigue life and strength of a part. In this case the extent of residual stresses may be closer to or even larger than the strength of the material.

17.12

CHAPTER SEVENTEEN

FIGURE 17.4 (a) Effect of shot-peening at different shot velocities on compressive residual stress in carburized 16MnCr5 steel (1.23% Mn, 1.08% Cr).30,31 (b) Residual stress as a function of distance from the surface of the carburized SAE 4320 in the as-carburized condition and after various shot-peening treatments.32,33

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.13

FIGURE 17.5 Residual stress distribution developed in (a) Ti-6Al-4V and (b) Inconel 718 test specimens.35 (Courtesy of P. S. Prevey, D. J. Hornbach, and P. W. Mason.)

Residual tensile stresses in the interior of a component also may be damaging because of the existence and consequence of defects that serve as stress raisers in the interior part. The uncommon phenomenon of delayed cracking, in the absence of adverse environments and large applied stresses, has now been attributed to the action of residual stresses on minute defects in the material.29 For example, a 17.5cm-diameter (6.9-in.) ¥ 125-cm-long (49.2-in.) steel shaft exploded into several pieces while lying free of any applied loads, on a laboratory floor. Under normal loading, it would have required a tensile strength larger than 150 MPa (22 ksi) to rupture the shaft. Hence, the understanding of residual stress formation is very

CHAPTER SEVENTEEN

17.14

important, and this must be given due consideration in the manufacture and performance analysis of processed parts.29

17.3.2

Development of Residual Stress in Processed Parts

Variations in stresses, temperature, and chemical species within the body during processing cause the production of macroresidual stresses. Various manufacturing processes such as forming, machining and assembling, heat treatment, shot-peening, casting, welding, flame cutting, and plating render their characteristic residual stress pattern to processed parts. Table 17.3 lists a summary of compressive and tensile residual stresses at the surface of parts fabricated by common manufacturing processes. In heat-treated parts, residual stresses may be classified as those caused by a thermal gradient alone and those caused by a thermal gradient in combination with a structural change (phase transformation). When a steel part is quenched from the austenitizing temperature to room temperature, a residual stress pattern is established due to a combination of thermal gradient and local transformation-induced volume expansion. Thermal contraction develops nonuniform thermal (or quenching) stress due to the different rates of cooling experienced by the surface and the interior of the steel part. Ferrite-to-austenite transformation involves contraction while the part is thermally expanding, hence the development of residual stresses. Because of the very low yield strength of steel at such temperatures, these residual stresses could con-

TABLE 17.3 Summary of Compressive and Tensile Residual Stresses at the Surface of Parts Created by Common Manufacturing Processes25 Compression at the surface Surface working: shot-peening, surface rolling, lapping, and so on Rod or wire drawing with shallow penetration† Rolling with shallow penetration† Swaging with shallow penetration† Tube sinking of the inner surface Coining around holes Plastic bending of the stretched side Grinding under gentle conditions Hammer peening Quenching without phase transformation Direct-hardening steel (not throughhardened) Case hardening steel Induction and flame hardening Prestressing Ion exchange † ‡

Tension at the surface Road or wire drawing with deep penetration Rolling with deep penetration Swaging with deep penetration Tube sinking of the outer surface Plastic bending of the shortened side Grinding: normal practice and abusive conditions Direct-hardening steel (through-hardened)‡ Decarburization of steel surface Weldment (last portion to reach room temperature) Machining: turning, milling Built-up surface of shaft Electrical discharge machining Flame cutting

Shallow penetration refers to ⱗ1% reduction in area or thickness; deep penetration refers to ⲏ1%. Depends on the efficiency of quenching medium. Reprinted by permission of Pergamon Press, Plc.

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.15

TABLE 17.4 Changes in Volume during the Transformation of Austenite into Different Phases4 Change in volume, %, as a function of carbon content, %C

Transformation Spheroidized pearlite Æ austenite Austenite Æ martensite Spheroidized pearlite Æ martensite Austenite Æ lower bainite Spheroidized pearlite Æ lower bainite Austenite Æ upper bainite Spheroidized pearlite Æ upper bainite

-4.64 + 2.21 ¥ (%C) 4.64 - 0.53 ¥ (%C) 1.68 ¥ (%C) 4.64 - 1.43 ¥ (%C) 0.78 ¥ (%C) 4.64 - 2.21 ¥ (%C) 0

Courtesy of the Institute of Materials, U.K.

tribute to distortion.20 Transformational volume expansion induces transformation stress arising from the transformation of austenite into martensite or other transformation products.37 Table 17.4 lists the changes in volume during the transformation of austenite into different structural constituents.38 17.3.2.1 Thermal (Contraction) Residual Stresses. The relation between the thermal stress sth during cooling (quenching) and the corresponding temperature gradient in the component is given by s th = E DT a

(17.1)

where E is the modulus of elasticity and a is the thermal coefficient of expansion of the material. It is thus apparent that thermal stresses are greatest for materials with high elastic modulus and coefficient of thermal expansion. Temperature gradient is also a function of thermal conductivity. Hence, it is quite unlikely for one to develop high-temperature gradients in good thermal conductors (e.g., copper and aluminum), but it is much more likely in steel and titanium.39 Another term involving thermal conductivity, called thermal diffusivity Dth, is sometimes used in context with temperature gradient. It is defined as Dth = k/rc, where k is the thermal conductivity, r is the density, and c is the specific heat. It is clear that low Dth (or k) promotes large temperature gradient or thermal contraction. It should be emphasized that large size of the part and high heating or cooling rates (severity) of quenching medium also augment temperature gradients, leading to large thermal contraction. Table 17.5 lists some of the relevant material properties that affect thermal and residual stresses.39 The variations with temperature are important to heat-treat distortion (and development of residual stress).36 17.3.2.2 Residual Stress Pattern due to Thermal Contraction. Residual stress is developed during heating or cooling of a solid part that involves thermal volume changes without solid-state phase transformation. This situation exists, e.g., when a steel part is heated to or cooled from a tempering temperature below A1. These residual stresses rely on there being a thermal gradient; once a part has attained a uniform temperature throughout (either once hot or once cooled), there are no thermal gradients and, therefore, no consequential residual stresses.20

CHAPTER SEVENTEEN

17.16

TABLE 17.5 Relevant Physical Properties in the Development of Thermal Stresses41 Modulus of elasticity

Coefficient of expansion

Thermal conductivity

Metal

GPa

psi ¥ 106

10-6/K

10-6/°F

W m-1 k-1

Btu in./ft2 · h · °F

Pure iron (ferrite) Typical austenitic steel Aluminum Copper Titanium

206 200 71 117 125

30 29 10 17 18

12 18 23 17 9

7 10 13 9 5

80 15 201 385 23

555 100 1400 2670 160

Courtesy of Wolfson Heat Treatment Centre, England.

FIGURE 17.6 Development of thermal and residual stresses in the longitudinal direction in a 100-mm-diameter (4-in.) steel bar on water-quenching from the austenitizing temperature of 850°C (1560°F). Transformation stresses are taken into consideration.42

When one is considering a part that is already austenitized and then quenched, the stresses developed during the quench are of a high magnitude and in excess of the hot yield strength of cooling austenite. Hence some yielding will occur.20 Figure 17.6 shows the development of longitudinal thermally induced residual stresses in a 100-mm-diameter (4-in.) steel bar on water-quenching from the austenitizing temperature of 850°C (1560°F).40 At the start of cooling, the surface temperature S falls drastically compared to the center temperature C (top left sketch of Fig. 17.6). At time w, the temperature difference between the surface and core is at a maximum of about 550°C (1020°F), corresponding to a thermal stress of 1200 MPa (80 tons/in.2) due to linear differential contraction of about 0.6%, if relaxation does not take place. Under these conditions, tensile stresses are developed in the case with a maximum value of a (lower diagram), corresponding to time w in the upper diagram, and the core will contract, producing compressive stresses

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.17

with a maximum of c (at time w). The combined effect of tensile and compressive stresses on the surface and core, respectively, will result in residual stresses as indicated by curve C, where a complete neutralization of stress will occur at some lower temperature u. Further decrease in temperature, therefore, produces longitudinal, compressive residual stresses at the surface and the tensile stresses at the core, as shown in the lower right-hand diagram of Fig. 17.6. Figure 17.7a is a schematic illustration of the distribution of residual stress over the diameter of a quenched bar due solely to thermal contraction in the longitudinal, tangential, and radial directions.22 The maximum residual stress attained on quenching increases as the quenching temperature and quenching power of the coolant are increased. Tempered glass is made by utilizing quenching techniques in which glass is heated uniformly to the annealing temperature and then surface-cooled rapidly by cold air blasts. This produces compressive surface stresses to counteract any tensile bending stress, if developed during loading of the glass, thereby increasing its load-carrying capacity.43

FIGURE 17.7 Schematic illustration of the distribution of residual stress over the diameter of a quenched bar in the longitudinal, tangential, and radial directions due to (a) thermal contraction and (c) both thermal and transformational volume changes. (b) Schematic illustration of orientation of directions.22 (Reprinted by permission of The McGraw-Hill Companies, Inc., New York.)

17.18

CHAPTER SEVENTEEN

17.3.2.3 Residual Stress Pattern due to Thermal and Transformational Volume Changes.42 During quench hardening of a steel (or other hardenable alloy) part, hard martensite forms at the surface layers, associated with the volume expansion, whereas the remainder of the part is still hot and ductile austenite. Later, the remainder austenite transforms to martensite, but its volumetric expansion is restricted by the hardened surface layer. This restraint causes the central portion to be under compression with the outer surface under tension. Figure 17.7c illustrates the residual stress distribution over the diameter of a quenched bar, showing volume expansion associated with phase transformation in the longitudinal, tangential, and radial directions.22 At the same time during the final cooling of the interior, its thermal contraction is hindered by the hardened surface layers. This restraint in contraction produces tensile stresses in the interior and compressive stresses at the outer surface (Fig. 17.7a). However, the situation shown in Fig. 17.7c prevails, provided that the net volumetric expansion in the interior, after the surface has hardened, is larger than the remaining thermal contraction. In some particular conditions, these volumetric changes can produce sufficiently large residual stresses that can cause plastic deformation on cooling, leading to warping or distortion of the steel part. While plastic deformation appears to reduce the severity of quenching stresses, in most severe quenching the quenching stresses are so high that they do not get sufficiently released by plastic deformation. Consequently, the large residual stress remaining may reach or even exceed the fracture stress of steel. This localized rupture or fracture is called quench cracking.42,43 It should be emphasized again that for a given grade of steel, both large size of the part and higher quenching speed contribute to the larger value of thermal contraction, compared to the volumetric expansion of martensite. In contrast, when the parts are thin and the quenching rate is not high, thermal contraction of the part subsequent to the hardening of the surface will be smaller than the volumetric expansion of martensite. Similarly, for a given quenching rate, the temperature gradients decrease with decreasing section thickness, and consequently the thermal component of the residual stress is also decreased.27 Figure 17.8a shows the continuous cooling transformation diagram of DIN 22CrMo44 low-alloy steel exhibiting austenitic decomposition with the superimposed cooling curves of the surface and center in round bars of varying dimensions. If the large-diameter (100-mm, or 4-in.) bar is water-quenched (i.e., for slack quenching) from 850°C (1562°F), martensite transformation occurs at the surface, and pearlitic + bainitic transformation occurs at the center, resulting in a residual stress pattern (top of Fig. 17.8b) similar to that due solely to thermal stress (Fig. 17.7a). During the rapid quenching of the medium-size (30-mm, or 1.2-in.) bar diameter, the start of bainite transformation at the center coincides approximately with the transformation of martensite on the surface. This results in compressive stresses at both the surface and center, with tensile stresses in the intermediate region (middle of Fig. 17.8b). When the smaller-diameter (10-mm, or 0.4-in.) bar is drastically quenched (e.g., in brine), the entire bar transforms to martensite. This is associated with very little temperature variation between the surface and the center of the part. In this situation, tensile residual stress is developed at the surface and compressive stress at the center of the bar (bottom, Fig. 17.8b).44,45 Although the shallower hardening steels exhibit higher surface compressive stresses, deep hardening steels may develop moderately high surface compressive stresses with severe water quenching. When these deep hardening steels are through-hardened in a less efficient quenchant, they may exhibit surface tensile stresses.27,41 Rose has pointed out the importance of transformations of core and surface before and after the stress reversal. According to him, the tensile surface

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.19

FIGURE 17.8 (a) Continuous cooling transformation diagrams of DIN 22CrMo44 steel showing austenitic decomposition with the superimposed cooling curves of the surface and center during water-quenching of round bars of varying dimensions. (b) The corresponding residual stress pattern developed because of thermal and transformational volume changes.44,45 (Reprinted by permission of Butterworths, London; after A. Rose.)

residual stress occurs when the core transforms after, and the surface transforms before, the stress reversal (Fig. 17.7c and bottom of Fig. 17.8), whereas compressive surface residual stress takes place when the core transforms before, and the surface transforms after, the stress reversal (top of Fig. 17.8b). His analysis is capable of explaining complex stress patterns for various combinations of part sizes, quenching rate, and steel hardenability.24 However, the residual stress pattern in the hardened steels can be modified either with different transformation characteristics or during the tempering and finish-machining (after hardening) operations.

17.20

CHAPTER SEVENTEEN

17.3.2.4 Residual Stress Pattern after Surface Hardening. In general, thermochemical and thermal surface-hardening treatments produce beneficial compressive residual stresses at the surface (Fig. 17.9). Carburizing and Quenching. When low-carbon steels are carburized and then quenched, first martensite (as an example) forms at some distance from the surface (Figs. 17.9 and 17.10),23,46 where the part temperature has dropped below the higher interior MS temperatures. The volume changes at this stage are quickly accommodated by the neighboring austenite due to its low flow stresses and the high temperatures. The surface austenite does not transform due to its low MS (reduced by ~ 470°C, or 878°F for a 1% increase in carbon). When the temperature falls below MS in the (higher-carbon) surface regions, the expansion of the formed martensite at the surface is constrained by the interior martensite that formed earlier. Consequently, the surface microstructure is held in compression. The contributory factors affecting this process and the position of maximum compressive stress include carbon and alloy contents which set MS temperatures, and steel hardenability; total case depths; quenching severity, which depends on the temperature of start; and the temperature-dependent plastic flow behavior of martensite and austenite. Despite the complexity of the interactions that affect the formation of residual stresses, hardened carburized parts with the martensite-austenite microstructure described earlier usually develop favorable compressive stresses.47

FIGURE 17.9 Relationship between carbon content, retained austenite, and residual stress pattern. It shows the development of peak compressive stress some distance away from the surface.23 (Reprinted by permission of Pergamon Press, Plc.)

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

FIGURE 17.10 process.46

17.21

Axial stress distribution (given in MPa) in carburized gear during quenching

According to Koistinen48 and Salonen,49 the peak compressive stress takes place at 50 to 60% of the total case depth, corresponding to about 0.5 to 0.6% carbon level, which produces a low retained austenite content and martensite hardness around the maximum. Another factor that might influence this compressive residual stress profile is that the martensite formed in lower-carbon regions of the case is of the lath type, which also affects the retained austenite content.23 The reversal sign of residual stress takes place at or near the case/core interface. Figure 17.10 shows the details of generation of axial stress distribution of a carburized gear (made from deeper hardening steel) during quenching. In the early stages, the contour lines of equal stress were largely unaffected by the surface profile. Later a zone of high compressive stress distribution occurred in the central portion of the teeth, which remained until the end of the quench.46 Nitriding and Nitrocarburizing. In nitriding, a compressive residual stress is set up in the surface layers due to the volume increase arising from the formation of nitrides in the shallow nitrided layer. High-temperature nitriding produces a little relaxation of stresses, whereas low-temperature nitriding imparts a maximum residual stress. Carburized surface hardening, on the other hand, relies on the martensite transformation for its hardness and its residual stresses.20 In nitrocarburizing, improvements in residual surface compressive stress and fatigue strength depend on the hardness and depth of the diffusion zone. These properties, in turn, decrease with increasing carbon and alloy content (i.e., increased hardenability). During quenching, after nitrocarburizing, a (macro-) compressive residual stress is produced in the compound layer and gamma prime phase.50 When nitrocarburized parts are rapidly quenched, the above properties are further enhanced.51 Boriding. In borided steel processed at 900°C (1650°F), a high compressive residual stress is developed at the surface layers (Fig. 17.11), which consists of FeB

17.22

CHAPTER SEVENTEEN

and Fe2B phases.50 This is attributed to the lower thermal expansion coefficient and the larger specific volume in a borided layer compared to those in a ferrite matrix.21,52 Induction Hardening. In an induction-hardened steel part, a compressive surface residual stress is produced when wear-resistant hard martensite (with slightly lower density) is formed on the surface of a section concurrently with volume expansion while the nonhardened core remains essentially unchanged (Fig. 17.12).53,54 The magnitude of the compressive stress, which is affected by both thermal contraction and martensite formation, may be a considerable fraction of the yield strength, which permits the application of significantly higher stresses than could normally be possible in fatigue loading. As in the carburizing practice, the surface compressive residual stresses are usually found to increase with depth below the surface54 (Fig. 17.12).53 A fairly sharp transition to a tensile state takes place near the hardness drop-off between the case and unhardened surrounding

FIGURE 17.11 Residual stress distribution of FeB and Fe2B layers in borided layers in borided steel processed at 900°C (1650°F).21,52

FIGURE 17.12 A typical hardness and residual stress profile in induction hardened (to 3-mm, or 0.12-in. case depth) and tempered (at 260°C, or 500°F) 1045 steel.53 (Reprinted by permission of ASM International, Materials Park, Ohio.)

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.23

material. With an increase in distance from the steep transition, the tensile condition gradually fades away toward zero stress.53 In induction hardening, an increase in hardenability changes the depth at which transition from compressive to tensile stress occurs. The increase in the rate of heating produces an increase in the maximum compressive and tensile residual stresses without affecting the mode of stress distribution.55 One of the problem areas with induction hardening of gears is the tendency to develop high-magnitude tensile residual stresses in a narrow zone beneath the hardened layer, thereby providing a potential failure site. This characteristic is taken into consideration when designing gears for induction hardening and selecting a case depth. It is also reflected in the design allowable stress value for bending.56,57 Plasma Nitriding and Electron Beam Hardening. In plasma nitrided and plasma nitrided plus electron beam treated steel, compressive residual stresses have been observed. However, the measured values have been reported to be considerably higher in the latter (Table 17.6).58 Laser Hardening. In laser hardened medium-carbon steel (AISI 1042), 4xxx series, and 80W Cr V8 tool steel, high surface compressive residual stress has been observed.59,60 The characteristics of the residual stress field depend on the material, the specimen type, and the laser hardening condition, including the laser processing parameters and the scanning pattern of the laser beam.60 17.3.2.5 Residual Stress in Other Processing Steps. As welding progresses, the temperature distribution in the weldment becomes nonuniform and varies as a result of localized heating of the weldment by the welding heat source. During the welding cycle, comprising heating and cooling, complex strains develop in the weld metal and adjacent areas. As a result, appreciable residual stresses remain after the completion of welding. Since the weld metal and heat-affected zone contract on cooling (Fig. 17.13a),61 they are restrained by the cool adjacent part. This produces

TABLE 17.6 Compressive Residual Stress in Plasma Nitrided and Plasma Nitrided plus Electron Beam Treated Fe-0.42C-0.96Cr-0.6Mn-0.37Si Steel58

Depth, mm

Phase

Retained austenite, wt%

Compressive residual stresses in nitrous ferrite (martensite), MPa

Plasma nitriding 50 200 400

a solid solution + g ¢ a solid solution + g ¢ a solid solution

— — —

435 638 520

Plasma nitriding + electron beam treatment 50 200 400

a solid solution + g solid solution a solid solution + g solid solution a solid solution

14

595

12

785

—

682

Note: a solid solution (nitrous martensite); g ¢-Fe4N phase; g solid solution (nitrous or retained austenite).

CHAPTER SEVENTEEN

17.24

tensile residual stress in the weldment region and compressive residual stress in the surrounding base-metal region (Fig. 17.13b). In general, a steep residual stress gradient is developed because of the steep tendency of the thermal gradient. This may, in turn, lead to hot cracking (between columnar grains) or severe centerline cracking in the weld area.62 Catastrophic failures of welded bridge and all-welded ships are mostly attributed to the existence of large and dangerous tensile residual stress in them.63 Poor design features such as square-corner hatches (in ships) are also to be avoided.20 The machining and grinding operations in manufacturing are important, since they are always utilized to produce the finished surface.† It has been shown that gentle surface grinding, using a soft sharp wheel and slow downfeed, produces compressive residual stress at the surface, whereas conventional (normal practice) and abusive grinding results in surface tensile stresses of very high magnitude (Fig. 17.14).25,64 However, the gentle grinding method is expensive from the viewpoint of operating time and wear of the wheel. † In gearing, at least, a gearing step is an unwanted groove along the bottom edge of a flank and somewhat in the tooth fillet; it is produced by flank grinding. It can happen when flank grinding is used to correct excessive heat treat distortion.20

FIGURE 17.13 (a) The transverse shrinkage occurring in butt weldments. (b) Longitudinal residual stress patterns in the weldment and surrounding regions. This also shows longitudinal shrinkage in a butt weld.61

90

Residual stress, MPa (Compression) (Tension)

600

60

400

Abusive 30

200

Conventional 0

0 –200 –400

Residual stress, ksi

800

Depth below surface, mil 3.15 6.3 9.45 12.6 120

Gentle –30 –60 0 80 160 240 320 Depth below surface, mm

FIGURE 17.14 Residual stress distribution after gentle, conventional, and abusive grinding of hardened 4340 steel.25 (Reprinted by permission of Pergamon Press, Plc.)

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.25

As a result of temperature gradients during cooling, castings develop compressive stresses at the surface and tensile stresses in the interior.25 However, transient temperature gradient and phase transformation occurring during the early stages of solidification and cooling of continuous steel castings in the mold give rise to the development of harmful residual stresses, leading to the formation of cracks.65 Chemical processes such as electroplating, scale formation, and corrosion of metals can produce residual stresses due to coherency strains arising from the matching tendency of crystal structures of the outer surface product with the crystal structure of the adjacent layer.25 Residual stresses are also introduced when heattreated parts are subjected to successive heating and cooling cycles during service conditions. 17.3.2.6 Residual Stress in the Heat-Treated Nonferrous Alloys. In nonferrous alloys, notably age-hardenable aluminum alloys, copper-beryllium alloys, certain nickel-base superalloys, and so on, a significant amount of thermal stress is generated during quenching prior to precipitation hardening. The quenching process in this condition does not invariably involve a phase change; rather, this is confined to the post-quenching aging treatment. In other nonferrous alloys such as uranium and titanium alloys, the final structural condition is not obtained by a slow cool. When high-strength titanium alloy is quenched from a solution annealing temperature of 850 to 1000°C (1560 to 1830°F), it develops large residual stress caused by poor thermal conductivity of titanium, leading to high-temperature gradient. This problem can, however, be avoided by stress-relief annealing at 650 to 700°C (1200 to 1290°F), which produces a slight reduction in mechanical properties. When a high-strength aluminum age-hardening alloy is rapidly quenched from the solution temperature, high thermal and residual stresses are induced due to the high coefficient of expansion of aluminum. Uphill quenching from liquid nitrogen temperature (-196°C, or -320°F) in a steam blast alleviates this problem. This induces stresses opposite in sign to those developed on water-quenching from the solutionizing and cancels out their effect. This is followed by aging of the alloy in the conventional manner.39 Fast polyalkylene glycol (PAG) quenching of solution-treated aluminum alloys tends to reduce residual stress levels because of its more uniform heat extraction rate (thermal shock is smaller, and thereby machining is less likely to produce further distortion), thereby helping solve major and long-standing distortion problems among aluminum workpieces.66

17.3.3

Control of Residual Stresses in Heat-Treated Parts

Table 17.7 lists some typical values of maximum compressive residual stresses developed in the surface hardened steels that have been reported in the literature.39 It is worth noting that there is a marked influence of tempering on the residual stress level. Tempering must be accomplished at 150 to 180°C (300 to 356°F) to maintain 50 to 60% retention of the residual stress level obtained after quenching because a higher tempering temperature greatly reduces surface compressive stresses. However, a higher thermal stress-relieving (at ~600°C, or 1110°F) of steel parts is used for fabrications, castings, heavy flame-cut sections, and so forth. Alternatively, serious residual tensile stresses may be avoided effectively by gentle grinding of the surface.

17.26

CHAPTER SEVENTEEN

TABLE 17.7 Compiled Summary of the Maximum Residual Stresses in Surface Heat-Treated Steels39 Residual stress (longitudinal) Steel 832M13 (type)

805A20 805A20

805A17 805A17

Heat treatment Carburized at 970°C (1780°F) to 1-mm (0.04-in.) case with 0.8% surface carbon Direct-quenched Direct-quenched, -80°C (-110°F) subzero treatment Direct-quenched, -90°C (-130°F) subzero treatment, tempered Carburized and quenched Carburized to 1.1–1.5 mm (0.043–0.06 in.) case at 920°C (1690°F), direct oil quench, no temper Carburized to 1.1–1.5 mm (0.043–0.06 in.) case at 920°C (1690°F), direct oil quench, tempered 150°C (300°F)

897M39 905M39

Nitrided to case depth of about 0.5 mm (0.02 in.)

Cold-rolled steel

Induction hardened, untempered Induction hardened, tempered 200°C (390°F) Induction hardened, tempered 300°C (570°F) Induction hardened, tempered 400°C (750°F)

MPa

ksi

280 340

40.5 49.0

200

29.0

240–340† 190–230

35.0–49.0 27.5–33.5

400 150–200

58 22–29

400–600 800–1000

58.0–87.0 116.0–145.0

1000 650 350 170

145.0 94.0 51 24.5

†

Immediately subsurface, that is, 0.05 mm (0.002 in.). Courtesy of Wolfson Heat Treatment Centre, England.

17.3.4 Measurement of Residual Stresses There are two methods of measuring residual stresses: the destructive method, also called the dissection method, and the nondestructive methods comprising mainly x-ray diffraction, neutron diffraction, and ultrasonic and magnetic methods. 17.3.4.1 Destructive (or Dissection) Method. The dissection method relies on removing layers of material from a part and measuring the dimensional or strain changes (by strain gauges, traveling microscope, micrometer, and whatever) resulting from each layer removal event. This method, being old but reasonably accurate, uses well-established methods and can be employed in confined situations at site.67 However, it is tedious, timeconsuming, and expensive.68 The other drawbacks are the destructive, or at best semidestructive, nature of the method and its ability to measure only the macroresidual stresses. The hole-drilling method is an example of the dissection technique, and it is used extensively for measuring residual stresses, which are less than nominally one-half the yield strength of the material. It consists of the mounting of strain gauges or a three-element strain-gauge rosette on the surface and measurement of strains. Then a rigidly guided milling cutter is used to drill a small, straight, circular,

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.27

perpendicular, and flat-bottomed hole not exceeding 3.2 mm (0.125 in.) at the center of the rosette and into the surface of the component being analyzed. Strain redistribution occurring at the surface in the surrounding area of the hole (resulting from the residual stress relief) is then measured with the previously installed strain gauges. The residual stress is calculated at a large number of points in a surface from the strain measurements using the well-established method, involving the magnitude and direction of strain, hole size, and materials properties.25,38 Layer removal can be effected by grinding (e.g., the outer layers of a cylinder can be gently removed in increments by grinding) and the strain change at each increment measured by strain gauges attached to the bore of the cylinder. To minimize the introduction of spurious strains by the grinding operation, the rate of metal removal should be less than 3.125 ¥ 10-4 m/s (1.23 ¥ 10-2 in./s), and readings are recorded after 15 min of the end of the grinding process to ensure that any heat generated has been dissipated.69 The sensitivity of the strain measurement and the accuracy of the stress calculation may be enhanced by employing the reverse-taper hole-drilling method.70 17.3.4.2 Nondestructive Methods. The method is described as nondestructive, but this will depend on what one is trying to achieve. If one wants to determine the surface stresses on a part, it is indeed nondestructive. If one wants to determine the residual stress gradient through a carburized case, it is destructive, i.e., unless it is executed in an area of the part where an incrementally produced groove is of no consequence to the fitness or purpose of the part.20 The main difficulty with the nondestructive methods is that measurements of crystallographic lattice parameters, ultrasonic velocities, or magnetization changes are made that are indirectly related to the residual stress. The above quantities are usually dependent on the stress and material parameters (such as metallurgical textures), which are difficult to quantify.68,71 The x-ray diffraction (XRD) method is probably the most well-established technique for measuring both macro- and microresidual stresses nondestructively. In most instances, the x-ray diffraction method has been employed to provide quantitative values for residual stress profiles in the surface or fully hardened components.72 This technique depends on the determination of lattice strains and the stress-induced differences in the lattice spacing. Macroresidual strain is measured from the shift of diffraction lines in the peak position using the so-called nonlinear Sin2 C method from which residual stress is calculated.72 For the measurement on microstrain the Voigt single-line method is applied.73 Precision in lattice strain measurement of the order of 0.2% is possible. XRD methods are capable of high spatial resolution, on the order of millimeters, and depth resolution, on the order of microns. The macroscopic residual stress and data related to the degree of cold working can be obtained simultaneously by XRD methods. XRD methods are applicable to most polycrystalline materials, metallic, or ceramic and are nondestructive at the sample surface.74 Possible sources of x-ray measurement errors are (1) grain size, (2) round surfaces (versus flat), (3) error in peak position, (4) stress relief due to aging, and (5) sample anisotropy.70 Portable x-ray diffraction equipment is now commercially available in various forms that allow stress measurements to be made very quickly (ranging from 4 to 30 s). The main drawbacks are that it cannot be applied to noncrystalline materials such as plastics, and it is only capable of measuring residual stresses of materials very close to the surface under examination. That is, the measurement is purely surface-related (a depth of 0.01 mm, or 0.4 mil, is commonly quoted).75

17.28

CHAPTER SEVENTEEN

Neutron radiography or diffraction, used for polycrystalline materials, has a much deeper penetration than x-rays, but has major safety problems and the disadvantage of being nonportable. Ultrasonic method for evaluating residual stress involves ultrasonic stress birefringence or sonoelasticity; this depends upon the linear variation of the velocities of sound in a body (i.e., ultrasonic waves) with the stress. This method has the potential for greater capability, versatility, and usefulness in the future.67,71 However, this has the disadvantage, in common with the magnetic methods, that it requires transducers shaped to match the surface being inspected.76 The magnetic method is based on the stress dependence of the Barkhausen noise amplitude. Each time an alternating magnetic field induced in a ferromagnetic material is reversed, it generates a burst of Barkhausen noise. The peak amplitude of the burst, as determined with an inductive coil near the surface of the component material, varies with the surface stress level. Since Barkhausen noise depends on composition, texture, and work hardening, it is necessary in each application to use calibrated standard (reference) samples with the same processing history and composition as the component being analyzed. This method is used to measure residual stresses well below the yield strength of the ferromagnetic material.† This method is rapid, and the measurements are made with the commercially available portable equipment. However, this method is limited to only ferromagnetic materials.71 Thermal evaluation for residual stress analysis (TERSA) is a new nondestructive method that is in an experimental stage. It has the advantage that it is completely independent, remote, and noncontacting. It consists of merely directing a controlled amount of energy from a laser energy source into the volume of the material being inspected and then making a precise determination of changes in the resulting temperature rise by infrared radiometry. However, the working instrument will also require some form of display to enable visual examination to be made of any highstressed regions.76

17.4 QUENCH CRACKING Anything that produces excessive tensile quenching stress is the basic cause of quench cracking, which may be considered as severe distortion. Quench cracking is mostly intergranular, and its formation may be related to some of the same factors that cause intergranular fracture in overheated and burned steels. Important contributors to cracking, apart from stress, in heat treatment are (1) part design, (2) steel grades, (3) part defects, (4) heat-treating practice, and (5) tempering practice.77 (See Sec. 13.11 for more details.) 1. Part Design. Features such as sharp corners; the number, location, and size of holes; deep keyways; splines; and abrupt change in section thickness within a part (i.e., badly unbalanced section) enhance the crack formation because while the one (thin) area is cooling quickly in the quenchant, the other (thick) area immediately adjacent to it is cooling very slowly. One solution to this problem is to change the material so that a less drastic quenchant (for example, oil) can be employed. An alternate solution is to prequench, that is, to cool it prior to the rest of the part. This will produce an interior of the hole or keyway that is residually stressed in com† Barkhausen noise property is also being developed to detect grinding burns in ground case hardened steels.20

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.29

pression, which is always desirable for better fatigue properties.77 The third solution is a design change, and the fourth is to use a milder quenchant. Tip cracking is a design-induced problem that begins with the gear designer. If the design cannot be modified, nor the steel changed, then it is up to the manufacturing people to find a way of avoiding tip cracking. Copper plating the blank is a good example. The Brown method of submerged induction hardening using a Delapena-type inductor, and employing hardened and tempered 4340 steel, and a good-quality quenching oil, very rarely encounters hardening problems.56,57 2. Steel Grades. Sometimes this can be checked by means of a spark test, whereas at other times a chemical analysis must be made. In general, the carbon content of plain carbon steels should not exceed the required level; otherwise, the risk of cracking will increase. The suggested average carbon contents for water, brine, and caustic quenching are given below: Method Induction hardening Furnace hardening

Shape

Carbon, %

Complex Simple Complex Simple Very simple, such as bar

0.33 0.50 0.30 0.35 0.40

A decrease in carbon content from 0.72 to 0.61% has been shown to slightly increase the thermal crack resistance of rim-quenched railroad wheels.78 Because of segregation of carbon and alloying elements, some steels are more prone than others to quench cracking. Among these steels, 4140, 4145, 4150, and 1345 appear to be the worst. A good option is to replace the 4100 series with the 8600 series. An additional disadvantage with the use of 1345 steels is the manganese floating effect, which leads to very high manganese content in the steel rolled from the last ingot in the same heat. Similarly, dirty (unclean) steels (that is, steels with more than 0.05% S, for example, AISI 1141 and 1144) are more susceptible to cracking than the low-sulfur grades. The reasons for this are that they are more segregated in alloying elements; the surface of this hot-rolled high-sulfur steel has a greater tendency to form MnS stringers and seams, which act as stress raisers during quenching; and they are usually coarse-grained (for better machinability), which increases brittleness and therefore promotes cracking.† If these high-sulfur grades are replaced by calcium-treated steels or cold-finished leaded steels, this problem can be obviated.77 3. Part Defects. Surface defect or weakness in the material may also cause cracking, for example, deep surface seams or nonmetallic stringers in both hot-rolled and cold-finished bars. Other defects are inclusions, stamp marks, and so forth. For largeseam depths, it is advisable to use turned bars or even magnetic particle inspection. The forging defects in small forgings, such as seams, laps, flash line, or shearing crack, as well as in heavy forgings, such as hydrogen flakes and internal ruptures, aggra† Cleanliness with respect to these steels is generally taken as oxide cleanliness. In this respect, it is usually (but not always) the case that machinable steels are not aluminum-treated (to ensure no abrasive alumina)—this is why they are generally described as coarse-grained steels.36

17.30

CHAPTER SEVENTEEN

vate cracking. Similarly, some casting defects, for example, in water-cooled castings, promote cracking.64 4. Heat-Treating Practice. Higher austenitizing temperatures produce a faster cooling rate during quenching and increase the tendency toward quench cracking. Similarly, steels with coarser grain size rate more prone to cracks than fine-grain steels because the latter possess more grain boundary area to stop the movements of cracks, and grain boundaries help to absorb and redistribute residual stresses. In other words, fine-grained steeels are tougher.36 An outstanding contributor to severe cracking is improper heat-treating practice, for example, nonuniform heating and nonuniform cooling of the component involved in the heat-treatment cycle. It is a good heat-treating practice to anneal alloy steels prior to the hardening treatment (or any other high-temperature treatment, for example, forging, welding, and so forth) because this produces a fine-grained microstructure and relieves stresses.79 Water-Hardening Steel. The water-hardening steels are most susceptible to cracks if they are not handled properly. Soft spots are most likely to occur in the water hardening steels, especially where the tool is grabbed with tongs for quenching. Normally the cleaned surface shows adequate hardening and the scaled surface insufficient hardening, which can be examined with a file. Soft spots may occur from the use of fresh water, or water contaminated with oil or soap. Most large tools emerging from hardening operations contain some soft spots. However, accidental soft spots in the wrong place should be investigated, and steps must be taken to eliminate them. Figure 17.15 shows the typical appearance of a thumbnail check as a soft spot on chipping chisels, which occurs on the bit near the cutting edge. The cracks enclosing the soft spots should be avoided by switching to brine quench.80 Oil-Hardening Steel. High-carbon low-alloy steels such as AISI 52100 (as well as carburized steels) are susceptible to microcrack during hardening. Lyman81 has developed a heat treatment cycle to eliminate this defect, which comprises quenching from the austenitizing temperature to a temperature just below the MS of the steel (sufficient to produce 30 to 40% martensite in the austenite) followed by transfer to a salt bath held at 260°C (500°F ) to temper that martensite,20 and final quenching to complete the formation of martensite.

FIGURE 17.15 Typical appearance of thumbnail check as a soft spot on chipping chisel.80 (Courtesy of Society of Manufacturing Engineering.)

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.31

Polymer and Salt Bath Quenching. Polymer quenchants have found wellestablished use in the quenching of solution-treated aluminum alloys, hardening of plain carbon steels with less than 0.6% C, spring steels, boron steels, hardenable stainless steels, all carburizing and alloy steels with section thickness greater than about 50 mm (2 in.), through-hardening steel parts, and induction and flame hardening treatments because of their numerous beneficial effects, including elimination of soft spots, distortion, and cracking problems associated with trace water contamination in quenching oils.82 Agitation is an important parameter in polymer quenching applications, both to ensure a uniform polymer film around the quench part and to provide a uniform heat extraction from the hot part to the adjacent area of quenchant, by preventing a buildup of heat in the quench region. Salt bath cooling of induction-hardened complex-shaped cast iron parts reduces the danger of cracking, which is usually experienced when air cooling followed by hot-water quenching is used.83 Air-Hardening Steel. Similarly, when air hardening steels are improperly handled, they are likely to crack. For example, avoidance of tempering treatment or use of oil quenching in air-hardening steel can lead to cracking. However, the common practice in the treatment of air-hardening steels is initially to quench in oil until “black” (about 540°C, or 1000°F), followed by air cooling to 65°C (150°F) prior to tempering. As compared to air cooling right from the quenching temperature, this practice is totally safe and minimizes the formation of scale. Decarburized Steel. Decarburization can occur at temperatures above about 700°C (1292°F) and usually arises from (1) drop of carbon potential of a furnace atmosphere below that necessary to maintain the carbon content of the surface;84 (2) insufficient protection as a result of plant failure (for example, leaks, air ingress, defective furnace or container seals, defective valves); (3) poor process control (for example, insufficient atmosphere-monitoring equipment, poor maintenance, and poor supervision); (4) incorrect diffuse stage of boost-diffuse carburization program; or (5) the existence of decarburizing agents such as CO2, water vapor, H2, and O2 in the furnace atmosphere.34,77,85 A partially decarburized surface on the part occurring during tool hardening also contributes to cracking because martensite transformation is completed therein well before the formation of martensite in the core. Significant decarburization causes inadequate surface microstructures and reduced residual compressive or even surface tensile residual stress. Decarburized surface on the tools has reduced hardness, which will lead to premature wear and scuffing. If surface carbon is >0.6%, the surface hardness should be acceptable. If surface carbon is £0.6%, all the important properties will be adversely affected; for example, bending fatigue limit could be reduced by 50%.34 Partial decarburization must be avoided, especially on all deep-hardening steels, by providing some type of protective atmosphere during the heating operation, stock removal by grinding, or a carbon restoration process. In addition to protective atmosphere, salt baths, inert packs, or vacuum furnaces may be used to obtain the desired surface chemistry on the tools or dies. The fact that the better and more consistent performance of the tools is observed after grinding reveals the existence of partial decarburization remaining. Other surface features to which this would apply include, for example, severe surface roughness and internal oxidation with associated high-temperature transformation products.20 Carburized Alloy Steel. Two types of peculiar cracking phenomena prevail in the carburized and hardened case of the carburized alloy steels: microcracking and tip cracking. Microcracking of quenched steels are small cracks appearing across or

17.32

FIGURE 17.16

CHAPTER SEVENTEEN

Microcracking in a Ni-Cr steel.85 (Courtesy of G. Parrish.)

alongside martensite plate (Fig. 17.16)85 and the prior austenite grain boundaries.86 They form mostly on those quenched steel parts that contain chromium and/or molybdenum as the major alloying elements with or without nickel content and where the hardening is done by direct quenching. Microcracks are observed mostly in coarse-grained structures and are associated with large martensite plates. This is presumably because of more impingements of the larger plates of martensite by other large plates. Another cause of microcracking is the increased carbon content of steel (exceeding eutectoid value such as AISI 52100) and that of martensite (i.e., increased hardenability), which is a function of austenitizing temperature and/or time.85 This finding was established for 8620H steel, which has a higher austenitizing temperature prior to quenching where there is a greater tendency to microcrack.87 This problem can be avoided by selecting a steel with less hardenability (i.e., with lower austenitizing temperature). Another solution is to change the heat-treating cycle to carburizing, slow cooling to black temperature, reheating to, say, 815 or 845°C (1500 or 1550°F), and quenching.77 Microcracking in case-hardened surfaces may be aggravated by the existence of hydrogen, which tends to absorb during carburizing and reheating in an endothermic atmosphere. However, this hydrogen-enhanced microcracking can be eliminated by tempering the carburized parts at 150°C (300°F) immediately after

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.33

quenching. Tempering exhibits an additional beneficial effect in that it has the ability to heal the smaller microcracks due to the volume changes and associated plastic flow that develop during the first stage of tempering.88 Apple and Krauss87 have observed that the reduction of microcracking leads to better fatigue resistance. On the other hand, Kar et al.89 have noticed an improvement in fracture resistance after elimination of both undissolved carbides and microcracks from an AISI 52100 steel microstructure. Tip cracking refers to the cracking that appears in the teeth of carburized and quenched (or nitrided) gears at the case/core interface; these cracks are crescentshaped and, when subjected to loading, progress in fatigue across the tooth section until breakage occurs. Many heat treaters have solved this problem to a great extent by decreasing the carbon content and case depth to the minimum acceptable design level, or by copper-plating the outer diameter of the gear blank prior to hobbing.83 Nitrided Steels. The nitrided cases are very brittle. Consequently, cracking may occur in service prior to realizing any improved wear and galling resistance. This can be avoided by a proper tool design, e.g., incorporating all section changes with a minimum radius of 3 mm (0.125 in.). Borided Steel. Monophase Fe2B layers are preferred to avoid brittleness, crack formation, and flaking of boride layer. Rapid heating to 900°C and the incorporation of diffusion anneal of 2 hr at 1000°C in argon during the pack boriding treatment tend to avoid the formation of brittle FeB and convert the FeB/Fe2B layer to monophase Fe2B layer, respectively.90 5. Tempering Practice. The longer the time the steel is kept at a temperature between room temperature and 100°C (212°F) after the complete transformation of martensite in the core, the more likely the occurrence of quench cracking. This arises from the volumetric expansion of retained austenite into martensite. There are two tempering practices that lead to cracking problems: tempering too soon after quenching, i.e., before the steel parts have transformed to martensite in hardening, and skin tempering, usually observed in heavy sections (≥50 mm, or 2 in. thick in plates and >75-mm, or 3-in., diameter in round bars). It is the normal practice to temper immediately after the quenching operations. In this case, some restraint must be exercised, especially for large sections (>75 mm, or 3 in.) in deep-hardening alloy steels. The reason is that the core has not yet completed its transformation to martensite which is accompanied by an expansion, whereas the surface and/or projections, such as flanges, begin to temper, which involves a shrinkage. These simultaneous, opposing volume changes produce radial cracks. This problem can become severe if rapid heating practice (e.g., induction, flame, lead, or molten salt bath) is used for tempering. Therefore, very large and very intricate deep-hardening alloy steel parts should be removed from the quenching medium, and tempering should be started while they are slightly warm to hold comfortably in the bare hands (~60°C, or 140°F). Skin tempering occurs in heavy section parts when the final hardness is >360 HB (or 39 Rc). This is due to insufficient tempering time and is usually determined when the surface hardness falls by 5 or more Rc points from the core hardness. This cracking often occurs several hours after the component has cooled from the tempering temperature and often runs through the entire cross section. This problem, however, can be removed by retempering for 3 hr at the original tempering temperature (provided no cracking is present), which is associated with a change in hardness of 2 Rc points maximum.77 Next time, the parts can be tempered for a longer time.

17.34

CHAPTER SEVENTEEN

17.5 DISTORTION IN HEAT TREATMENT Distortion can be defined as an irreversible and usually unpredictable dimensional change in the component during processing from heat treatment and from temperature variations and loading in service. The term dimensional change is used to denote changes in both size and shape.91 Distortion is therefore a general term often used by engineers to describe all irreversible dimensional change in a component as a result of heat treatment operations.92 Although it is recognized as one of the most difficult and troublesome problems confronting the heat treater and heat-treatment industries on a daily basis, it is only in the simplest thermal heattreatment methods that the mechanism of distortion is understood. Changes in size and shape of ferrous parts may be either reversible or irreversible. Reversible changes, which are produced by applying stress in the elastic range or by temperature variation, neither induce stresses above the elastic limit nor cause changes in the metallurgical structure. In this situation, the initial dimensional values can be restored to their original state of stress or temperature. Irreversible changes in size and shape of heat-treated parts are those that are caused by stresses in excess of the elastic limit or by changes in the metallurgical structure (e.g., phase changes). These dimensional changes sometimes can be corrected by mechanical processing to remove extra and unwanted material or by heat treatment (annealing, tempering, or cold treatment) to redistribute residual stresses. When heat-treated parts suffer from distortion beyond the permissible limits, it may lead to scrapping of the article, rendering it useless for the service for which it was intended, or it may require necessary correction. Allowable distortion limits vary to a large extent, depending on service applications; in cases where very little distortion can be tolerated, specially desired tool steels are used. These steels possess metallurgical characteristics that minimize distortion.

17.5.1 Types and Causes of Distortion Distortion can be classified into two categories: size distortion, which is the net change in specific volume between the parent and transformation product produced by phase transformation without a change in geometrical form, and shape distortion or warpage, which is a change in geometrical form or shape (of the workpiece) and is revealed by changes of curvature (angular relations) or curving, bending, twisting, and/or nonsymmetrical dimensional change without any volume change.92,93 The former may be considered a natural occurrence with some degree of predictability while the latter, being much less predictable, is a major processing concern. Usually both types of distortion occur during a heat-treatment cycle. The distortion or the dimensional changes may be attributed to numerous factors introduced into a component before, during, or even after heat treatment which are illustrated in Fig. 17.17.94 The important factors related to steel, machining, and heat treatment are described here. Steel-Related Factors. mation temperature.

These include as-cast shape, hardenability, and transfor-

1. As-cast factor on distortion is small compared to other factors. 2. Hardenability. The effect of hardenability on dimensional change may be explained by the change of transformation temperature. Once the hardenability

PRE-TREATMENT

MACHINING

MATERIAL

Tube

Feeds & Speeds

Part Orientation Pre Heat

Diam/Section Ratio

Fixturing

Form

Parameters

Support

Heating

Reduction Ratio Normalized

Rigidity Ramp Heat

Condition

Human Factor

Sub Critical Anneal

Chemistry

Hardenability

Handling

Pressure

Residuals

Tim

Chucking

Stress Relieve

Metallurgical

Hoo Count

Cutting Fluid

Vibratory Treatment

Tramp Elements

Stress Relieved Microstructure

3 Point

Type

Range

Temperature

Carbide Morphology

Wrap Around

Full Anneal

Dummy Carb

Forging

Tool Type

%S, Pb etc.

Stock Material

Grain Size

Cleanliness

Machinability

17.35

DISTORTION

Shot Blast Cleaning

Tempering

Marsizing

Die Quench

Surface Treatment

Surface Temp Swaco

Grinding

Peening

Chucking

Vibration

Bainitic

Circulation

Jacking

POST-HARDENING

Single/Double Q. Temperature/Time

Homogeneity

Equipment Stress Relaxation

Carburizing

Metallurgical

Turbulent/Laminar

Straightening

Uniformity

HARDENING

Velocity

Design

Pit/IQ Case Uniformity

Martemp Interrupted

Tank Design

Equipment

Gas

Cone Dies

Baffles/Draft Design

Induction

Parameters Quench Method

Pt. Design

Shot Peening

Process

Media Liquid

Fixture Coolant

Temp

Gleason

Parameters

Process

Volume

Atmosphere

Carbide Soln Circulation

AUSTENITIZING

FIGURE 17.17 Ishawaka diagram summarizing some of the many factors that should be considered when faced with a distortion problem.94

Type

17.36

CHAPTER SEVENTEEN

becomes adequate to produce a fully martensitic structure, the low rate of dimensional change arises from the relatively small effect of alloying elements on the Ms temperature. That is, hardenability control provides an essentially similar transformation in a component of a given size (shape and dimension) under constant heat treatment conditions.36 3. Transformation temperature plays an important role in dimensional change, thereby establishing the importance of hardenability control in minimizing variability. Since the hardenability is related to the consistent cooling rate, increasing the cooling rate will lead to the occurrence of transformations at lower temperatures. Thus cooling rates have dual roles of changing both thermal strains and transformation temperature.95 Machining-Related Factors. During machining, the distortion of a part is a function of geometry, order of material removed, and stress state in the material removed. If the change of shape which takes place is not accommodated, the part may be scrapped during machining. Measurement of the initial residual stress distribution and the use of finite element modeling enable the development of machining procedures which reduce or eliminate distortion. Distortion occurs due to removal of stressed material from the forging. Quenching stresses, when machined away, usually cause the largest amount of distortion, not necessarily as a result of the magnitude of the quenching stresses, but due to the large amount of material where quenching residual stresses are present. Turning and shot-peening stresses, although typically higher in magnitude than quenching stresses, are present in a shallow layer of near-surface material, and therefore, they exert less influence on distortion than quenching.95 Dimensional Changes Caused by Changes in Metallurgical Structure during Heat Treatment. Various dimensional changes produced by a change in metallurgical structure (or transformation stress) during the heat treatment cycle of steel parts are described below.96 1. Heating (austenitizing). When annealed steel is heated from room temperature, thermal expansion occurs continuously up to Ac1, where the steel contracts as it transforms from body-centered cubic (bcc) ferrite to face-centered cubic (fcc) austenite. The extent of decrease in volumetric contraction is related to the increased carbon content in the steel composition (Table 17.4). Further heating expands the newly formed austenite. 2. Hardening. When austenite is cooled quickly, martensite forms; at the intermediate cooling rates, bainite forms, and at slow cooling rates, pearlite precipitates. In all these transformation sequences, the magnitude of expansion increases with the decrease in carbon content in the austenite (Table 17.4). However, the net volume change for the ferrite through austenite to martensite condition is an increased expansion as the carbon content increases (assuming that all carbides are dissolved during austenitizing).20 The volume increase is maximum when austenite transforms to martensite, intermediate with lower bainite, and least with upper bainite and pearlite (Table 17.4). The volume increases associated with the transformation of austenite to martensite in 1 and 1.5% carbon steels are 4.1 and 3.84%, respectively; the volume increases involved in the transformation of austenite to pearlite in the same steels are 2.4 and 1.33%, respectively. Such volume increases are less in alloy steels and least in 2C-12Cr and A10 tool steels. Note

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.37

that plastic deformation (strain) occurs during such transformations at stresses that are lower than the yield stress for the phases present.97 The occurrence of this plastic deformation, called the transformation plasticity effect, influences the development of stresses during the hardening of steel parts.98 During quenching from the austenite range, the steel contracts until the Ms temperature is reached, then expands during martensitic transformation; finally, thermal contraction occurs on further cooling to room temperature. As the hardening temperature increases, a greater amount of carbide goes into solution; consequently, both the grain size and the amount of retained austenite are increased. This also increases the hardenability of steel. More trouble with distortion comes from the quenching or hardening operation than during heating for hardening, in which the faster the cooling rate (i.e., the more severe the quenching), the greater the danger of distortion. When the milder quenchants are used, the extent of distortion is lessened. The severity of quenching thus influences the distortion of components. The quench distortion remains a complex phenomenon involving numerous factors such as material, thermal, plastic, and elastic properties and phase transformations. The properties of the quenchants and the part/quenchant interface are also very significant and often difficult to characterize or quantify. This process, however, can be successfully predicted by computer simulation provided necessary input data are available and reasonably accurate.99 The dependence of volume increase, particularly in steel parts of different dimensions, on grain size (or hardenability) is another important factor. Variations in volume during quenching of a fine-grained shallow-hardening steel in all but small sections are less than for a coarse-grained deep-hardening steel of the same composition. 3. Tempering. There is a certain correlation among the tempering temperature, volume change, and its state of stress. The magnitude of the distortion during the tempering process depends on the steel composition and the austenitizing temperature. Tempering causes a continuous reduction in the state of stress and reduces the volume of martensite, but not adequately enough to equalize completely the prior volume increase as a result of martensitic transformation, unless the components are completely softened. Low-Alloy and Plain (Medium- and High-) Carbon Steels. During the first and third stage of tempering, a decrease in volume occurs that is associated with the decomposition of: high-carbon martensite into low-carbon martensite plus e-carbide in the former stage, and aggregate of low-carbon martensite and e-carbide into ferrite plus cementite in the latter stage. In the second stage, however, an increase in volume takes place (due to the decomposition of retained austenite into bainite) that tends to compensate for the early volume reduction. As the tempering temperature is increased further toward A1, more pronounced volume reduction occurs. High-Alloy Steels. In some highly alloyed tool-steel compositions, the volume changes during martensite formation are less striking because of the large proportion of retained austenite and the resistance to tempering of alloy-rich martensite. These hardened steels show sharp increases in both hardness and volume between 500 and 600°C (930 and 1100°F) owing to the precipitation of very finely dispersed alloy carbides from the retained austenite. This produces a depleted matrix in alloy content, raising the Ms temperature of retained austenite. During cooling down from the tempering temperature, further transformation of retained austenite into martensite will occur with an additional increase in volume.

CHAPTER SEVENTEEN

17.38

TABLE 17.8 Typical Volume Percentages of Microconstituents Existing in Four Different Tool Steels after Their Standard Hardening Treatments82

Hardening treatment

As-quenched hardness, HRC

Martensite, vol%

Retained austenite, vol%

Undissolved carbides, vol%

W1

790°C (1450°F), 30 min; WQ

67.0

88.5

9

2.5

L3

845°C (1550°F), 30 min; OQ

66.5

90

7

3.0

M2

1225°C (2235°F), 6 min; OQ

64

71.5

20

8.5

D2

1040°C (1900°F), 30 min; AC

62

45

40

Steel

15

Note: WQ, water quenched; OQ, oil quenched; AC, air cooled. Courtesy of ASM International.

17.5.1.1 Size Distortion. Size distortion is the result of a change in volume generated by a change in microstructural structure during heat treatment. For a given component and a given steel grade, heat-to-heat variations of hardenability affect the relative proportions of microstructural constituents, e.g., martensite, bainite, ferrite, and pearlite, with the consequence that the dimensions of the part will be altered by an amount relating to the predominant microstructure.20 Table 17.8 shows the typical volume percentages of microconstituents present in four different tool steels after their standard hardening treatments. Typical dimensional changes during hardening and tempering of several tool steels are given in Table 17.9. It is apparent that some steels such as M3 and M41 high-speed steels show appreciable increase in size of about 0.2% after hardening and tempering between 540 and 595°C (1000 and 1100°F) to produce complete secondary hardening. Other types, such as A10, expand very little when hardened and tempered over the entire temperature range up to 595°C (1100°F). Excessive size changes in oil-hardening nonshrinkable tool steel are usually caused by lack of stress relief (when necessary) and hardening and/or tempering at the incorrect temperature. The golden rule is to learn to be suspicious of tools that are seriously off-size in only one dimension. It is further noted that alloying addition in steel brings about a change in the specific volume of many microconstituents, but to a lesser extent than carbon.100 This table provides comparative data on size distortion in a variety of steels; however, this information cannot be used alone to predict the shape distortion factor. 17.5.1.2 Shape Distortion or Warpage. This is sometimes called straightness, or angularity change. It is found particularly in nonsymmetrical components during heat treatment. From the practical viewpoints, warpage in water- or oil-hardening steels is normally of greater magnitude than is size distortion and is more of a problem because it is usually not predictable. This is caused by the sum effect of more than one of these factors: 1. Rapid heating (or excessive heating), drastic (or careless) quenching, or nonuniform heating and cooling cause severe shape distortion. Slow heating as well as preheating of the parts prior to heating to the austenitizing temperature yields the most satisfactory result. Rapid quenching produces thermal and

TABLE 17.9 Typical Dimensional Changes during Hardening and Tempering of Several Tool Steels82 Hardening treatment Tool steel

17.39

O1 O1 O6 A2 A10 D2 D3 D4 D5 H11 H13 M2 M41

°C

°F

Quenching medium

Total change in linear dimensions after quenching, %

815 790 790 955 790 1010 955 1040 1010 1010 1010 1210 1210

1500 1450 1450 1750 1450 1850 1750 1900 1850 1850 1850 2210 2210

Oil Oil Oil Air Air Air Oil Air Air Air Air Oil Oil

0.22 0.18 0.12 0.09 0.04 0.06 0.07 0.07 0.07 0.11 -0.01 -0.02 -0.16

Temperature

Courtesy of ASM International.

Total change in linear dimensions, %, after tempering at 150°C 300°F

205°C 400°F

260°C 500°F

315°C 600°F

370°C 700°F

425°C 800°F

480°C 900°F

510°C 950°F

540°C 1000°F

565°C 1050°F

595°C 1100°F

0.17 0.09 0.07 0.06 0.00 0.03 0.04 0.03 0.03 0.06 ... ... ...

0.16 0.12 0.10 0.06 0.00 0.03 0.02 0.01 0.02 0.07 ... ... ...

0.18 0.13 0.14 0.08 0.08 0.02 0.01 -0.01 0.01 0.08 ... ... ...

... ... 0.10 0.07 0.08 0.00 -0.02 -0.03 0.00 0.08 ... ... ...

... ... 0.00 ... 0.01 ... ... ... ... ... ... ... ...

... ... -0.05 0.05 0.01 -0.01 ... -0.4 0.3 0.3 ... ... ...

... ... -0.06 0.04 0.02 -0.02 ... -0.03 0.03 0.01 0.00 ... ...

... ... ... ... ... ... ... ... ... ... ... -0.06 -0.17

... ... -0.07 0.06 0.01 0.06 ... 0.05 0.05 0.12 0.06 0.10 0.08

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

... ... ... ... 0.02 ... ... ... ... ... ... 0.16 0.23

... ... ... ... ... ... 0.14 0.21

17.40

CHAPTER SEVENTEEN

mechanical stresses associated with the martensitic transformation. In the case of low- and high-hardenability steels, respectively, this problem becomes severe or very small. 2. Nonsymmetrical carburizing may also contribute to distortion.95 3. Residual stresses are present in the component before heat treatment. These arise from machining, grinding, straightening, welding, casting, spinning, forging, and rolling operations, which will also furnish a marked contribution to the shape change.101 4. Residual stresses developed during heat treatment are caused by: (1) thermal gradients within the metal, (2) nonuniform changes in the metallurgical structure, and (3) nonuniformity in the composition of the metal itself, such as that caused by segregation.102 5. Applied stress causes plastic deformation. Sagging and creep of the components occur during heat treatment as a result of improper support of components or warped hearth in the hardening furnace. Hence, large, long, and complex-shaped parts must be properly supported at critical positions to avoid sagging or preferably are hung with the long axis on the vertical. 6. Nonuniform agitation/quenching or nonuniform circulation of quenchant around a part results in an assortment of cooling rates that creates shape distortion.103 Uneven hardening, with the formation of soft spots, increases warpage. Similarly, an increase in case depth, particularly uneven case depth in case hardening steels, increases warpage on quenching.104 7. There is tight (or thin and highly adherent) scale and decarburization, at least in certain areas. Tight scale is usually a problem encountered in forgings hardened from direct-fired gas furnaces having high-pressure burners. Quenching in areas with tight scale is extremely retarded compared to that in the areas where the scale comes off. This produces soft spots, and in some cases, severe unpredicted distortion. Some heat treaters coat the components with a scale-loosening chemical prior to their entry into the furnace.103 Similarly, the areas beneath the decarburized surface do not harden as completely as the areas below the nondecarburized surface. The decarburized layer also varies in depth and produces an inconsistent softer region compared to the region with full carbon. All these factors can cause a condition of unbalanced stresses with resultant distortion.103 8. There are long parts with small cross sections (>L = 5d for water quenching, >L = 8d for oil quenching, and >L = 10d for austempering, where L is the length of the part and d is its diameter or thickness). 9. There are thin parts with large areas (>A = 50t, where A is the area of the part and t is its thickness). 10. There is unevenness of, or greater variation in, section. Figure 17.7 summarizes a large number of possible contributory factors.

17.5.2 Examples of Distortion 1. Ring die. Quenching of ring die through the bore produces the reduction in bore diameter as a result of formation of martensite, associated with the increased volume. In other words, metal in the bore is upset by shrinkage of the surrounding metal and is short when it cools.24 However, all-over quenching causes the outside

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.41

diameter to increase and the bore diameter to increase or decrease, depending upon the precise dimensions of the part. When the outside diameter of the steel part is induction- or flame-hardened (with water quench), it causes the part to shrink in outer diameter.79 These are the examples of the effect of mode of quenching on distortion.105 2. Thin die. A thin die, with respect to wall thickness, is likely to increase in bore diameter, decrease in outside diameter, and decrease in thickness when the faces are hardened. If the die has a very small hole, insufficient quenching of the bore may enlarge the hole diameter because the body of the die moves with the outside hardened portion. 3. Bore of a finished gear. Similarly, the bore might turn oval or change to such an extent that the shaft cannot be fitted by the allowances that have been provided. Even a simple shape such as a diaphragm or orifice plate may, after heat treatment, lose its flatness in such a way that it may become unusable. 4. Production of long pins. In the case of the production of long pins (250 mm long ¥ 6-mm diameter, or 10 ¥ 1/4 in.) made from medium-alloy steel, it was found, after conventional hardening, that when the pins were mounted between centers, the maximum swing was over 5 mm (0.20 in.). However, the camber could be reduced to within acceptable limits by martempering, intense or press quenching. 5. Hardening and annealing of long bar. When a 1% carbon steel bar, 300 mm long (or more) ¥ 25-mm diameter (12 in long or more ¥ 1-in. diameter), is water quenched vertically from 780°C (1435°F), the bar increases in both diameter and volume but decreases in length. When such bars are annealed or austenitized, they will sag badly between the widely spaced supports. Hence, they should be supported along their entire length in order to avoid distortion. 6. Hardening of half-round files. Files are usually made from hypereutectoid steel containing 0.5% chromium. Files are heated to 760°C (1400°F) in an electric furnace after being surface coated with powdered wheat, charcoal, and ferrocyanide to prevent decarburization. They are then quenched vertically in a water tank. On their removal from the tank, the files appear like the proverbial dog’s tail. The flat side has curved down, the camber becomes excessive, and the files can no longer be used in service. One practical solution is to give the files a reverse camber prior to quenching. The dead flat files could, however, be made possible, and the judgment with regard to the actual camber needed depends upon the length and the slenderness of the recut files.106 Similarly, when a long, slender shear knife is heat-treated, it tends to curve as a dog’s tail, unless special precautions are taken. 7. Hardening of chisels.79 Chisels about 460 mm (18 in.) long and made from 13mm (0.5-in.) AISI 6150 bar steel are austenitized at 900°C (1650°F) for 1.5 hr and quenched in oil at 180°C (360°F) by standing in the vertical position with chisel point down in special baskets that allow stacking of two 13-mm (0.5-in.) round chisels per 650 mm2 (1 in.2) hole. Subsequently, hardened chisels are tempered between 205 and 215°C (400 and 420°F) for 1.5 hr. These heat-treated parts show 55 to 57 Rc hardness but are warped. The reasons for this distortion are that • The portion of the bar that touches the basket cools slowly, producing uneven contraction and thermal stress. • The martensite formation is delayed on the inner or abutting side of the bar, causing unequal expansion during transformation. This distortion can be elimi-

17.42

CHAPTER SEVENTEEN

nated or minimized by loading the parts in the screen-basket in such a way that the stacking arrangement permits sufficient space between each part and by slightly decreasing the austenitizing temperature.78 Distortion can also be minimized by austempering the part, provided that the carbon content is on the high side of specification, to produce the lower bainitic structure of 55 to 57 Rc. If higher yield stress is not warranted, only chisel ends need hardening and subsequent tempering.79 8. Hardening of a two-pounder shot. The hardness of a two-pounder shot was specified at 60 Rc on the nose and 35 Rc at the base. A differential hardening technique was performed on the shot made of a Ni-Cr-Mo steel. This technique consisted of quenching the shot in the ice-cold water by its immersion in a tank up to the shoulder, followed by drawing out the water from the tank at a stipulated rate until the waterline reached the base of the nose. The final step involved withdrawing the shot from the tank when completely cold. The back end was then softened by heating in a lead bath after initial tempering. The first few shots hardened in this way were observed to split vertically across the nose. The failure was, however, avoided by withdrawal of the shot before attaining ice-cold temperature and its subsequent immersion in warm water.106 9. Hardening of a burnishing wheel. In the manufacture of railway axles, the gearing surface on which the axle rests in the housing has to be given a high burnishing polish, employing a circular pressure tool that is made of 1.2C-01.5Cr steel. For satisfactory results, the hardness of the tool surface should be about 60 Rc. It has been found that the tool usually cracks before its withdrawal from the coldwater quenching bath. This problem may, however, be avoided by quenching the tool in water for 10 s prior to transferring it to an oil bath for finish quenching. Time quenching can be judiciously applied for many heat treatment problems of distortion or cracking. Stress-relieving treatment after the use of the tool for some time may also enhance its performance life. As indicated above, martempering is also one of the solutions for this problem.105 10. Hardening of case-carburized mild steel. If oil-hardening steels are not available for making a component, mild steel parts are carburized and water quenched to obtain the desired hardness, possibly resulting in excessive distortion. Press straightening of case hardened shaft is fairly common. Shaft lengthto-diameter ratio is important (short, fat shafts are difficult; long slender shafts are easier). Straightening depends very much on the yield strength of the core; the core is caused to yield while the case remains elastic. Therefore, a ferritic core (mild steel) should lend itself to straightening better than a bainitic or martensitic core (alloy steel). Some heat treaters prefer to straighten after quenching and before tempering, because tempering raises the yield strength of the hardened material.20 11. Hardening of carburized low-carbon steel rollers. The best course of quenching carburized En32 steel rollers (25-mm dia. ¥ ≥600 mm long, or 1-in. dia. ¥ ≥2 ft long), employed in textile printing, is to roll them down skids into water quenching tanks because this produces less warpage than when quenched slowly with the bar either in vertical, horizontal, or inclined positions. These are the procedures adopted for hardening of cylinders with length considerably greater than the diameter. 12. Hardening of helix gears. The distortion of the helical gears made of IS 20MnCr1 Grade steel (similar to AISI 5120) used as the third speed gear in the

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.43

gear box of Tata trucks is an unavoidable natural consequence of the hardening process after carburizing. This type of distortion is linked with increased length and decreased diameter and occasionally increased helical angle.107 If the extent of distortion can be controlled, a constant correction to the helix angle can be imparted in the soft-stage manufacturing (machining) prior to heat treatment, so that this correction can compensate for the distorted angle and may result in a gear with the desired helix angle. Thus a constant magnitude of distortion without minimization is assured in every job of every batch of production in commercial manufacturing. However, the residual stress system and metallurgical properties such as core strength, case depth, surface hardness, proper microhardness in the surface regions, and so forth are assured.108 Similarly, when heavy-duty tooth gear is gas carburized and quenched to harden the surface layer, the diameter and tooth span increase and tapering and bending also occur. 13. Nitriding. A rolling mill screw, after liquid nitriding, may also show a small decrease in length, which causes pitch errors in the screw.107 14. Induction and flame hardening of spur gears. Spur gears, after induction and flame hardening, exhibit increased circular pitch, the error being maximum for the tooth groove quenched first. Similarly, in line-heating process, the thin plate undergoes convex bending; and the thick plate, concave bending.107 15. Head hardened rail. The rail head is hardened for its increased abrasion resistance. This is done by two methods. One is through heating of the rail followed by spray quenching the rail head only. In this case, the quench distortion of the rail tends to be head convex type (Fig. 17.18a), due to faster cooling of rail head compared to the rail base. The second process involves induction heating of the rail head followed by spray quenching. Here, the quench distortion tends to be a head concave type

FIGURE 17.18 Quench distortion of head hardened rail.109 (Courtesy of ASM International, Materials Park, Ohio.)

17.44

CHAPTER SEVENTEEN

(Fig. 17.18b), due to the slower cooling of the rail head compared to the rail base, even though the rail head is water spray quenched.109 17.5.3 Precautions 1. Inadequate support during the heat treatment cycle, poorly designed jigs and quenching fixtures, or incorrect loading of the parts may cause distortion.93 In general, plain carbon and low-alloy steels have such a low yield strength at the hardening temperature that the parts are capable of distorting under their own weight. Every care, therefore, must be taken to ensure that parts are carefully supported or suspended during heating. Long parts are preferably heated in a vertical furnace or with the length in the vertical plane.110 They should be quenched in the vertical position with vertical agitation of the quenchants. Also, it must be remembered that many tool steels are spoiled by failure to provide enough support when they are taken out of the furnace for quenching. Thus every precaution is taken to ensure that parts are adequately supported during the entire heat treatment by employing well-designed jigs, fixtures, and so on. 2. Large parts must be raised off the hearth plate to ensure adequate heat circulation and more even heating and cooling. Care must be taken in transferring the load. Preferably, the parts should be placed on trays that can be grasped to remove the load. If the individual part must be handled by tongs, avoid holding it at the thinner section, which will lose heat rapidly and might bend more easily.102 3. Tool steels should be heated to hardening temperature slowly, or in steps, and uniformly. Hot salt baths are used to render fast, uniform heat input. 4. It is best to heat small sections to the lower region of the recommended hardening temperature range and to heat large sections to the higher temperature range. Overheating by employing too high a temperature or too long a heating time must be avoided. 5. It is a good practice to protect the surface of the component from decarburization by (packing in) cast iron chips, controlled furnace atmosphere, or using a vacuum furnace. If a separate preheating furnace is not available, the part can be put in a cold furnace, after which the temperature is raised to proper preheating temperature and kept at that temperature to attain uniform heating throughout, prior to proceeding to the hardening temperature.111 6. With the slower cooling rate, which is consistent with good hardening practice, a lower thermal gradient will be developed, thereby producing less distortion. (Note: Slower cooling rate is possible with steels having adequate hardenability to give required microstructure and hardness.20) 7. Thus rapid heating and cooling rates of irregularly shaped parts must be avoided. 8. Proper selection of quenchant with desirable quenching properties and adequate agitation during hardening must be provided. 9. Recent innovations in quench tank designs are oscillating quench elevators, reversible and variable. Oil circulation, ultrasonic agitation, very high-flowrate quenching, and flood quenching minimize and ensure uniform breakdown of vapor phase and provide better uniformity and uniform circulation of quench media.94 10. The ideal method of cooling is a spray quenching; especially computer-aided quenching (CAQ) is more promising.109

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.45

17.5.4 Methods of Preventing Distortion106,112 Currently, there are various methods of prevention of distortion (or warping) that are employed in the industry: straightening, support and restraint fixtures, quenching fixtures, pressure quenching, press quenching, rolling die quenching, and stressrelieving. These are described below. 1. Straightening. Straightening is one method to remove or minimize distortion. Since straightening (after hardening) can largely relieve the desirable residual compressive stresses (in plain-carbon and low-alloy steels) that may cause breakage, it would be better to accomplish this before the steel cools below the Ms temperature, i.e., when the steel is in the metastable austenitic state.45 This temperature is above 260°C (500°F) for most tool steels and is preferably about 400°C (750°F) for long shear knives, which are usually made of 2C-12Cr steel. Warping on parts such as shafts and spindles can be corrected by straightening during or after hardening, followed by grinding to size.108 Mostly high-alloy steels are straightened after hardening due to the higher percentage of retained austenite and their comparatively low yield stress. Straightening also can be accomplished during the tempering process.45 However, straightening of hardened parts with higher strength will cause a loss of fatigue properties and possibly initiation of cracks at the surface. Hence, straightening after the hardening treatment must be very carefully controlled and should be followed by a low-temperature tempering treatment. The case-hardened (e.g., nitrided, carburized) parts can be straightened to a very large extent as a result of their lower core hardness. Nitrided parts may be straightened at 400°C (750°F).45 As a rule, the parts subjected to straightening after heat treatment are likely to be distorted at subsequent stages of machining.113 2. Support and restraint fixtures. Fixtures for holding finished parts or assemblies during heat treatment may be either the support or the restraint type. For alloys that are subjected to very rapid cooling from the solution-treatment temperature, it is common practice to use minimum fixturing during solution treatment and to control dimensional relations by using restraining fixture during aging. Support fixtures are used when restraint type is not needed or when the part itself renders adequate self-restraint. Long, narrow parts are very easily fixtured by hanging vertically. Asymmetrical parts may be supported by placing them on a tray of sand or a ceramic casting formed to the shape of the part.80 Restraint fixtures may require machined grooves, plugs, or clamps. Some straightening of parts can be accomplished in aging fixtures by forcing and clamping slightly distorted parts into the fixture. The threaded fasteners for clamping should not be used because they are difficult to remove after heat treatment. It is preferable to use a slotted bar held in place by a wedge.80 The bore of a hub, the most important dimension in the hardening of thin spur gears, can be mechanically plugged to prevent the reduction of the bore and keep the out-of-roundness close to tolerance limits. When hardening large hollows, either restraining bands on the outside during tempering or articulated fillers serve the same purpose. 3. Quenching fixtures. When water quenching or oil quenching is essential, distortion can be minimal by employing properly designed quenching fixtures that forcibly prevent the steel from distortion.114 Figure 17.19 shows a typical impingement-type quenching fixture. The requirements essential for the better design of this type of fixture are as follows:103

17.46

CHAPTER SEVENTEEN

FIGURE 17.19 A typical impingement-type quenching fixture. (Reprinted by permission of Fairchild Publications, New York.)

a. There must be an accurate positioning of the part in the fixture. Whenever possible, round bars should be rotated during quenching to level out variations in jet pressure around the part. b. There should be an unhindered flow of quenchant through the sufficiently large holes (3.3 to 6.4 mm, or 0.13 to 0.25 in., in diameter). Jets as large as 12.25 mm (0.50 in.) in diameter may be employed with furnace-heated heavy sections (e.g., plates). A large portion of the excess quenchant with these large jets is for the removal of scale.115 c. Spacing between the holes should be reasonably wide (for example, 4d, where d is the hole diameter). d. For oil-quenching fixtures, the facility to submerge the part is required to reduce fumes and flashing. e. There must be the provision for efficient cleaning of the holes. f. A facility must be available to drain out the hot quenchant for effective quenching performance with cold quenchant. 4. High-pressure gas quenching. High-pressure gas quenching, using helium or hydrogen up to 20 bar, is claimed to approach the efficiency of oil and reduce distortion. However, pressures up to 40 bar are being considered to provide quench rates equivalent to or faster than those of oil.95 This is economical and fast, provides even cooling, and offers a unique design, minimum distortion, and improved metallurgical qualities. As a result of these beneficial effects, this is suited to quench large-diameter tooling for the aluminum extrusion industry; quench large-diameter carburized gear, large fasteners, and precision gears to be jigged vertically; harden high-speed steel tools (such as saw blades, dies, and other parts with edge configuration) and 718 jet engine compressor blades.116 This is also employed to quench (vacuum processed) large sections of titanium alloy castings for aircraft applications.117 Figure 17.20 is a pressure-quench

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.47

FIGURE 17.20 Pressure-quench module for attachment into standard vacuum-sealed quenched and continuous vacuum furnaces.116 (Courtesy of Hayes, Inc.)

module that may be attached to vacuum-sealed quenched and continuousvacuum furnace as a replacement for the oil quench section.116 (See Sec. 13.2.3.7 for more details.) High-pressure gas quenching coupled with vacuum heat treatments is commercially used. These facilities may be expensive, but they minimize or eliminate post-heat treatment corrective machining.95 5. Press quenching. Press quenching is widely employed in preventing and controlling quench distortion in components whose geometry renders them particularly prone to distortion.118 Although press quenching maintains the geometry (shape, flatness, etc.), “uniform” dimensional changes still occur. Press quenches usually pulsate the restraining tools—to allow the component to move (e.g., contract) during quenching.36 Press quenching is used on bevel and spiral bevel gear wheels, large crown wheels, and annular gears.20,36 Flat circular diaphragms of spring steel used in the control or measurement of pressure are press quenched between two copper blocks, which cannot be accomplished by direct quenching.104 6. Rolling die quenching. A rolling die quench machine can provide uniform water quenching with minimal distortion for large-production runs. When a heated part is placed on the rollers, the die closes and the rolls turn. This removes any distortion incurred during heating. According to manufacturers of rolling die quench machines, symmetrical parts with the following straightness can be achieved in production: l TIR = K Ê ˆ Ë d¯

(17.2)

where TIR is the total indicator reading of straightness, l is the length (in.), d is the diameter (in.), and K is a constant = 10-4.

17.48

CHAPTER SEVENTEEN

For minimum yield strength requirements of 310 MPa (45 ksi), air hardened or normalized parts with negligible distortion can be produced.103 7. Stress relieving. The presence of residual stresses in the parts caused by coldworking, drawing, extrusion, forging, welding, machining, heading operations, or air cooling following normalizing greatly increases the tendency of distortion. However, these residual stresses can be relieved by subcritical annealing or normalizing treatment just before the final machining operation, which decreases the distortion to an appreciable extent. This is of special importance for intricate parts with close dimensional tolerances.104 Stress reduction is necessary to avoid distortion during hardening and to avoid cracking resulting from the combination of residual stress to the thermal stress produced during heating to the hardening temperature. In the event that stress relieving is not performed after heat treatment, large distortions of the part can be removed by heavy grinding. However, the drawbacks of this operation are possible elimination of most, if not all, of the hardened case of the carburized and hardened part; and danger of burning and crack formation on the surface layers. Hence, it is customary to stress-relieve plain carbon or low-alloy steel parts at a temperature of 550 to 650°C (1020 to 1200°F) (for 1 to 2 hr), hot-worked and high-speed steels at 600 to 750°C (1110 to 1380°F), and the heavily machined or large parts at 650°C (1200°F) (for 4 hr) prior to final machining and heattreatment operations. Subresonant stress relieving may also be employed to neutralize thermally induced stress without changing the mechanical properties or the shape of the component. These components include: large workpieces, premachined or finishmachined structural or tubular, nonferrous, hardened, nonsymmetrical or varying section thickness, stationary, or assembled. However, this does not work on copper-rich alloys and the edges of burned plates.119 Distortion due to residual stress that occurs during nitriding can be minimized or eliminated by previously stress-relieving the workpiece at a sufficiently high temperature.120 The stress relief technique has been found to be ineffective for heat-treating asymmetrical parts which require adequate support at critical locations to avoid sagging, or preferably are hung with the long axis in the vertical position.121 However, fulfilling these requirements cannot stop distortion, because the main reason of distortion, i.e., high plastic deformation of the asymmetrical component after hardening, is not removed. Consequently, straightening is needed. Straightening, however, introduces regions of high stress concentrations, which may initiate development of surface and initial cracks.113 17.5.5 Control of Distortion In order to remove or minimize distortion, the modern trend is to shift from water quenching practice to milder quenching (if the steel in question has adequate hardenability), for example, oil quenching, polymer quenching, martempering, austempering, or even air-hardening practice. One may also need to change the grade of steel. Milder quenchants produce slower and more uniform cooling of the parts, which drastically reduces the potential distortion. Other strategies of controlling distortion for age-hardening aluminum, beryllium, and other alloys include: alloy and temper selection, fixturing, age hardening temperatures, proper machining, and stamping operations.121

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.49

The fewer the number of reheats applied to components in case-hardening steels following carburizing, the smaller the distortion on the finished part. When top priority is given to minimum distortion, it is desirable to make the parts from oilhardening steels with a controlled grain size and to harden them by martempering direct from carburizing. Currently, polyalkylene glycol-base quenchants, such as UCON quenchants HT and HT-NN, are variously used for direct quenching from the forging treatment, continuous cast quenching, and usual hardening of forged and cast steels and cast iron. In this case, boiling does not take place at the component surface but rather at the external surface of the deposited polymer film. More uniform cooling occurs, and thermal stresses are released. Because of the lower boiling point and high thermal conductivity, UCON quenchants act through the martensite zone more rapidly than oil.122 Distortion during ferritic nitrocarburizing is minimal because of low treatment temperature and the absence of subsequent phase transformations.70 For many applications the distortion due to induction hardening is small and acceptable, requiring no special measures to control or correct. However, the shape and the ratio of hardened to unhardened thickness are important factors.20 There are many methods of reducing distortion in induction-hardened components; these methods are usually found by experience with variables such as the hardening temperature and the type and temperature of quenching medium employed. Methods of reducing distortion in induction hardened parts include: the hardening of small spindles held vertically in jigs; the plug-quenching of gears to prevent the bores from closing in; the flattening of cams by clamping them together during tempering; and the selective hardening of complex shapes.123 As a replacement of medium- or slow-quenching oils, UCON quenchants E and E-NN can be readily used in induction- and flame-hardening operations, in both spray and immersion types, for high-carbon and most alloy steels and traditional hardening of cast iron and cast or forged steels of complex geometry with better distortion-reduction properties. Agitation of quenchant should be carried out by motor-driven stirrers to move the medium with respect to the part being quenched or by pumps that force the medium through the appropriate orifice. Alternatively, the parts are moved through the medium, and for some applications, spray quenchant is recommended. Water additives are sometimes employed in salt baths to increase heat extraction.80 Ultrasonic quenching is also effective in controlling distortion, which involves the introduction of ultrasonic energy (waves with a frequency of 25 kHz) in the quenching bath. This breaks down the vapor film that surrounds the part in the initial stages of water or oil quenching.111 17.5.6

Distortion after Heat Treatment

Straightening. When every possible case has been employed to minimize distortion, it may still be essential to straighten after heat treatment, which has already been discussed. Grinding after Heat Treatment. In the case of carburized or nitrided parts, the metallurgist and designer, together with the production engineer, must collaborate regarding the amount to be removed by grinding after heat treatment. This grinding allowance must be taken into account when determining the initial dimensions and when specification for the case depth is to be applied.

17.50

CHAPTER SEVENTEEN

Distortion may also occur after heat treatment, with time, owing to the completion of any unfinished transformation or the effect of increased temperature during grinding. For example, fully hardened components such as blade shears may be damaged by characteristic crazing pattern because of heavy and careless grinding. Local overheating results in the transformation of undecomposed austenite, and the accompanying changes in volume produce sufficient stresses to cause cracking and development of a crazed pattern. Dimensional Stability. To achieve dimensional stabilization or stability† (i.e., retention of exact size and shape) over long periods, which is a vital requirement for gauges and test blocks, the amount of retained austenite in heat-treated parts must be reduced because retained austenite slowly transforms and produces distortion when the material is kept at room temperature, heated, or subjected to stress. Dimensional stabilization also reduces internal (residual) stress, which causes distortion in service. The excessive retained austenite in steels directly quenched after carburizing arising from high case carbon content, high substitutional alloy content, high carburizing temperature, or geometry may be reduced by several methods, such as tempering, refrigeration or subzero treatment, intercritical temperature reheating, and shot-peening (Fig. 17.21). Process selection is dependent on the required performance level of the carburized part.32 Multiple tempering (with prolonged tempering times) is needed to achieve stabilization for, say, high-speed steel tools. However this is not essential for, say, case hardened gears made from conventional steels. The first tempering reduces internal stress and facilitates its transformation to martensite on cooling. The second and third retempering reduce the internal stress produced during the transformation of retained austenite. It is the usual practice to carry out a single or repeated cold treatment after the initial tempering treatment. In cold treatment, the part is cooled below Mf, which will cause the retained austenite to transform to martensite; the extent of transformation depends on whether the tool part is untempered or first tempered. Cold treatment is normally accomplished in a refrigerator at a temperature of -70 to 120°C (-100 to -184°F). Tools must be retempered immediately after return to room temperature following cold treatment in order to reduce internal stress and increase the toughness of the fresh martensite. Finally, they are ground to size. Subzero treatment should be employed judiciously, and according to Parrish and Harper, considered as a last resort for case hardened parts, especially commercial gears. Reduced fracture and bending fatigue resistance have been noted in refrigerated-treated carburized steels.21 If material selection, carbon gradient, and quenching are right, the retained austenite content and surface hardness will each be acceptable. Hence, there is no need to subzero-treat. The need to use refrigeration on a regular basis could be a sign that something is not quite right; e.g., surface carbon control is suspect, or target surface carbon is high. The heat treater should endeavor to get the processing right rather than resorting to the easier option of refrigeration. On the other hand, there may be a good reason for running with a high carbon for a given application.20 A recent innovation in subzero treatments avoids thermal tempering and replaces it with subzero treatment in the presence of a cyclic magnetic field. Rota†

For full dimensional stability, the residual stresses in a body should be zero.20

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.51

FIGURE 17.21 Retained austenite as a function of distance from the surface of a carburized SAE 4320 in the as-carburized condition and after various shot-peening treatments.32,33 (Courtesy of ASM International, Materials Park, Ohio.)

tion beam test results suggest that with this treatment, fatigue lives comparable to those of conventionally tempered parts could be achieved.34 Vibratory stress-relieving process (VSRP) has been claimed to provide the ultimate in precision and long-term dimensional stability and to serve as an effective alternative to thermal treatment for component stabilization in large castings and fabrications. However, it does not offer any metallurgical benefits such as driving off hydrogen and modification of microstructure.20,124

17.5.7

Distortion and Its Control in Heat-Treated Aluminum Alloys

The high levels of residual stress and distortion that are produced in the waterquenched aluminum extrusion and forgings (such as 2000, 6000, and 7000 series) and aluminum castings can be reduced 60 to 100% by using proper selection of polyalkylene glycol quenchant or polyvinylpyrrolidone 90 concentration [for example, 25% solutions for wrought alloys, 20 to 30% UCON quenchant A for thicknesses up to 25 mm (1 in.), and 17 to 22% for larger than 25-mm (1-in.), section thickness in casting alloys] with sufficient agitation, lower bath temperature, proper fixture (throughout solutionizing, quenching, and age hardening treatments), and straightening (in the as-quenched state after taking out from the fixture) procedure. The initial cost of these polymer solutions as a replacement to the conventional hot-water quenching method is easily compensated for by other advantages, such as reduced scrap, reduced machining (compared to two machining operations

17.52

CHAPTER SEVENTEEN

required—one before and another after heat treatment—in the conventional waterquenching method), and increased fatigue life as a result of reduced convective heat transfer or film coefficient between the part and the quenchant, more uniform quench, precise control of quench rates, and improved heat-transfer qualities from the deposition of liquid organic polymer on the surface of the part being quenched.125–127 This method costs less, therefore saves time, and allows easy shaping, bending, and twisting of the parts without establishing residual stresses. Such parts as leading-edge wing skins, spars, and bulkheads are used in the aerospace industries.123

17.6 IMPORTANCE OF DESIGN The wrong design of the component may result in the establishment of nonuniform heating and cooling of the components, which produces overload and/or internal stresses, leading to distortion and failure during or after hardening. Correct consideration at the design stage plays an important role in lessening the distortion and danger of cracking. The basic principle of successful design is to plan shapes that will minimize the temperature gradient through the part during quenching. Fundamental rules such as maintaining a simple, uniform, regular, and symmetrical section with comparatively few shape changes; ensuring small and smooth cross-sectional size changes; and using large radii are still too frequently overlooked at the design stage. Thus, successful heat treatment demands a rational design that avoids sharp corners as well as sudden and undue changes of section. It is often possible for tool designers to compensate for size distortion. For example, in preparing precision hobs for gear cutting, dimensional accuracy must be kept within very close tolerances. On linear longitudinal growth, it is the general practice to go out-of-round in the following high-speed steel bars as much as 0.3% in M1 type, 0.2% in M2 type, and 0.15% in T1 type during heat treatment. These data will alter slightly with changes in the design of the hobs, but essentially the growth in tungsten-base high-speed steel is lower than that of the molybdenum-base high-speed steel (M1 and M2). This does not create any difficulty if the growth is compensated for and if the steel is consistent in its growth.112 The distortion produced in the surface hardening of long shafts by the scanning method can be a great problem if the equipment is not in very good condition. Due consideration must be given so that locating centers run concentrically, in line, and at the appropriate speed; the coil must be accurately aligned, and the quench must be correctly designed with sufficient number of holes of suitable size and angle. For long shafts with a relatively small diameter (e.g., half shafts, which are likely to distort), the use of hydraulically operated restraining rolls usually overcomes this.128 The designer should bear in mind the following rules while designing a die or machine part that is to be heat-treated: 1. Distribution of the material should be as uniform as possible. 2. Provide fillets (larger radii) at the base of keyways, cutter teeth, and gear teeth to avoid stress concentration; semicircular keyways, which permit the use of roundcorner keyways, are the right choices. Ideally, drives using involute splines are preferred to keyways. 3. Avoid abrupt changes of section; in other words, provide smooth changes of section.

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.53

4. Large holes (such as drawing or cutting openings in die rings or plates) must be centrally located from the outer contour. In some cases, holes are drilled through the heaviest section of the tool in order to help fairly balance the weight of the section rather than to unbalance it.80 Deep blind holes should always be avoided because they cause nonuniform quenching. If this is not possible, the hole can be ground in after hardening. Drilled hole junctions in a steel part should be avoided because they enhance very high and undesirable cooling conditions. The problem with these cross-holes is to get sufficient quenchant into them. The inside surface of the holes tends to be in a state of high tensile stress, usually leading to cracking, at least with water quenching. As a minimum, the corner at the junction of the holes with outer diameter of the part should be given a generous radius to better distribute the tensile stress.116 Similarly, grooves and keyways in highly stressed areas should be avoided; or, if possible, they should be located in low-stressed areas of the part. Alternatively, fixtures should be used that make it possible for the hole or the inside of the groove to be quenched in the beginning or more rapidly than the rest of the part.27 5. Round off all the holes, corners, and outer edges. 6. If sharp corners are unavoidable, provide relief notches in place of sharp edges. 7. The insertion of identification marks on the hardened component is recommended, preferably after hardening with tools having well-rounded edges and minimum deformation (shallow penetration depth), and at positions far away from the high-stress concentration zones (reentrant angles, bends, and so on).129 8. Large intricate dies should be made up in sections, which frequently simplifies heat treatment.80

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

N. P. McLeod and J. Nutting, Met. Technol., vol. 9, 1982, pp. 399–404. G. E. Hale and J. Nutting, Int. Met. Rev., vol. 29, no. 4, 1984, pp. 273–298. R. W. Gardiner, Met. Technol., vol. 4, 1977, pp. 536–547. T. J. Baker and W. D. Harrison, Met. Technol., vol. 2, no. 5, 1975, pp. 201–205. T. J. Baker and R. Johnson, JISI, vol. 211, 1973, pp. 783–791. I. S. Brammar, JISI, vol. 201, September 1963, pp. 752–761. R. C. Andrews, G. M. Weston, and R. T. Southin, J. Aust. Inst. Met., vol. 21, 1976, pp. 126–131. R. C. Andrews and G. M. Weston, J. Aust. Inst. Met., vol. 22, 1977, pp. 171–176. G. D. Joy and J. Nutting, in Effects of Second Phase Particles on the Mechanical Properties of Steels, Iron and Steel Institute, London, 1971, pp. 95–100. R. N. O’Brien, D. H. Jack, and J. Nutting, in Proceedings of Heat Treatment ‘76, Metals Society, 1976, London, pp. 161–168. C. L. Briant and S. K. Banerjee, Met. Trans., vol. 10A, 1979, pp. 1151–1155. B. J. Sultz and C. J. McMahon, Jr., Met. Trans., vol. 4, 1973, pp. 2485–2489. N. P. McLeod, Ph.D. thesis, University of Leeds, 1978. A. Preece and J. Nutting, JISI, vol. 164, 1950, pp. 46–50. R. Prestner, Met. Mater., April 1974, p. 229.

17.54

CHAPTER SEVENTEEN

16. A. H. Bodimeade, Ph.D. thesis, University of Leeds, 1974. 17. G. D. Joy, Ph.D. thesis, University of Leeds, 1971. 18. D. R. Glué, C. H. Jones, and H. K. M. Lloyd, Met. Technol., vol. 2, 1975, pp. 416–421. 19. Carbon and Alloy Steels (ASM Specialty Handbook), ASM International, Materials Park, Ohio, 1996, pp. 308–328. 20. G. Parrish, private communication, 2000. 21. T. Hanabusa and H. Fujiwara, in Proceedings 32d Jpn. Congr. Mater. Res., 1989, pp. 27–36. 22. G. E. Dieter, Engineering Design, McGraw-Hill, New York, 1982. 23. G. Parrish and G. S. Harper, Production Gas Carburizing, Pergamon Press, Oxford, 1985. 24. B. Hildenwall and T. Ericsson, in Proceedings of Hardenability Concepts with Applications to Steel, eds. D. V. Doane and J. S. Kirkaldy, TMS-AIME, 1978, pp. 579–606. 25. E. B. Evans, in Encyclopaedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4183–4188. 26. R. F. Kern and M. E. Suess, Steel Selection, Wiley-Interscience, New York, 1979. 27. R. F. Kern, Selecting Steels and Designing Parts for Heat Treatment, ASM, Metals Park, Ohio, 1969. 28. R. B. Liss, C. G. Massieon, and A. S. McClosky, The Development of Heat Treat Stresses and Their Effect on Fatigue Strength of Hardened Steel, Presented at Society of Automotive Engineers midyear meeting, 1965. 29. R. W. Shin and G. H. Walter, in Proceedings of Residual Stresses for Engineers and Metallurgists, ed. J. Vande Walle, ASM, Metals Park, Ohio, 1981, pp. 1–20. 30. Carbon and Alloy Steels (ASM Specialty Handbook), ASM International, Materials Park, Ohio, 1996, pp. 365–389. 31. B. Scholtes and E. Macherauch, in Case Hardened Steels: Microstructural and Residual Stress Effects, ed. D. E. Diesburg, TMS-AIME, 1984, pp. 141–151. 32. G. Krauss, Advanced Mats. & Process., July 1995, pp. 48U–48Y. 33. J. A. Sanders, M.S. thesis, Colorado School of Mines, Golden, 1993. 34. G. Parrish, Carburizing: Microstructures and Properties, ASM International, Materials Park, Ohio, 1999. 35. P. S. Prevey, D. J. Hornbach, and P. W. Mason, Proceedings: 17th ASM Heat Treating Conference, eds. D. Milan et al., 1997, ASM International, 1998, pp. 3–12. 36. W. T. Cook, private communication, 2000. 37. R. W. K. Honeycombe and H. K. D. H. Bhadeshia, Steels: Microstructure and Properties, 2d ed., Arnold, London, 1995. 38. B. S. Lement, Distortion in Tool Steel, American Society for Metals, Metals Park, Ohio, 1959. 39. H. C. Child, Heat Treat. Met., 1981.4, pp. 89–94. 40. A. Rose and H. P. Hougardy, in Proceedings of the Transformation and Hardenability in Steels Symposium, Climax Molybdenum Company, Ann Arbor, Mich., 1967, pp. 155–167. 41. H. P. Kirchner, Strengthening of Ceramics: Treatment Tests and Design Applications, Marcel Dekker, New York, 1979. 42. W. Baldvin, Jr., Residual Stresses, in Proceedings of the American Society for Testing and Materials, vol. 49, 1949, pp. 539–583. 43. R. E. Reed-Hill and R. Abbaschian, Physical Metallurgy Principles, 3d ed., PWS-Kent Publishing, Boston, 1992. 44. A. Rose, Hart.-Tech. Mitt., vol. 21, no. 1, 1966, pp. 1–6. 45. K. E. Thelning, Steel and Its Heat Treatment, Butterworths, London, 1985.

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

17.55

46. T. Yamaguchi, Z. G. Wang, and T. Inoue, in Proceedings of the 27th Japan Congress on Materials Research, 1984, p. 147; Mater. Sc. Technol., vol. 1, 1985, pp. 872–876. , 47. D. E. Diesburg, C. Kim, and W. Fairhurst, Proceedings of Heat Treatment 81, Metals Society, London, 1983, pp. 178–184. 48. D. P. Koistinen, Trans. ASM. vol. 50, 1958, pp. 227–241. 49. L. Salonen, Acta Polytech. Scand. Ser., vol. 109, 1972, pp. 7–26. 50. H. C. F. Rozendaal, P. F. Colijn, and E. J. Mittemeijer, Surf. Eng., vol. 1, 1985, pp. 30–42. 51. Case Hardening of Steel, ed. H. E. Boyer, ASM International, Metals Park, Ohio, 1987. 52. T. Endo and M. Kawakami, J. Soc. Mater. Sci. Jpn., vol. 312, 1983, p. 114. 53. E. D. Walker, in Proceedings of Residual Stresses for Engineers and Metallurgists, ed. J. Vande Walle, ASM, Metals Park, Ohio, 1981, pp. 41–50. 54. S. L. Semiatin and D. E. Stutz, Induction Heat Treatment of Steel, ASM, Metals Park, Ohio, 1985. 55. M. Melander, Mater. Sci. Eng., vol. 1, 1985, pp. 877–882. 56. G. Parrish et al., Heat Treat. Met., vol. 25, 1998.1, pp. 1–8. 57. G. Parrish et al., Heat Treat. Met., vol. 25, 1998.2, pp. 43–50. 58. P. Petrov, D. Dimitrov, M. Aprakova, and S. Valkanov, Mater. and Manufact. Process., vol. 13, no. 4, 1998, pp. 555–564. 59. J. Grum, R. Sturm, and P. Zerovnik, in Proceedings Second International Conference on Quenching and Control of Distortion, eds. G. E. Totten et al., ASM International, Materials Park, Ohio, 1996, pp. 181–191. 60. R. L. Peng and T. Ericsson, Scand. Jn. of Met., vol. 27, 1998, pp. 223–232. 61. K. Masabuchi, in Encyclopaedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4180–4183. 62. L. Karlsson, in Thermal Stresses I, vol. 1, ed. R. Hetnarski, Elsevier, Amsterdam, 1986, pp. 299–389. 63. L. Novikov, Theory of Heat Treatment of Metals, Mir Publishers, Moscow, 1978. 64. R. N. Mittal and G. W. Rowe, Met. Technol., vol. 9, 1982, pp. 191–197. 65. J. O. Kristiansson, J. Therm. Stresses, vol. 5, 1982, pp. 315–330. 66. Polymer Quenchant User Report, Tenaxol, Inc., Milwaukee, Wisc. 67. R. G. Bathgate, Met. Forum, vol. 6, 1983, p. 11. 68. L. Mordfin, in Proceedings of Residual Stresses for Engineers and Metallurgists, ed. J. Vande Walle, ASM, Metals Park, Ohio, 1981, pp. 189–210. 69. F. Abbasi and A. J. Fletcher, Mater. Sci. Technol., vol. 1, 1985, pp. 770–779. 70. G. E. Totten and M. A. H. Howes, Steel Heat Treatment Handbook, eds. G. E. Totten and M. A. H. Howes, Marcel Dekker, New York, 1997, chap. 5, pp. 251–292. 71. L. Mordfin, in Encyclopaedia of Materials Science and Engineering, Pergamon Press, Oxford, 1986, pp. 4189–4194. 72. E. J. Mittemeijer, J. Heat Treat., vol. 3, no. 2, 1983, pp. 114–119. 73. T. H. De Keijser, L. J. Langford, E. J. Mittemeijer, and A. B. P. Vogels, J. Appl. Crystallogr., vol. 15, 1982, pp. 308–314. 74. D. J. Horrnbach and P. S. Prevey, 17th ASM Heat Treating Conference, eds. D. L. Milan et al., ASM International, Materials Park, Ohio, 1997, pp. 13–18. 75. T. R. Finlayson, Met. Forum, vol. 6, 1983, pp. 4–10. 76. D. S. Mountain and G. P. Cooper, Strain, vol. 25, no. 1, 1989, pp. 15–19. 77. R. F. Kern, Heat Treat., vol. 17, no. 4, 1985, pp. 38–42.

17.56

CHAPTER SEVENTEEN

78. D. H. Stone, in Proceedings of the 1988 ASME/IEEE Joint Railroad Conference, American Society of Mechanical Engineers, New York, 1988, pp. 43–53. 79. C. E. “Joe” Devis, Ask Joe, ASM, Metals Park, Ohio, 1983. 80. Chapter 8, in Troubleshooting Manufacturing Processes, 4th ed., ed. L. K. Gillispie, Society of Manufacturing Engineers, Dearborn, Mich., 1988. 81. J. Lyman, J. Eng. Mats. and Tech., vol. 106, January 1984, pp. 253–256. 82. A. K. Sinha, in ASM Handbook, vol. 4: Heat Treating, ASM International, Materials Park, Ohio, 1992, pp. 601–619. , 83. G. Wahl and I. V. Etchells, in Proceedings of Heat Treatment 81, Metals Society, 1983, pp. 116–122. 84. M. J. Gilersleeve, Mats. Sc. and Technol., vol. 4, no. 4, 1991, pp. 307–310. 85. G. Parrish, The Influence of Microstructure on the Properties of Case-Carburized Components, ASM, Metals Park, Ohio, 1980. 86. R. P. Brobst and G. Krauss, Met. Trans., vol. 5, 1974, pp. 457–462. 87. C. A. Apple and G. Krauss, Met. Trans., vol. 4, 1973, pp. 1195–1200. 88. T. A. Balliett and G. Krauss, Met. Trans., vol. 7, 1976, pp. 81–86. 89. R. J. Kar, R. M. Horn, and V. F. Zackay, Met. Trans., vol. 10A, 1979, pp. 1711–1717. 90. W. Fichtl, Hart-Tech. Mittelungen, vol. 33, 1978, pp. 1–8. 91. G. E. Hollox and R. T. Von Bergn, Heat Treat. Met., 1978.2, pp. 27–31. 92. T. Bell, Survey of Heat Treatment of Engineering Components, Iron and Steel Institute, 1973, pp. 69–72. 93. K. W. Chambers, Heat Treatment of Metals, Iron and Steel Institute, 1966, pp. 94–95. 94. H. W. Walton, in 2d International Conference on Quenching and the Control of Distortion, eds. G. E. Totten et al., ASM International, Materials Park, Ohio, 1996, pp. 143–148. 95. W. T. Cook, Heat Treat. Met., 1999.2, pp. 27–36; in Proceedings of the 18th Conference on Heat Treating, eds. R. A. Wallis and H. W. Walton, ASM International, Materials Park, Ohio, 1999, pp. 12–22. 96. R. W. Wilson, Metallurgy and Heat Treatment of Tool Steels, McGraw-Hill, New York, 1975, pp. 93–95. 97. P. G. Greenwood and R. H. Johnson, Proc. Royal Soc., vol. A283, 1965, p. 403. 98. B. L. Josefson, Mater. Sci. Technol., vol. 1, no. 10, 1985, pp. 904–908. 99. D. Huang, K. Arimota, D. Lambert, and M. Narazaki, Heat Treating Conference and Exposition, St. Louis, Mo., Oct. 9–12, 2000, pp. 1–5. 100. A. Ferrante, Met. Progr., vol. 87, 1965, pp. 87–90. 101. B. R. Wilding, Heat Treatment of Engineering Components, Iron and Steel Institute, 1970, pp. 20–25. 102. B. A. Becherer and L. Ryan, in ASM Handbook, vol. 4: Heat Treating, ASM International, Materials Park, Ohio, 1992, pp. 761–766. 103. R. F. Kern, Heat Treat., vol. 17, no. 3, 1985, pp. 41–45. 104. D. J. Grieve, Met. Mater. Technol., vol. 7, no. 8, 1975, pp. 397–403. 105. F. D. Waterfall, Met. Treat Drop Forg., April 1985, pp. 139–144. 106. S. Visvanathan, TISCO J., vol. 23, no. 4, 1976, pp. 199–204. 107. Y. Toshioka, Mater. Sci. Technol., vol. 1, no. 10, 1985, pp. 883–892. 108. R. Verma, V. A. Swaroop, and A. K. Roy, TISCO J., October 1977, pp. 157–160. 109. S. Owaku, 2d International Conference on Quenching and the Control of Distortion, eds. G. E. Totten et al., ASM International, Materials Park, Ohio, 1996, pp. 149–154. 110. Section 8, in Cassels Handbook, 9th ed., ICI Ltd., 1964.

DEFECTS AND DISTORTION IN HEAT-TREATED PARTS

111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129.

17.57

R. F. Harvey, Met. Progr., vol. 79, no. 6, 1961, pp. 73–75. A. K. Sinha, Tool Alloy Steels, August 1980, pp. 219–224. A. Khersonsky, Met. Heat Treat., July/August, 1996, pp. 43–45. G. F. Melloy, Hardening of Steel, Lesson 5, in Heat Treatment of Steels, Metals Engineering Institute, American Society for Metals, 1979, pp. 1–28. R. F. Kern, Heat Treat., vol. 18, no. 9, 1986, pp. 19–23. Hayes, Inc., private communication, October 2000. , J. M. Neiderman and C. H. Luiten, Proceedings of Heat Treatment 84, Metals Society, London, 1984, pp. 43.1–43.8. Met. Mater., vol. 9, July/August 1975, pp. 52–53. T. E. Hebel, Heat Treat., vol. 21, no. 9, 1989, pp. 29–31. D. Pye, in Steel Heat Treatment Handbook, eds. G. E. Totten and M. A. H. Howes, Marcel Dekker, New York, 1997, Chap. 10, pp. 721–764. F. Dunlevey, Heat Treat., vol. 21, no. 2, 1989, pp. 34–35. UCON Quenchants for Ferrous and Nonferrous Metals, Tenaxol, Inc., Milwaukee, Wisc., 1988. R. Creal, Heat Treat., vol. 18, no. 12, 1986, pp. 27–29. R. A. Claxton, Heat Treat. Met., 1991.2, pp. 53–59; 1991.3, pp. 85–89. C. E. Bates, J. Heat Treat., vol. 5, no. 1, 1987, pp. 27–40. Information on Polymer Quenchants, Tenaxol, Inc., Milwaukee, Wisc., 1989. C. E. Bates and G. E. Totten, Heat Treat. Met., 1988.4, pp. 89–97. P. D. Jenkins, Metallurgia, vol. 45, no. 4, 1978, pp. 196–199. F. Strasser, Heat Treat. Met., 1980.4, pp. 91–96.

CHAPTER 18

SURFACE MODIFICATION AND THIN-FILM DEPOSITION

18.1 INTRODUCTION Surface modification and thin-film deposition involve the alteration of surface composition or structure through the introduction of high-energy or particle beams, physical vapor deposition (PVD), and/or chemical vapor deposition (CVD) techniques that can serve to improve selected mechanical properties and wear and hardness, reduce friction, and increase fatigue, corrosion, and oxidation resistance of the critical surface area. Ion beam processes are used to produce surface-modified thinfilm deposits. PVD is the production of a condensable vapor by physical means and subsequent low-temperature deposition of elements and alloys, as well as compounds using reactive deposition processes at low to ultrahigh vacuum. CVD is the deposition of atoms or molecules by the high-temperature chemical reaction at atmospheric pressure or low vacuum to ultrahigh vacuum. This chapter describes the ion beam, physical vapor deposition, and chemical vapor deposition processes with respect to the principles of operation, the effects on the surface properties, advantages and disadvantages, and applications.

18.2 ION BEAM PROCESSES Ion beam processes offer various possibilities as industrial surface treatments for improving wear and corrosion resistance of critical components. Ion beam processes can be grouped into four categories: ion implantation, ion beam mixing, ion beamassisted deposition, and ion plating. All of them allow independent control of process parameters such as particle energy, particle flux, gas pressure, and substrate temperature. Figure 18.1 shows a schematic view of ion implantation, ion beam mixing, and ion beam-assisted deposition after coating.1

18.2.1 Ion Implantation Ion implantation is an important surface modification, precise doping, or surfacealloying process for metals and alloys, semiconductors, ceramics, dielectrics, optical materials, and polymers, which involves the generation of high-energy gaseous or 18.1

18.2

CHAPTER EIGHTEEN

FIGURE 18.1 Schematic illustration of coating, ion implantation, ion beam mixing, and ion beam-assisted deposition.1 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

metallic ion beams in an ion source or gun and acceleration, under very good vacuum (10-6 torr or better) through the scanner plates toward a highly charged substrate, where the subsequent bombardment of the surface of the substrate material occurs (Fig. 18.1), independent of thermodynamic criteria such as solid solubility and diffusivity. The basic equipment requirements include an ion source—an evacuated chamber containing an ionized gas, or plasma, a high-voltage accelerator, a mass separator (usually magnetic deflection), and a beam sweeper. Figure 18.2 is a schematic illustration of an ion beam accelerator.2 McHargue has described a review dealing with the effect of ion implantation in (a) metals and alloys with respect to microstructure, hardness, wear and friction, fatigue, oxidation, and corrosion and (b) ceramics with respect to structure and surface mechanical properties.3 The mean penetration depth of ion-implanted layers is a function of ion energy and atomic mass, as well as the mass and atomic number of the substrate material and the angle of incidence. Typical penetration depths are 1017 ions/cm2 are mostly required with ion implantation. Thus, high concentrations are obtained with less accelerator time. The configurations in Fig. 18.5a were employed to alloy Al and Pt by Mayer et al.,46 as shown in Fig. 18.6. With Xe ion mixing, RBS analysis suggests that the initial sharp interface between the layers (the nearly vertical solid lines at the right edge of the Pt signal and left edge of the Al signal) becomes graded with the formation of a Pt-Al alloy. In this case, a stoichiometric compound (PtAl2) forms, and the mixing proceeds at a constant composition with increasing Xe influence, as shown by the steps in the two signals. The ion beam mixed phases can be quite different from those formed by thermally reacting the layers, and, in this system, PtAl4 forms when the Pt/Al layers are annealed, instead of PtAl2.47 Many experimental studies have been carried out in many systems to determine the influence of temperature and irradiation parameters. Different mechanisms identified in IBM include the ballistic effects, radiation-enhanced diffusion, thermal spike mixing, and chemical effects. The ballistic effects denote mass transport. They are independent of temperature and consist of both recoil mixing and cascade mixing. The former occurs by a direct collision between an incident ion on a target atom. However, the number of atoms transported by this mechanism is smaller than that of the cascade mixing. Sigmund et al.48 have developed a collision model of cascade mixing. If one assumes that the atom relaxation process is basically equivalent to a vacancy-interstitial pair

FIGURE 18.5 Schematic configurations used for ion beam mixing: (a) bilayer and (b) multilayer.45

18.12

CHAPTER EIGHTEEN

FIGURE 18.6 RBS spectra for Al/Pt layers ion beam-mixed with 300 keV Xe. The steps show the formation of PtAl2.46

formation and that the phenomenon is almost isotropic, an apparent diffusion coefficient D0 for cascade mixing can be given by the Andersen formula49: D0 t 0.42 FdRc2 = 6 NEd f

(18.3)

where Rc is the distance between a stable vacancy-interstitial pair, Ed is the displacement threshold energy, Fd is the amount of elastic energy lost per unit length, t is the irradiation time, and f is the ion fluence. Since the number of atomic displacements is much greater than the number of implanted ions, the cascade mixing exerts a very efficient effect in homogenizing the multilayers to an atomic scale even at lower temperatures. When the mixing is carried out at a temperature where radiation effects are mobile, an additional temperature-dependent mechanism takes place due to the extra cascade transport resulting from the long-range migration of radiationinduced defects. The diffusion coefficient, D*(T), for radiation enhanced diffusion mixing is expressed by D* (T ) = Cv (T )Dv (T ) + Ci (T )Di (T )

(18.4)

where Cv and Ci are the concentrations of vacancies and interstitials, respectively, and Dv and Di are the corresponding diffusivities. However, if a large number of low-temperature mixing experiments can be understood on the basis of ballistic mechanism, there appear to be many instances where chemical driving forces have

SURFACE MODIFICATION AND THIN-FILM DEPOSITION

18.13

to be considered. Recent calculations suggest that mixing at low temperature must assume a diffusion effect during the thermal spike of the cascade in which local thermal equilibrium is set up and a temperature is defined. Thus the effective diffusion coefficient, D, is described by 2DH m ˆ ˘ D = D0 ÈÍ1 - Ê Î Ë kT ¯ ˙˚

(18.5)

where D0 is the diffusion coefficient for the pure ballistic effect and T is an effective temperature defined by (3/2) kT = qD, where qD is the energy density in the cascade and DHm is the heat of mixing. This diffusion coefficient D can be either positive or negative depending on the sign of DHm. The above equation can explain why in binary systems with a sufficiently large positive heat of mixing, the value of D is negative and there are difficulties in mixing, as in the Cu-W system.9,50,51 The phases of ion beam-mixed alloys are often the same as those of ionimplanted or -irradiated alloys with the same composition, as well as the immiscible alloy systems.9 The advantages of ion beam mixing include (1) the ability to markedly enhance film/substrate adhesion properties due to improvement of the solid solubility and ion-induced compound formation at the film/substrate interfaces and formation of compressive residual stresses within the film, and (2) the independent control of processing parameters such as ion flux and energy.52 18.2.3 Ion Beam-Assisted Deposition. Ion beam-assisted deposition (IBAD) is a process in which ion bombardment and physical vapor deposition are combined. The extra energy imparted to the deposited atoms leads to atomic displacements at the surface and in the bulk, as well as improved atom migration along the surface. These resulting atomic movements result in the improved film properties such as better adhesion and cohesion of the film, higher density, and modified residual stress, when compared to similar films formed by PVD without ion bombardment. When the ion beam or the evaporant is a reactive species, compound semiconductors such as Si3N4 can be synthesized at very low temperatures. In addition, adjustment of the ratio of reactive ions to atoms arriving at the substrate surface allows adjustment of the stoichiometry of IBAD films. Detailed reviews of the IBAD process can be found elsewhere.53–57 There are two primary ways to accomplish the IBAD process. One way is based on the low-energy (0.5 to 5 keV) ion source such as the broad-beam Kaufman-type ion gun (Fig. 18.7a), and is used without mass separation. The second way is based on high-energy ion implanters. Figure 18.7a and b are schematic IBAD configurations with a simultaneous and alternating evaporation and inert gas ion bombardment arrangement.1 IBAD processing can be grouped into three types4: 1. Nonreactive IBAD, where inert gas ions (Ar+) are used to influence the nucleation and growth of the deposited elements or compounds. 2. Reactive IBAD, i.e., when the vacuum chamber is backfilled with a reactive gas and where the ion beam is used to both influence film growth and provide atoms for the growth of a compound film (e.g., Si or Ti deposition with N+ ion bombardment to form Si3N4 or TiN) in a reactive gas atmosphere. 3. Modified reactive IBAD, where atoms in the form of a backfilled molecular gas (e.g., N2 or O2) are provided so that these atoms are introduced (i.e., activated) into the growing film by bombardment with energetic ions (inert or reactive).

CHAPTER EIGHTEEN

18.14

(a)

(b)

FIGURE 18.7 Classical IBAD geometry with simultaneous evaporation and inert gas ion source (left side) and alternating IBAD arrangement (right side).1 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

FIGURE 18.8 Single-beam (left side) and dual-beam (right side) arrangement for IBAD with sputter deposition.1 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

This can be employed to produce a stoichiometric compound when the evaporant is adequately reactive.58,59 It can also be employed to compensate for the loss of constituent elements (e.g., oxygen) from the evaporating compound (e.g., Al2O3 or SiO2), which tends to decompose at high temperatures yielding a metalrich coating in the absence of such an O2 backfill.4 The typical ion energy range employed for IBAD/RIBAD is 100 eV to 30 keV, because even in the alternating arrangement the deposited intermediate layer is very thin, thereby requiring less energy. The deposition technique employed for IBAD/RIBAD is evaporation of elements or alloys, but it can also be used for sputter deposition in either dual- or single-beam arrangements (Fig. 18.8).1 (See Sec 18.3.2 for more details on sputter deposition.) Recent experimental studies have shown that the ion beam assistance energy and the arriving ions (I) per arriving atoms (A) ratio are important parameters in modifying the properties of the deposited coating.9

SURFACE MODIFICATION AND THIN-FILM DEPOSITION

18.15

Advantages and Limitations. The advantages of the IBAD process over competing coating processes are (1) low deposition temperature, (2) high adhesion, (3) coating densification, (4) residual stress reduction, (5) columnar microstructure elimination, and (6) controlled microstructure and crystalline film orientation.4 Even polymers with low melting points can be coated, due to lower deposition temperatures (below ⬃100°C or 212°F). The properties of adhesion, density, and stress are superior to those of PVD films, and there is a greater control over the microstructure. Based on the deposition parameters, films can be deposited as (1) metastable crystalline, (2) amorphous, (3) textured crystalline or epitaxial (for some materials), and (5) nanocrystalline.53 Eventually, the composition, or crystalline phases, can be precisely varied as a function of thickness to form functionally gradient materials with properties such as graded hardness, density, tensile strength, stress, coefficient of thermal expansion, and refractive index.53 The limitations of the IBAD process are (1) moderately higher cost than PVD, (2) line-of-sight processing, and (3) limited users.53 Applications. The widespread industrial growth in ion beam-assisted coatings is justified by their excellent tribological and mechanical properties (hardness and adhesion), corrosion resistance, reproducible and stable optical properties, and decorative value. There is promise of useful dielectric applications. The films are very smooth, with a fine equiaxed microstructure rather than a columnar one, and they are applied at low temperatures.60 Applications can be described in the areas of optical films, oxidation and corrosion-protection coatings, and tribological coatings. 1. Optical films. The IBAD process has been used in optical thin films for two types of applications. In the first type of applications, where densification is the primary concern, lowenergy argon or oxygen ions are typically employed to bombard the optical thin films during deposition.61 As a result, the density is enhanced, sometimes approaching the bulk material. The main advantage is certainly not the increase in refractive index to near-bulk values, but rather the stability of the refractive index value under humidity and temperature variations because of the freedom from voids or pores in the film that absorb water vapor. This makes optical-coating design simpler and favors better control and reproducibility in the fabrication process. Another attraction of using low-energy ions is the improved adhesion to the substrate, which also assists in the increased productivity. In the second type of applications, where graded refractive index profiles is the main requirement, the IBAD process can be extensively used to fabricate graded index antireflection coatings, reflection filters, and mirrors.62 Some of these devices can be tens of microns thick, which implies that stress control is a major factor. Usually, low energies are preferred for the deposition of optical films to reduce absorption owing to radiation damage. 2. Diamond-like carbon (DLC) coating. Currently, IBAD, ion-induced deposition (IID), and filtered cathodic-arc (or vacuum-arc) deposition processes can be used to deposit diamond-like carbon (DLC) and zirconia for the biomedical and corrosion applications due to their high hardness, low coefficient of friction and wear rate, and chemical inertness.4,63 A low-friction DLC coating is one of the more successful applications of IBAD technology. In a classical IBAD, using a broad-beam ion source of Kaufman type (Fig. 18.7a), the coating material itself is evaporated with simultaneous bombardment at a low-energy (0.1 to 2 keV) argon ion beam to deposit DLC at rates between 0.1

18.16

CHAPTER EIGHTEEN

FIGURE 18.9 Setup for depositing DLC by ion-induced deposition using hydrocarbon vapor.63 (Reprinted by permission of The Metallurgical Society, Warrendale, Pennsylvania.)

and 1 nm/s (1 and 10 Å/sec).63 In ion-induced deposition of DLC, the ion beam is used to both decompose and provide impact energy to an organic evaporant gas which is directed toward the substrate. Figure 18.9 exhibits a setup using a 10- to 40-keV nitrogen or argon ion beam and hydrocarbon vapor to form thin films of DLC that are highly halogenated and may contain some amounts of silicon and oxygen. The films are less hard than the filtered cathodic arc amorphous carbon but are very lubricious with lower residual stress, which permits the deposition of thicker films without peeling off from the substrate.63 In filtered cathodic-arc (also called vacuum-arc) DLC, pure-carbon ions with energies of 20 to 30 eV are emitted from graphite cathode and impacted on the negatively biased substrate (see Fig. 18.12). This type of amorphous DLC is characterized by high hardness, high residual stress, and good adhesion with a density of ≥3.0 g/cm3 and a specific chemical (sp3) bonding content of 80 to 90%. (See Sec 18.3.1.3 for more details on vacuum arc deposition.) The potential applications of DLC coatings include (1) friction- and wearresistant coatings for engine components, tools and dies, pump and wear components, and bearings and gears; (2) coatings for audio speakers, x-ray windows, and heat sinks for electronic devices; (3) mold release coatings for compression and injection molds; (4) coatings for resistance to chemical attacks; (5) coatings in optical fields for sunglasses, ophthalmic lenses, infrared windows, laser optics, and fiber optics, and magnetic and optical recording disks and heads; and (6) coatings for medical (prosthetic) devices.28,64 This process is also used for hard, protective coatings for optics and windscreens on vehicles. Although DLC absorbs strongly in the visible range, thin coatings (20 to 200 nm or 200 to 2000 Å) remain transparent to serve as protective transmission coatings. The advantages of these coatings are high scratch hardness, low porosity (reduced number of pinholes), and superior adhesion to most substrates.53 3. Wear- and corrosion-resistant coatings. Several researchers have investigated the parameters required to achieve hard wear-resistant IBAD coatings (e.g., TiN coating of 316 stainless steel electric razor screens and on Al2O3) by using bombardment with either (1) a nitrogen ion, or (2) an argon ion to accelerate incorporation of ambient nitrogen gas with a reactive species (e.g., TiN).65,66 They are also attractive for corrosion-protection applications.

SURFACE MODIFICATION AND THIN-FILM DEPOSITION

18.17

Pt, TiC, TiN, B, diamond-like carbon, chromium nitride, boron nitride, chromium oxide, silicon nitride, and silicon coated on metals using the IBAD process offer excellent corrosion resistance.67–72 IBAD chromium nitride (CrxNy) is found to be a candidate for electroplated hard chromium (EHC) replacement due to its high hardness, good corrosion resistance, and overall similarity to EHC.4 4. High-temperature oxidation resistance. Oxidation protection of titanium alloys with IBAD coatings of chromium nitride,73 TiN,73 and silicon nitride74 has been reported. 5. Ion-induced CVD. Several researchers have adopted reactive IBAD to produce unique hydrocarbon or ceramic films.75 This process involves the introduction of a gas into the chamber, cooling the substrate to induce condensation of the gas, and bombarding the surface with an ion beam. During the process, hydrocarbon bonds are collapsed, volatile species are released, and a coating is effected. For silicone oil vapor, the films can vary between very-low-friction solid lubricants and very-hard, corrosion-resistant silicon oxycarbide (SixOxCy) coatings, depending on the arrival ratio of ions to vapor-condensed atoms. This process is akin to CVD, where the high temperature of the substrate provides the energy to begin chemical reactions leading to the film formation. In the ion beam case, the same or similar reactions can be beam-induced at room temperature, opening the likelihood of depositing CVD-like films on polymers and other temperature-sensitive substrates. 6. Friction and wear resistance. The IBAD process can be used to deposit solid lubricant coatings (e.g., molybdenum disulfide), which exhibit greater adherence to the substrate and longer lifetime.76,77 18.2.4 Ion Plating. Ion plating or ion vapor deposition* is the name given to a class of ion-assisted PVD processes. The bombarding species as well as the depositing species can be from numerous sources. Bombardment can occur in plasma or vacuum conditions. The process is often termed IBAD if bombardment takes place in vacuum.78 In plasma-based ion plating, the substrate remains in contact with a plasma, and the ions are accelerated from the plasma and strike the substrate surface. In vacuum-based ion plating, the film deposition occurs in a vacuum and the bombardment is from an ion or plasma “gun.” In reactive ion plating, the plasma or ion/plasma gun produce ions of a reactive species to both bombard and react with the depositing material to form a compound film material. In some instances, such as when using low-voltage, high-current electron-beam evaporation or arc vaporization, a sufficient amount of the vaporized source material can be ionized to permit bombardment by “film ions.” Often the term ion plating is associated with modifying terms such as reactive ion plating, sputter ion plating, chemical ion plating, arc ion plating, and alternating ion plating, which denote the method employed to bombard the film, the source of depositing material, or other specific conditions of the deposition. Figure 18.10a shows a simple plasma-based ion plating system with a resistively heated vaporization source, in which the negatively biased substrate can be placed in the plasma generation regime or in a remote or downstream location outside the active plasma generation area. Figure 18.10b shows a schematic vacuum-based configuration with an electron-beam evaporation source.79 Figure 18.10c is a schematic depiction of a cathodic- (multiple) arc (or ion bond) (source) ion plating unit.78 This * In the case of ion plating, the ionized fraction is usually less than 5%.

CHAPTER EIGHTEEN

18.18

Variable leak

Gas

Cathode dark space

Insulator Movable shutter Ground shield Substrate

_ High voltage supply

+

Plasma

Evaporator filament

Current monitor

Chamber High current feedthroughs

Vacuum Filament supply

(a)

+ Power

supply

– Darkspace shield

Water cooling

Gas inlet 10 m Torr Ar Substrate Shutter Pump out valve Pressure barrier

Electron beam heated source –5

10 Torr Vacuum pumps (b)

FIGURE 18.10 Schematics showing typical ion plating installations: (a) plasma-based configuration with resistively heated vaporization source;78 (b) vacuum-based configuration with electron-beam-heated evaporation source;79 and (c) ion-bond PVD process.80 [(a) Reprinted by permission of ASM International, Materials Park, Ohio; (b) Courtesy of Institute of Physics Publishing, Bristol; (c) Courtesy of Ion Bond, Inc.]

is a cost-effective method for hard-coating tools and wear parts, and closely toleranced components made from temperature-sensitive materials. For the deposition of compound coatings such as TiN, TiC, ZrN, HfN, Si3N4, or Al2O3, metal vapors are released into a “reactive” plasma produced in argon plus an appropriate “reactive” gas, i.e., N2 for nitride coating; CH4, C2H6, or C2H2 for carbides; O2 for oxides; and possibly BCl3 for borides. For the deposition of oxide

SURFACE MODIFICATION AND THIN-FILM DEPOSITION

18.19

Plasma

Reactive gas

Neutral gas Substrate Evaporator

Vacuum pump

Coating material Evaporated Material

Power supply (c)

FIGURE 18.10 (Continued) Schematics showing typical ion plating installations: (a) plasma-based configuration with resistively heated vaporization source;78 (b) vacuumbased configuration with electron-beam-heated evaporation source;79 and (c) ion-bond PVD process.80 [(a) Reprinted by permission of ASM International, Materials Park, Ohio; (b) Courtesy of Institute of Physics Publishing, Bristol; (c) Courtesy of Ion Bond, Inc.]

coatings or when trying to deposit nonconducting substrates made from glass, ceramic, or plastic, generally the dc power supply is replaced by an rf power supply, usually rated at 13.56 MHz, which minimizes the likelihood of charging or arcing.81 The extent of the reaction is a function of the plasma conditions, bombardment conditions, and the availability of the reactive species. By controlling the availability, the film composition can be changed. For example, in the reactive ion plating of TiN, first the availability of the nitrogen is decreased in the plasma at the commencement of deposition, to form an initial layer of titanium; next, the availability of nitrogen in the plasma is increased to form a “graded” interface.78 Advantages and Limitations of Plasma-Based Ion Plating. Plasma-based ion plating is the widely used ion plating configuration. The advantages of plasma-based ion plating are the following79,82,83: 1. Superior surface-covering ability (“throwing power”) of the evaporant under the appropriate conditions to provide more uniform coating of complex geometry 2. Ability to sputter clean the substrate surface prior to deposition 3. Ability to achieve good adhesion even in many difficult-to-deposit systems 4. Ability to introduce heat and defects into the first few monolayers of the surface to improve nucleation, reaction, and diffusion 5. Improvement of reactive deposition process (activation of reactive gases, bombardment-improved chemical reaction, and adsorption of reactive species)

18.20

CHAPTER EIGHTEEN

6. Flexibility in tailoring film properties by adjusting bombardment conditions (such as density, morphology, and residual stress) 7. Equipment requirements similar to those of sputter deposition 8. Deposition source that can be from resistance or induction-heated thermal evaporation, arc vaporization, sputtering, or chemical vapor precursor gases78. The limitations of plasma-based ion plating are the following: 1. The intense energy of ion vapor deposition may impart substantial energy to the substrate in excess of that of evaporation or sputtering. This can easily melt plastic or anneal aluminum substrates. 2. The energy of ion deposition rapidly heats the deposition chamber and tooling, driving off the absorbed gas and contaminants. These gases activate and contaminate the plasma and react with the films. Thus ion plating needs the same cleanliness and operating care as sputtering. The heat buildup in the chamber and tooling also necessitates water cooling, means to avoid burns, and extended cooling cycles, slowing production. Additionally, the exceptional throw of evaporant flux drives the film far under deposition masks.79 3. To bombard growing films of electrically insulating materials from a plasma, the surfaces must either reach a high self-bias or a bias with pulsed dc power or an rf potential.78 4. Uniform bombarding and availability of reactive species may be difficult to achieve with a complex substrate surface. 5. High residual compressive growth stresses may be built into the film layer due to “atomic peening”.78 Applications. Typical applications of the ion plating process are the following: 1. Good adhesion between a film and substrate (e.g., Ag on steel for mirrors and bearings; Ag on Be for diffusion bonding) 2. Electrically conductive coatings (Al, Ag, and Au) on semiconductors and plastics 3. Hard, wear-resistant, and abrasion-resistant coatings such as TiN, TiNx, TiCxNy, (TiAl)CxNy, and Ti0.5Al0.5N on metal cutting and working tools, molds, dies, jewelry, and certain aircraft components, and CrN + Cr2O3 on piston rings78 4. Low-shear solid film lubricants (e.g., Ag, Au, and Pb) 5. Corrosion-resistant coatings such as Al on U, mild steel, and Ti, and carbon and tantalum on biological implants 6. Decorative coatings (TiN provides gold-colored deposits, TiCxNy provides rosecolored deposits, TiC provides black deposits, ZrN provides brass-colored deposits) applied to hardware, cutlery, jewelry, and guns 7. Deposition of electrically conductive diffusion barriers (such as TiN and HfN on semiconductor devices) 8. Deposition of insulating coatings (such as SiO2, Al2O3, and ZrO2) 9. Deposition of permeation barriers on webs 10. Deposition of optically clear, electrically conductive films (indium-tin oxide, ITO)

SURFACE MODIFICATION AND THIN-FILM DEPOSITION

18.21

11. Deposition of Al on very large structural parts for corrosion protection (as a substitute for electroplated Cd) 12. Base deposition for further deposition by other means such as electroplating and painting 13. Filling vias and trenches on semiconductor surfaces by sputter deposition78 14. Improved strength of the copper/cordierite-type glass ceramic interface relative to evaporation52

18.3 PHYSICAL VAPOR DEPOSITION In physical vapor deposition (PVD) processes, the depositing material is created by physical means (either vacuum evaporation or sputtering) and transported from the source (target) through a vacuum or low-pressure gaseous (plasma) environment to the substrate, where subsequently the vapor phase is condensed onto a substrate material to form a thin film. The basic PVD processes, which can be considered as complementary techniques, fall into three broad categories: thermal evaporation, sputter deposition, and molecular beam epitaxy (MBE). Thermal evaporation is grouped into resistance evaporation, electron beam evaporation, ion vapor evaporation (described in an earlier section), vacuum arc deposition, and laser ablation. Sputter deposition is divided into dc sputtering, rf sputtering, and magnetron sputtering. Molecular beam epitaxy is grouped into conventional MBE, solid-source MBE (SSMBE), metalorganic MBE (MOMBE), and gas source MBE (GSMBE). Table 18.3 is a comparison of typical deposition characteristics of hard coatings produced by electron beam evaporation, sputtering, and cathodic arc evaporation deposition techniques. Advantages and Disadvantages. The major advantages of PVD processes are the following: 1. The ease with which a wide variety of deposited graded, multicomponent, multilayered, and nonequilibrium coatings and alloys can be produced. For example, nonequilibrium Al alloy PVD coatings show high tensile strength, improved corrosion resistance, and excellent thermal stability.84 2. Relatively low temperature on the substrate material surface (100 to 550°C) during deposition43 allows the use of many substrates including heat-sensitive materials and brazing filler metal to manufacture brazed carbide tools; this also produces crack-free coatings with a uniform fine-grained microstructure.85,86 3. PVD coatings exhibit smooth finish and, therefore, produce less friction during machining.86 4. PVD coatings produce compressive surface stress, which assists in resisting crack initiation and propagation.86 5. PVD coating retains the transverse rupture strength of carbide, while the CVD coating usually reduces the transverse rupture strength. This makes the PVD coatings more desirable for milling inserts, which need higher impact strength due to interrupted cutting.86 6. PVD coatings can be applied uniformly over sharp cutting edges, whereas CVD coated inserts need honing.86 7. Multilayered PVD coatings of TiN/TiCN and TiN/TiAlN combine the beneficial effects of the individual layers into a multilayered coating.86

18.22

CHAPTER EIGHTEEN

TABLE 18.3 Characteristics of Hard Coatings Produced by Three PVD Processes E-beam State of source Type of source

Solid Metals

Flux composition Heating and conditioning Source voltage (V) Source current (A) Mean particle energy (eV) Degree of ionization (I/N %) Substrate voltage (V) Substrate temperature (°C) Substrate current density (mA/cm2) Deposition rate (mm/min) Ion to neutral ratio (I/N) Source location Number of sources Control of stoichiometry System throughput Operation complexity Fixturing complexity Ion generation Ion energy levels (I/N) Process time (3/4 mm TiN) Alloy source evaporation Gas phase reaction control

Sputtering

Arc evaporation

Atoms, ions Argon ions 70–100 140 83 MPa for WC-Co coatings, and 70 MPa for Al2O3 coatings) and lowest porosities (usually no porosity for Mo coatings, ~0.5% for WC-Co coatings, and 1.8 mm (0.070 in.). Irrespective of the method or type of sealant employed, pores or interconnected porosities that are not linked to the exterior surface are incapable of being sealed, and machining or wear in service may open these with a consequent loss of corrosion protection.1 Table 19.10 lists a comparison of dielectric constant and electrical loss factors of APS-Al2O3 coatings impregnated with silicone under vacuum and at atmospheric pressure.13

THERMAL SPRAY COATINGS

19.31

TABLE 19.10 Comparison of Dielectric Constants and Loss Factors for APS-Al2O3 Coatings Impregnated with Silicone under Vacuum and at Atmospheric Pressure13 Pressure in the chamber while impregnating, hPa 1011 0.13

Dielectric constant

Loss factor at 1 kHz

6.24 ± 0.43 4.31 ± 0.17

0.039 ± 0.008 0.019 ± 0.002

Reprinted by permission of John Wiley & Sons, England.

19.4.3 Finishing Thermally sprayed coatings must be finished for most service applications. The common finishing methods include grinding, lapping, polishing, machining, abrasive brushing, peening, or vibratory finishing. However, special care is needed in such endeavors to avoid damage of the coatings, resulting in excessive surface porosity because of extraction of coating particles or thermal stress-related cracking. The ultimate surface finish obtained with a thermal spray coating depends on its composition, deposition parameter (which are solely responsible for the size and amount of true porosity in the coating), and the cohesive strength or particle-to-particle bonding within the coating. 19.4.3.1 Grinding. Table 19.11 provides typical machining parameters for some thermal sprayed metallic coatings. Generally, lower in-feed rates are used with wrought materials. Burnishing is frequently employed with soft materials such as Sn, Zn, and babbit metals to form a smooth, dense bearing surface; they are often more cost-effective and provide better finishes for many other coatings. If grinding does not make a very smooth surface, it may be worthwhile to lap the coating after grinding. Again, it is advisable to consult with the manufacturers of lapping materials for specific recommendations. Diamond or CBN (cubic boron nitride) wheels are used to grind the ceramic coatings while SiC wheels are used to grind metallic coatings. 19.4.3.2 Polishing and Lapping. Polishing is used with the same machines as used for grinding, but with a finer abrasive wheel. Close control of the polishing parameters leads to surfaces with Ra < 0.2 mm (for chromium oxide coatings). Smoother surfaces can be obtained with lapping (using very fine diamond or Al2O3 particles in an oil or grease mixture).13

19.4.4 Other Methods Ohmori et al.73 reported a heat treatment at 1073 K coupled with electric current conduction through APS-8% Y2O3-stabilized ZrO2 coatings that led to increased adhesion with copper and stainless steel substrates. Similarly, Burman et al.74 have reported the elimination of oxide present in the as-sprayed APS-FeCrAlY coatings by exposing it to electron beam remelting.13

TABLE 19.11 Typical Ranges of Speeds and Feeds Used in Machining Thermal Sprayed Metal Coatings1 Carbide tool (WC-6 Co)

High-speed steel tool Speed Coating metal

m/s

Feed sfm

mm/rev

Speed

Feed

in./rev

m/s

sfm

mm/rev

in./rev

Steels

19.32

Low carbon, medium carbon, low alloy High carbon, stainless

0.25–0.50

50–100

0.075–0.125

0.003–0.005

0.25–0.50

50–100

0.075–0.125

0.003–0.005

—

—

—

—

0.15–0.20

30–40

0.075–0.100

0.003–0.004

Nonferrous metals Brass, bronze, nickel, copper, monel Lead, tin, zinc, aluminum, babbit

0.50–0.75

100–150

0.075–0.125

0.003–0.005

1.25–1.80

250–350

0.050–0.150

0.002–0.006

0.75–1.00

150–200

0.075–0.175

0.003–0.007

1.25–1.80*

250–350*

0.050–0.100

0.002–0.004*

* Aluminum only. Reprinted by permission of ASM International.

THERMAL SPRAY COATINGS

19.33

19.4.5 Coating Repair The repair of thermal spray coatings by coating over service-worn or in-process damaged coatings is usually not recommended, even though the predeposited coating is reference ground, cleaned, and grit blasted. Hence the recommended method is to strip the existing coating and apply a completely new coating. It should be noted that when applying a multilayered coating, it is necessary to apply each new layer over the as-deposited surface of the previous layer; grinding or grit blasting between layers is not recommended.1

19.5 COATING CHARACTERISTICS Coatings exhibit lamellar or flattened grains tending to flow parallel to the substrate. The coating structure is heterogeneous with respect to wrought and cast materials. This is attributed to variations in the condition of the individual particles on impact. The variations in porosity and oxide content as well as chemical composition of the coating significantly influence the coating’s properties, and in the case of corrosion, the underlying substrate.5 Coating characteristics can be classified into the following: 1. Microstructure [complex, composite structures consisting of a metal substrate (usually), a metallic bond coat, and a ceramic overlay, porosities, deposit cohesion and adhesion, unmelted particles, and oxidation levels] 2. Mechanical properties (hardness, modulus, bond strength)

19.5.1 Microstructures Thermal spray coatings consist of a layered structure composed of splats along with pores, oxides, cracks, deposit cohesion and adhesion, and unmelted particles, which lead to their anisotropic and heterogeneous structure. The service performance of these coatings depends strongly on the microstructure of the coating and the properties of the coating/substrate interfaces.65 A detailed examination of microstructural details by both optical and scanning electron microscopy is necessary in order to understand the mechanism of coating formation and the coating properties such as bond strength, hardness, and corrosion resistance, and thereby the coating or bulk properties. Cross-sectional microstructures of several thermal sprayed coatings are shown in Figs. 19.1 and 19.13. Usually, the higher-particle-velocity coating processes provide the densest and superior bonded coatings, both cohesively (splat-to-splat) and adhesively (coating-to-substrate). Metallographically calculated porosities for detonation gun coatings and some HVOF coatings are £2%, whereas for most plasma-sprayed coatings porosities are in the range of 5 to 15%.1 Porosity.* Porosity is a characteristic feature and structural index of thermally sprayed coatings.75 Porosity strongly influences the ultimate physical and mechani* There are two types of porosity. (1) Open porosity comprises coarse pores formed between the lamellae and are interconnected through small channels to the surface. (2) Closed porosity, so called microporosity, comprises spherical pores formed by entrapped gas within the lamellae. Porosity (or voids), cracks, and coating decohesion are usually recognized by metallographic techniques.

CHAPTER NINETEEN

19.34

(a)

(c)

(b)

(d)

FIGURE 19.13 Microstructures of a detonation-gun-deposited tungsten carbide/cobalt cermet coating: (a) as polished and (b) etched. A mechanically mixed chromium carbide/NiCr cermet coating: (c) as polished and (d) etched.1 (Reprinted by permission of ASM International.)

cal properties of thermally sprayed deposits; it is a function of thermal, fluid flow, and solidification conditions. Most thermally sprayed coatings (except VPS, postheat-treated, or fused coatings) contain porosity (up to 25%). Porosity can be harmful in coatings with respect to (1) corrosion (sealing of coatings recommended), (2) machined finish, and (3) strength, microhardness, ductility, and wear characteristics. However, porosity can be beneficial with respect to (1) lubrication (porosity appears as a reservoir for lubricants), (2) reduction of stress levels and increase of thickness limitations, (3) increase of thermal barrier properties, (4) abradability in clearance control coatings, (5) increase of shock resistance properties, and (6) applications in prosthetic devices, etc.19 Porosity can be reduced to a minimum by optimizing the spray-deposited conditions (such as sprayed material, spraying parameters, and pressure induced on the surface of substrate during droplet impingement) or by thermomechanical processes.75a It has been shown that porosity decreases with increasing particle velocity and temperature. High droplet temperature prior to impingement, generated by short spraying distance, reduces the porosity level due to increased filling

THERMAL SPRAY COATINGS

19.35

of the cavities by the molten particles.76 In other words, the apparent increase in the porosity level is primarily attributed to the reduction in impact energy and temperature prior to impingement,77 shadowing effect (unmelted particles/spray angle), and shrinkage and stress-relief effects.19 Plasma-sprayed ceramic coatings are usually associated with porosity and some microcracks, which may be both beneficial and harmful to the performance of ceramic coatings. For example, pores can serve as crack arresters, and thus improve fracture toughness, whereas cracks give rise to either transformation toughening (e.g., for PSZ) or fracture initiation and fatigue failures in the part.36 Unmelted Particles. These can sometimes be observed in the coating microstructure, due to partial melting during the spraying process.13 Unmelted particles seem to break the chemical homogeneity and the overall microstructure. In many instances, unmelted particles reduce the cohesive strength of the coating.13 They also create shadow porosity.31a Oxides. Oxidation of thermal sprayed metals can significantly affect the microstructure, phase composition, properties, and performance of sprayed coatings. In many applications, metal oxides (often dominated along splat boundaries) can be detrimental toward corrosion, strength, and machinability.19 (Some oxides are lubricious.31a) The extent of oxidation occurring during spray coating is a function of the material being deposited, the method of deposition, and the particular deposition process. Oxidation may occur due to the oxidizing potential of the fuel-gas mixture in flame spraying, HVOF, or detonation gun spraying or due to air inspirated into the gas stream in plasma spraying or any other methods. It is noted that the latter cause can be corrected by using inert-gas shrouds or low-pressure chambers with plasma spraying. Use of carbon-rich gaseous mixtures in oxy-fuel processes can produce carburization rather than oxidation with some metallic coatings. Metallic coatings are perhaps most prone to oxidation, but carbide coatings may suffer a remarkable loss of carbon that is not especially prominent in metallographic examination. Oxidation during deposition can result in higher porosity, weaker coatings due to brittleness, incompatible thermal coefficients of expansion, and disruption of the chemical uniformity of surfaces exposed to a corrosive environment.1,78 Most of the thermal spray processes result in very rapid quenching of the particles on impact. Quench rates have been approximated to be 106 to 108 °C/s for metallics and 104 to 106 °C/s for ceramics. Consequently, the materials deposited may be in thermodynamically metastable states, and the grains within the splats may be submicron in size or even amorphous. The metastable phases present may not possess the predicted characteristics, especially corrosion characteristics, of the material, and this factor should be borne in mind during the selection of coating compositions.1 Cooling and solidification of most materials are associated with shrinkage or contraction. The tensile strength in the coating increases with coating thickness up to a level exceeding the bond or cohesive strength and leads to coating failure. Highstrength materials such as austenitic stainless steels are prone to a high degree of stress buildup and are therefore limited to low coating thicknesses. Usually thin coatings are more durable than thick coatings.19 The stress buildup in coatings is a function of the spraying process and coating microstructure. Dense coatings usually exhibit more stress buildup than porous coatings. It should be noted that FS coatings usually have greater thickness limitations than PS coatings. On the other hand, the systems with high kinetic energy and low thermal energy [HVOF and high-energy plasma (HEP)] can produce very dense and relatively stress-free coatings. This is attributed to the compressive stresses

CHAPTER NINETEEN

19.36

developed from mechanical deformation (as in shot peening) during particle impact deposition offsetting the tensile shrinkage stresses due to cooling and solidification.19

19.5.2

Mechanical Properties

The mechanical properties of thermal spray coatings are not well documented except for their hardness and bond strength. The bond strength between the coating and the substrate depends on the selection of coating parameters, the properties of the materials used for spraying, the thermophysical properties, and the surface conditions of the substrate materials.13 Substrate temperature during spraying has a significant effect on the mechanical properties (strength and ductility) of sprayed coatings. Substrate preheating improves the adhesion of ceramic coating and also modifies the splat morphology. The splat morphology and particularly the splat/splat and splat/substrate interface are critical to properties such as wear, erosion, corrosion, and bond strength.5 Coating hardness seems to be determined by the highest temperature to which the coatings are subjected. Industrially, rough surfaces for thermal spray coating adhesion are usually made by grit blasting, or high-pressure water-jet. EDM* machining of Al alloys can be used as an industrial surface preparation method prior to thermal spray coating for higher coating/substrate adhesion strength.79 Table 19.12 lists the typical mechanical properties data for a wide range of plasma-sprayed coatings.5 However, the sensitivity of the properties of the coatings to particular deposition parameters makes general cataloging of properties by simple chemical composition and common process (e.g., WC-12Co by plasma spray) practically meaningless. The situation becomes more complex because the properties of coatings on test specimens may vary somewhat from those on parts due to differences in geometry and thermal conditions. However, coatings made by competent suppliers using adequate quality control will be quite reproducible, and thus the measurement of various mechanical properties of these standardized coatings may be of great importance in the selection of coatings for specific applications. Properties that may be of importance include the modulus of elasticity, strain-tofracture, modulus of rupture, and hardness. Examples of some of these are listed in Table 19.13.1 The anisotropic structure of coatings results in a difference in mechanical properties in the longitudinal and transverse directions. Strength in the longitudinal direction can be 5 to 10 times that of the transverse direction.

19.6 APPLICATIONS Thermal spray technologies have expanded their applications to many industries. The thermal sprayed coatings offer various properties such as tribological (wear resistance, abradable or abrasive wear resistance, corrosion resistance, oxidation resistance, and heat resistance), thermal behavior, electrical conductivity or resistivity, textured surfaces, dimensional restoration, copying of intricate surfaces, * Electric discharge machine or electric discharge machining.

THERMAL SPRAY COATINGS

19.37

TABLE 19.12 Typical Mechanical Properties of Plasma-Sprayed Coatings5 Bond tensile strength* Material

MPa

ksi

Rockwell macro/micro hardness

Pure metals Aluminum Copper Molybdenum (fine) Molybdenum (coarse) Nickel (fine) Nickel (coarse) Niobium Tantalum Titanium Tungsten

8.3 21.4 57.2 55.2 23.4 33.1 54.5 46.9 41.4 40.0

1.2 3.1 8.3 8.0 3.4 4.8 7.9 6.8 6.0 5.8

45/58 HRH 65/142 HRB 70/1450 HR15N 65/1448 HRA 84/ . . . HR15T 81/ . . . HR15T 61/1344 HRC 65/1585 HRA 78/ . . . HR15N 50/500 HRA

2.48 7.20 9.90 8.96 7.95 7.48 7.06 14.15 4.17 16.90

155 449 618 559 496 467 441 883 260 1055

Alloy metals 304 stainless 316 stainless 431 stainless 80Ni-20Cr (fine) 80Ni-20Cr (coarse) 40Ni-60Cu 35Ni-5In-60Cu 10Al-90Cu (fine) 10Al-90Cu (coarse) Hastelloy 31 (fine) Hastelloy 31 (coarse) 5Al-95Ni 20Al-80Ni 6Al-19Cr-75Ni 12Si-88Al 5Al-5Mo-90Ni Hastelloy X Hastelloy C 420 stainless 0.9C stainless Cast iron Ti-6Al-4V Monel 0.2C steel

17.6 23.4 31.0 31.0 29.0 24.1 24.1 28.3 22.1 41.4 23.4 68.3 47.6 49.6 16.5 37.9 42.7 42.1 22.1 33.8 35.9 33.1 44.8 22.1

2.55 3.4 4.5 4.5 4.2 3.5 3.5 4.1 3.2 6.0 3.4 9.9 6.9 7.2 2.4 5.5 6.2 6.1 3.2 4.9 5.2 4.8 6.5 3.2

88/ . . . HR15T 70/ . . . HR30T 35/ . . . HRC 90/ . . . HR15 90/ . . . HR15T 72/ . . . HRB 83/ . . . HR15T 88/ . . . HR15T 81/ . . . HR15T 79/ . . . HR15T 79/ . . . HR15T 80/490 HRB 80/510 HRB 90/250 HRB 78/60 HR15T 80/200 HRB 89/ . . . HR15T 90/ . . . HR15T 70/ . . . HR15N 35/ . . . HRC 28/ . . . HRC 35/ . . . HRC 35/ . . . HR15N 95/ . . . HRB

7.22 6.80 6.25 7.48 7.19 7.89 7.94 6.73 6.30 7.65 7.83 7.51 6.92 7.51 2.49 7.43 7.65 8.25 7.10 7.05 7.00 4.30 8.50 6.90

451 425 390 467 449 493 496 418 393 478 489 469 432 469 155 464 478 515 443 440 437 268 531 431

Metal composites 95Ni-5Al 80Ni-20Al 65Ni-35Ti 75Ni-19Cr-6Al 75Ni-9Cr-7Al-5Mo-5Fe 90Ni-5Al-5Mo

33.8 32.4 32.1 42.7 27.6 48.3

4.9 4.7 4.65 6.2 4.0 7.0

80/500 HR15T 86/500 HR15T 72/660 HR15N 92/250 HR15T 80/250 HRB 80/200 HRB

7.39 7.02 6.62 7.71 6.90 7.40

461 438 413 481 431 462

Density g/cm3

lb/ft3

19.38

CHAPTER NINETEEN

TABLE 19.12 Typical Mechanical Properties of Plasma-Sprayed Coatings5 (Continued) Bond tensile strength* Material

MPa

ksi

Rockwell macro/micro hardness

Carbide powders and blends 88WC-12Co (cast, fine) 88WC-12Co (cast, coarse) 88WC-12Co (sintered) 83WC-17Co 75Cr3C2-25NiCr (fine) 75Cr3C2-25NiCr (coarse) 75Cr3C2-25NiCr (composite) 85Cr3C2-15NiCr

44.8 44.8 55.2 68.9 41.4 34.5 — —

6.5 6.5 8.0 10.0 6.0 5.0 — —

88/ . . . HR15N 81/ . . . HR15N 85/ . . . HR15N 85/950 HR15N 84/950 HR15N 80/1850 HR15N . . . /1850 HR15N 80/1850 HR15N

Ceramic oxides ZrO2 (calcinated) Chromium oxide 80ZrO2-20yttria TiO2 Al2O3 (white) 87Al2O3-13TiO2 60Al2O3-40TiO2 50Al2O3-50TiO2 Al2O3-gray (fine) Al2O3-gray (coarse) Magnesium zirconate

44.8 44.8 15.2 — 44.8 15.5 27.6 — 6.9 — 17.2

6.5 6.5 2.2 — 6.5 2.25 4.0 — 1.0 — 2.5

70/ . . . HR15N 90/ . . . HR15N 80/ . . . HR15N 87/ . . . HR15N 90/ . . . HR15N 90/850 HR15N 85/ . . . HR15N 87/193 HR15N 85/ . . . HR15N 75/ . . . HR15N

Density g/cm3

lb/ft3

13.75 12.41 14.55 11.10 6.41 6.23 — 5.80

858 775 908 693 400 389 — 362

5.30 4.80 5.00 4.10 — 3.50 3.50 4.0 3.30 3.30 4.20

331 300 312 256 — 218 218 250 187 187 262

* Over a grit-blasted surface roughened to 2.5 to 4.1 mm (100 to 160 min.) AA (arithmetic average). Source: Ref. 28 in Ref. 5. Reprinted by permission of ASM International.

catalytic and prosthetic properties, and so forth, based on metallic, carbidecontaining, ceramic, or composite materials.19,29 Wear Resistance. Thermal spray coatings are used to resist practically all types of wear, such as abrasive, erosive, and adhesive, in virtually any type of industry. The materials used vary from soft metals to hard metal alloys to carbide-based cermets to oxides. Usually, the wear resistance of the coatings increases with their density and cohesive strength; for example, the higher-velocity coatings such as HVOF and especially D-gun® coatings offer the best wear resistance for a given composition, in contrast to plasma spray coatings (Table 19.14). Table 19.15 shows examples of erosive wear data, obtained by various laboratory tests for some thermal spray coatings.1 Wear-resistant WC-M (M = Ni, Co, Cr, Mo, or Co-Cr) coatings, produced by HVOF or APS techniques, are used to reduce wear or modify friction in many sliding, abrasive, and corrosive applications such as compressor piston rods, pump plungers, shaft sleeves on centrifugal pumps and fans, and midspan dampers on jet engine compressor fans and blades.80 These coatings are primary candidates for replacing chrome plating for use on aircraft landing gear components. The plain and alloyed WC-Co and Cr3C2-NiCr cermet coatings using HVOF,

TABLE 19.13 Mechanical Properties of Representative Plasma, D-Gun®, and High-Velocity Oxy-Fuel Coatings1 Type of coating: Tungsten-carbide-cobalt

Parameter

Alumina

Nominal composition, wt%

W-7Co-4C

W-9Co-5C

W-11Co-4C

W-14Co-4C

Al2O3

Al2O3

Thermal spray process

Detonation gun

High-velocity combustion

Plasma

Detonation gun

Detonation gun

Plasma

19.39

Rupture modulus, 103 lb/in.2*

72

—

30

120

22

Elastic modulus, 106 lb/in.2*

23

—

11

25

14

17 7.9

Hardness, kg/mm2, HV300

1,300

1,125

850

1,075

>1,000

>700

Bond strength, 103 lb/in.2†

>10,000‡

>10,000‡

>6500

>10,000

>10,000‡

>6500

* Compression of free-standing rings of coatings. † ASTM C633-89, “Standard Test Method for Adhesion or Cohesive Strength of Flame-Sprayed Coatings,” ASTM, 1989. ‡ Epoxy failure. Source: Publication 1G191, National Association of Corrosion Engineers. Reprinted by permission of ASM International.

CHAPTER NINETEEN

19.40

TABLE 19.14 Abrasive Wear Data for Selected Thermal Spray Coatings1 Material Carballoy 883 WC-Co WC-Co WC-Co WC-Co

Type

Wear rate, mm3/1000 rev

Sintered Detonation gun Plasma spray Super D-Gun HVOF

1.2 0.8 16.0 0.7 0.9

ASTM G65 dry sand/rubber wheel test, 50/70-mesh Ottawa silica, 200 rpm, 30-lb load, 3000-revolution test duration. Reprinted by permission of ASM International.

TABLE 19.15 Erosive Wear Data for Selected Thermal Spray Coatings1 Material Carballoy 883 WC-Co WC-Co AISI 1018 steel

Type

Wear rate, mm/g

Sintered Detonation gun Plasma spray Wrought

0.04 1.3 4.6 21

Silica-based erosion test; particle size, 15 mm; particle velocity, 139 m/s; particle flow, 5.5 g/min; ASTM Recommended Practice G75. Reprinted by permission of ASM International.

plasma, and D-Gun® processes are more popular (for protection against abrasive fretting and erosive wear environments). D-Gun® and HVOF-Cr3C2-NiCr cermet coatings are commonly used due to their high-temperature (530 to 815°C) wear and erosion protection applications in the aircraft engine industry such as for exhaust flaps on turbine engines, turbine compressors, midspan stiffeners, pump seals and liners, and knife edge seals. Such coatings using DG processes are used in numerous industrial sectors such as steel plant machinery, printing rolls, and the petrochemical industry.36 WC-Co cermet coating is preferably used in service conditions below 530°C due to the degradation of WC by oxidation.81 Recently, TiC-Ni coatings using HVOF and D-Gun processes have generated more interest due to their combination of wear and corrosion resistance, highdensity, low-friction properties, good sprayability by different thermal spray processes, low production cost, etc., using optimized spray conditions and powder compositions.82 They appear to be an alternative to plain and alloyed WC-Co and Cr3C2-NiCr coatings which, in APS, are subjected to phase changes due to oxidation and decarburization. Mo-based coatings have been used for several adhesive wear conditions. Flamesprayed Mo-wire and plasma-sprayed Mo-pseudo-alloys are employed in the automotive, paper and pulp, and aerospace industries. For example, Mo-based coatings find applications in automotive piston rings to provide scuff resistance and to decrease adhesive sliding wear.36,83 FeCrMo coating is used for adhesive wear resistance on large-bore cylinder liners and stamping dies, NiCrMo coating for both wear and corrosion resistance, and

THERMAL SPRAY COATINGS

19.41

FIGURE 19.14 Steady-state erosion rates versus constituent composition.86 (Reprinted by permission of The Metallurgical Society.)

WC-Co/NiCrMo coatings for a high degree of abrasive and/or adhesive wear resistance on rotors for positive displacement pumps, and on knife edges.84 Co-based alloys such as CoCrW and CoCrMo alloys, when deposited by HVOF and PTA processes, are expected to have superior resistance to a combination of harsh environments such as wear and corrosive conditions at elevated temperatures. They provide better erosion and erosion-corrosion properties than the Ni-based metal phase coatings. The metal composition providing the best erosion and corrosion properties is an 8.5Cr-6.5Co matrix.85 A comparison between the post-sprayed powder and the actual coating microstructure has revealed that the retention of the FeCrAlY matrix is much better than that of the Cr3C2 particles, which can form oxides during HVOF thermal spraying.86 Erosion tests show that both oxides and carbides increase the erosion rate of the coating (Fig. 19.14) and that the small amounts of hard constituents are desirable for erosion resistance.5,86 The use of chromia and Al2O3 coatings have some drawbacks: toxicity of certain chromium oxides, the likelihood of metallic Cr in chromia coatings, and the brittleness of pure Al2O3 coatings.87 D-Gun® sprayed Al2O3-ZrO2-TiO2 coatings have been reported to exhibit very high abrasive wear resistance; the corresponding APS coatings have also shown superior abrasion wear resistance to plain APS-Al2O3 coatings.88 Friction Control. Many industries such as the automotive, aircraft, and pulp and paper industries, require sprayed coatings to reduce the frictional coefficient and improve the scuff resistance properties of the substrate material. Mo and Mo-based alloys are usually selected for these applications.89 Thermal spray coatings are employed in some applications to give specific frictional characteristics to a surface, ranging from low to high friction (with a coefficient of friction up to 1.42 for the plasma-sprayed WC-Co coatings).90 The textile industry offers, as an example, a variety of applications involving the complete range of friction characteristics and surface topographies to handle very abrasive synthetic fibers. Oxide coatings such as alumina are mostly used with surfaces that change from very smooth to very rough, depending on the coefficient of friction required.1 Low-friction Co-based coatings find specific applications in the protection against adhesive and fretting wear, galling and seizure of gas turbine and jet engine parts, or the like from Ti alloys.91 The application of APS-Al2O3 and/or Cr2O3 coatings allows the replacement of

19.42

CHAPTER NINETEEN

steel by Al as the rotating drum material. The APS-Al2O3-TiO2 coating is used for Al friction dampers in buildings to decrease their failures during or after earthquakes. Corrosion Resistance. Flame-sprayed Zn, Al, Zn-Al, and Zn-Al-Mg alloy coatings (in the 5- to 20-mil thickness range) are often used to provide long-lasting, lowmaintenance protection from direct chemical and sacrificial galvanic corrosion (with or without a passive barrier layer)92 on bridges, ships, tanker trucks, hulls of fishing vessels, the interiors of steel fresh water tanks and conduits, and other large steel structures. Other corrosion-resistant applications for thermal spray coatings are oxidation and sulfidation resistance in power boilers, shielding exhaust manifolds and stacks, industrial flue gas stacks and ducts, automotive valve seats, turbine blades of jet engines, and cylinder heads of marine diesel engines.93 In a comparison of several different thermal spray processes, it was reported that high-temperature corrosionresistant coatings must incorporate compositions that favor the formation of protective oxides at splat boundaries, be sufficiently dense to form protective oxides within and to fill voids, and be sufficiently thick to postpone the diffusion of corrosive species to the substrate material along the fast diffusion paths of the coating porosity.94 Infrastructure Maintenance. For infrastructure maintenance, thermal spraying is used for underwater tunnels, river dams, concrete and steel gates on a river lock and gate system, river dams, steel piping running underground and aboveground, off-shore oil rigs, water towers, railroad cars, and ships.36 Dimensional Restoration. Thermal spray coatings are frequently used to restore the dimensions of worn parts (such as valves and giant turbines used to regulate flow at hydroelectric dams, fuser rolls in copier machines, nonrotating air seals in aircraft turbine engines, carbon steel railway motor turbocharger shafts, and surfaces of printing press cylinders.93 Sometimes, a coating with low residual stress and/or low cost is employed to restore the worn area, followed by the application of a thin, more wear-resistant coating over it. In any buildup application, it should be noted that the properties of the coatings are perhaps quite different from those of the substrate, and that the coating will not contribute any structural strength to the part. Obviously, if care is not exercised, the coating may reduce the fatigue strength of the part.1 Thermal Barrier Coatings. The conventional duplex thermal barrier coating (TBC) comprises a metallic bond coat (typically, MCrAlY usually applied by VPS or EB-PVD) and a thick (≥125 mm) ceramic top coat (CaO-, MgO-, Y2O3-, or CeO2stabilized ZrO2 by APS or by EB evaporation). This class of ceramic has a low thermal conductivity to effectively insulate the underlying superalloy substrate from the high-temperature environment. For gas turbine applications, the thickness of the TBCs lies in the range of 125 to 250 mm (0.005 to 0.010 in.). At this thickness range, for hot sections operating at temperatures around 1000°C, the surface of the superalloy component can be lowered by ~100°C, thereby extending the lifetime at the turbine temperature or allowing the turbine to function at a higher, more efficient temperature. The tenacious feature of the TBC is attributed to the porosity, microcracks, and inherent toughness of the specific ceramic (YSZ).36 However, ceramic coatings sprayed onto metals have certain drawbacks: (1) high residual and thermal stress and (2) relatively low bond strengths. Surface cracking and debond-

THERMAL SPRAY COATINGS

19.43

FIGURE 19.15 Concept of multi-layered thermal barrier coatings for gas turbine components used at high temperature.95 (Reprinted by permission of ASM International.)

ing have been common experiences of mechanical failure. One way to overcome these problems may be to fabricate functionally gradient materials (FGMs). Figure 19.15 exhibits the concept of a multi-layered TBC on a superalloy substrate with different unique functions corresponding to the individual layers.95 Thermal cycle tests have established that the lifetimes of four-layer TBC are about twice those of two-layer TBC comprising a bond layer and a ceramic layer. The temperatures of substrates with the four-layer TBC are reported to be about 95°C lower than those of uncoated substrates.96 Functionally Graded Materials (FGMs).* Functionally graded materials with either continuously or stepwise variations of composition and/or microstructure provide solutions to numerous engineering problems confronting coating systems with large differences in the coefficient of thermal expansion (CTE). A classical example is the TBCs, in which large differences in the CTEs between the substrate and coating can result in failure during the thermal cycle. By continuously grading the composition of the coating from the metallic bond coat at the substrate/coating interface to that of the ceramic TBC at the outer surface, the stresses resulting from mismatch are reduced. One of the major goals in the fabrication of FGMs is that the final structure should vary in a regular and consistent fashion.97,98 The growing applications of FGMs include turbine components, rocket nozzles, chemical reactor tubes, burner nozzles, molds, furnace walls,99 advanced batteries, solid oxide fuel cells (SOFCs) (to produce power cost-effectively),100 thermoelectric devices (comprising alternate layers of semiconductor materials like FeSi2), and tubular laminate-superalloy composite gun barrels.101 Examples of thermally sprayed FGMs for burner nozzle applications and for thermoelectric devices are shown in Figs. 19.16 and 19.17.11,12 Electrical Applications. Like thermal properties, the electrical conductivity of thermal spray coatings is anisotropic and lower than that of their wrought or * The fabrication of FGMs can also be realized by powder metallurgy which allows dissimilar materials to be integrated while minimizing the stress and allowing normally compatible properties, such as hardness and corrosion resistance, to be incorporated in the same material.

19.44

CHAPTER NINETEEN

FIGURE 19.16 Schematic of a thermally sprayed FGM for burner nozzle applications.11 (Reprinted by permission of ASM International.)

FIGURE 19.17 Other applications of thermally sprayed FGMs for thermoelectric devices: (a) strain-control coating, (b) sprayed electrodes for SOFCs, and (c) plasma-sprayed alternating layers.11 (Reprinted by permission of ASM International.)

THERMAL SPRAY COATINGS

19.45

sintered counterparts due to their lamellar microstructure and porosity. Metallic or conductive cermet coatings are, however, employed as electrical conductors where both wear resistance and electrical conductivity are required. Thermal spray oxide coatings are used as electrical insulators, but it is essential to seal the coating to prevent moisture, even from the air, from penetrating the coating and reducing its insulating propensity. Thermal spray coatings have also been used to make hightemperature thermocouples and strain gages. Electromagnetic or radio-frequency shielding can also be achieved by flame or electric-arc spray coatings of zinc, tin, or other metals.1 Electronics Industry. Examples of applications here include: metal- (Cu, Al, steel, KovarTM) ceramic (Al2O3) substrates for hybrid microelectronics; plasma-sprayed sputtering sources of Ta-Hf nitrides and YBa2Cu3Ox targets (in the deposition of high-quality PVD films); microwave integrated circuits (MICs) (such as using plasma-sprayed Mg-Mn ferrite inserted into an Al2O3 substrate and striplines with sprayed ferrites and dielectrics); capacitor electrodes (such as APS-Al coating in a small-size double-layer capacitor); heater rolls [such as APS-NiAlMo (bond coat)/MgO.Al2O3 (electrical insulating spinal coat)/(Cu-Zn ferrite, cermet TiO2 + NiCr, or cermet Al2O3 + NiCr) heater coat] for use in energy-saving small office machines (such as laser printers, photocopier machines, faxes); conductor paths for hybrid electronics (such as VPS- and APS-copper coatings used as conductor paths, on sintered Al2O3 substrates); and plasma-sprayed Al2O3-28% MgO coated integrated circuit (IC) brackets for electrical insulation.13 Energy Industry. Boilers in power generation plants (APS- or HVOF-310 stainless steel coatings or APS-NiCr coatings on boiler tubes); MHD generators (APSTBC deposit comprising NiCrAl, 25 mm thick as the base coat and Y2O3-stabilized ZrO2 (YSZ), 100 to 150 mm thick, as the top coat for conversion of thermal energy of a plasma (1800 K, 800 to 900 m/s) to electricity; stationary gas turbines [VPSpowder (Ni-38Cr-11Al-0.25Y) coatings on stationary gas turbine (Udimet 520 alloy) blades]; SOFCs (such as VPS-La0.84Sr0.16MnO3 coatings on porous sintered ceramics with APS-perovskite coatings as the cathode, dense ZrO2-8% Y2O3 as a solid electrolyte, and ZrO2-30 vol% Ni composites as the anode, for conversion of chemical into electrical energy with a high fuel utilization factor of 87.1% and a high power generation efficiency of 38%)102; and perovskite as the oxidizer electrode (cathode), rf induction plasma spray coating of a dense Y2O3-stabilized-ZrO2 (YSZ) membrane as the electrolyte, and Ni/YSZ cermet as the fuel electrode (anode).103 Petrochemical and Chemical Industries. Tools in petroleum search installations such as super D-Gun® spray-WC-15 Co powder coatings on drill bit cones; flame spray and fuse (with powder self-fluxing alloy of composition, Ni-43.5W-6Cr-1.35B1.9Fe-6.25Co-1.8Si-3.1C) coatings on polycrystalline diamond cutters and postspraying fusion in furnace; super D-Gun® spray-powder coatings of composition W-20 Cr-7Ni-6C on rotors; APS-Cr2O3 coatings on drilling components (diameter ~45 mm) made of hardened steel (Rc 52 to 59) for substantially prolonged service lifetimes; HVOF-316 stainless steel coatings on the chemical refinery vessels to provide a good corrosion resistance against sulfur and NH3;13 HVOF-sprayedstainless steel 316L powder coatings (760 mm thick) and Hastelloy C-276 coatings on the ends of gas-well tubing for the extension of the corrosion life of vessels in chemical refineries and gas-well tubing in gas research drilling, respectively.13 Thermal spray Ti coatings are being used in various applications, especially to produce corrosion-resistant surfaces in the chemical process industries.

19.46

CHAPTER NINETEEN

Automotive Industry. Examples of applications here are: APS-Al2O3 coating on aluminum midplates for diode assembly in automotive alternators to provide resistance against salt corrosion and moisture absorption; pinion shafts; ZrO2-coated disk brake pads; electric arc spray coating of Fe-C wire on Al valve lifters; turbocompressor housing (of combustion engines); cylinder head gaskets; AlSi-50% Mo plasma spraying of synchronizer ring and hydraulic torque converter pump impeller; plasma-sprayed ceramic (Al2O3 + TiO2) coating at distributor rotors’ discharge tips (to reduce electromagnetic interference and suppress noise); and thermal sprayed Fe3O4 coating on torque sensors.104 Aerospace Industries. Thermal barrier coatings have found widespread applications in the aerospace and other industries: TBCs of 8 wt% Y2O3-stabilizedZrO2 (top ceramic layer, 1000 mm thick), CoCrAlY as bond coatings (and also as corrosion-resistant layers) and particulate composites containing volume ratios of 85/15 and 40/60 of both components and thickness ~500 mm as intermediate coatings for gas turbine combustors, shrouds, blades, nozzles, and vanes, and on internal combustion cylinders and valves as well as for piston crowns and cylinder heads in adiabatic diesel engines. In other cases, they may be used to diffuse heat as either surface conductors or thermal emitters. Because of their unique lamellar microstructure and porosity, the thermal conductivity of thermal spray coatings is usually anisotropic and significantly lower than that of their wrought or sintered counterparts.1 The duplex TBCs, which comprise a MCrAlY bond coating and a Y2O3, CaO, or MgO-stabilized-ZrO2 top coating, are useful for having high durabilities. The concept of multilayered TBCs is shown in Fig. 19.15.105 The advantages and improvements by the use of TBCs on various parts of gas turbines (such as blades, nozzles, vanes, and combustion chambers) are (1) an increase of engine power and efficiency due to increased operating temperature and increased thermal cycling (between ambient and operational temperatures over 1100°C) lifetime, (2) an increase in compressor efficiency due to a reduced air flow for turbine cooling, and (3) a longer service life of the metallic substrate material due to a decreased thermal fatigue load.106 MCrAlY alloy coatings are used for increased oxidation and hot-gas corrosion resistance of components operating in high temperatures and corrosive environments. For example, LPPS®-sprayed MCrAlY alloy coatings are used in many hot parts in gas turbines and aircraft engines as the undercoatings of thermal barrier coatings.107 High-temperature materials such as gas turbine blades and diesel engine components are usually protected by plasma spray or PVD techniques with an intermediate NiCoCrAlY alloy bond coating to improve adhesion strength and to decrease oxidation. The primary TBC material is (6 to 8%) YSZ for thermal insulation in the hot areas of gas turbine components. They are characterized by considerable toughness, low micro-porosity, high melting points, low thermal conductivity, and good thermal shock resistance.108 Iron and Steel Industries. Applications here include the following: APS-Al2O3 + 25% ZrO2 coating on cooling rolls of a continuous annealing line (CAL), with the increased roll lifetime from 3 months (for a roll with an electroplated Cr coating) to up to 2 years; APS-(Al2O3 + TiO2) (300-mm thick) coating on stave cooling pipe of blast furnace shells for extended protection against carburization; D-Gun flamesprayed CoCrAlY-Y2O3/CrB2 coating on hearth rolls for an extended service life to 6 years (compared to 1.5 to 3 years with the conventional ceramic/cermet coating)109;

THERMAL SPRAY COATINGS

19.47

thermal spray (>1 mm thick) coating on chute liners, grizzly bars (subjected to abrasive wear by ore fines), sintering fans and cooler fans (subjected to attack by dustladen gas), both at sinter plants109; and HVOF-WC-cermet coating onto the rolling surface of steel rolls used in the steel industry to increase the abrasion and friction resistance of the rolls.110 Nonferrous Metals Industries. Applications in nonferrous metals industries include the following: Multicoating with a 50-mm NiCr bond coating, 100-mm Al2O3 top coat and between them three 100-mm particulate composite coatings using NiCr and Al2O3 powders blended in ratios of 25 : 75, 50 : 50, and 75 : 25 for hot extrusion dies for extended life and remarkable cost savings; thermal spray Cr2O3-TiC composite/ NiCrAl bond protective coatings onto mild steel substrate against liquid copper, for an extended life up to 5 hr; multicoating with a 400-mm-thick Mo coating sprayed with IPS under Ar atmosphere as the bond coat, a 200-mm-thick particulate composite of Mo + 50 wt% ZrO2 (stabilized with calcia), and a 400-mm-thick ZrO2 stabilized with calcia as the top coat for protection against liquid zirconium13; HVOF-WC-Co cermet (particle size £45 mm) sprayed coatings (~200 mm in thickness) on mild steel sink rolls for protection against the molten zinc bath and for maintaining a smooth surface longer111; plasma spray ceramic [such as triple spinel (MgO-Al2O3-ZrO2) and zircon (ZrSiO4)] coatings on H13 steel and cast iron parts in squeeze casting machines, cast iron heating elements, thermocouple protectors, piston molds, and ladles (as replacements for ceramic and ceramic-slurry coated tools) for the longest life in molten aluminum and for lower aluminum casting operation costs.112 Shipbuilding Industry. Applications in the shipbuilding industry include the following: SPS- and VPS-Co-25Cr-10.5Al-2.5Hf-5Pt coating on the blades of the second-stage high-pressure turbines of the LMN2500 marine engine in a U.S. Navy test ship for more corrosion resistance (at high-temperature, low-contamination conditions) than provided by PVD coatings; APS-multicoating with a Ni-9Cr-7Al5.5Mo-5Fe bond coat, ZrO2-18TiO2-10Y2O3 top coat, and sealing on 410-type stainless steel valve stems of U.S. Navy aircraft carrier ships for refurbishment with extended performance life; and AS-aluminum coatings on nonskid helicopter flight decks of U.S. Navy ships for increased corrosion resistance and service life.13 Machine Building Industry (Textile, Pump Construction, Agroalimentary, etc.). Applications here include the following: APS-Cr2O3 coating on rotating aluminum drum, changing the direction of the fiber in the textile machine to achieve increased productivity and fiber velocity; APS-Al2O3-TiO2 coating onto the fiber guides; APSbiocompatible ceramic coating on piston and piston liners of agroalimentary pumps for transporting liquids (e.g., yogurt, chocolate, etc.) to achieve increased three-body abrasion (due to particles in the liquid) wear protection; and APS-multicomponent ceramic coatings on the surfaces of piston and piston liners of vacuum pumps for improved sliding adhesive wear.13 Printing Industry. Applications in the printing industry include the following: APS-Al2O3 (up to 2 mm thick) coating on Corona rolls for sufficient wear-resistant properties; APS-Cr2O3 coating on Anilox rolls (ARs) for applications in printing machines working with flexographic system. [The PlazJet (Praxair) high-power plasma-sprayed Al2O3-40% TiO2 and high-purity Cr2O3 coatings at 12 lb/hr (5.5 kg/hr) and 18 lb/hr (8.2 kg/hr), respectively, yield significant cost and time savings compared to conventional plasma spraying, providing important advantages in terms of paper making and print rolls.113]

CHAPTER NINETEEN

19.48

TABLE 19.16 Tentative Coating Specification for the Production of Anilox Rolls114 Condition number

Coating property

Property specification

1

Porosity

Low (e.g., 50 kW, and condition 6 by the use of appropriate grinding and polishing methods.13,114 Paper Industry. Examples of applications in the paper industry are as follows:APS multicoatings with Ni-6.5Al-6Mo bond coat (80 to 130 mm thick), 8% Y2O3- and 1.7% HfO2-stabilized (17% porous, 380-mm-thick) ZrO2, and 50-mm-thick, 5 to 7% porous ZrO2 of the same ZrO2 on Yankee dryers to eliminate the delamination of the paper (actually, thermally sprayed coatings on experimental rolls offer increased drying efficiency and production of good-quality paper); HVOF-WC-NiCr or WCCo (wear-resistant) coatings [of 100-mm thickness and mirror finish (Ra = 0.01 to 0.03 mm)] on gloss Calendar rolls for achieving two-years’ service lifetime; FS(kanthal M, i.e., Fe-22Cr-6Al alloy) wire coating on steel tubing in boilers* for superior corrosion-erosion resistance;13 plasma-sprayed Al2O3-2–4% TiO2 coatings (up to 10 mm thick and diamond-ground) on steel and cast iron rolls to impart high hardness and hydrophilic properties for more uniform release of processed paper with an improved worker safety environment.115 Decorative Coatings. Applications in decorative coating industries are the following: FS-copper coatings onto glass artware during the glass-blowing stage for achieving different color deposits; APS-ceramic oxide coatings (Al2O3 for white, TiO2 for gray, Al2O3-TiO2 for blue, Cr2O3 for black, and ZrO3 for yellow coloration).13

* Boilers are used in the pulp industry to fire black liquor and inorganic compounds. The boiler serves as a chemical reactor, in which the organic portion of the fuel, i.e., black fuel, is burned and the inorganic portion is reduced to sodium sulfide.

THERMAL SPRAY COATINGS

19.49

Mining Industry. Applications in the mining industry include the following: APS(400-mm) composite coatings [of 60 vol% self-fluxing NiCrSiB, 35 vol% bronze (Cu-10Sn), and 5 vol% MoS2] on the internal parts of hydraulic steel props used in coal mines for excellent corrosion resistance in aggressive mine water116 and for increased service life13; APS-Cr2O3 coatings on hardened-steel drilling components in the petroleum mining industry to increase the service life times of those parts13. Medical. Medical applications include the following: plasma-sprayed Zrreinforced hydroxyapatite [Ca2(PO4)6(OH)2]*117,118 composite coatings (of a chemical composition similar to that of bone) of thickness 50 to 400 mm with a range of porosity levels and compositions on (1) pure Ti or Ti-6Al-4V for orthopedic implants (for hip, knee, tooth, and other prostheses) and (2) ceramic substrates of ZrO2 and Al2O3 for use in artificial joints and dental roots.119 Ceramic Industry. Applications in the ceramic industry are as follows: APSceramic rolls and high-energy plasma (HEP) and sprayed ceramic (Y2O3) tubes as free-standing bodies; APS-WC-17Co coatings on replaceable mild steel wear plates to be applied on brick clay (cast mild steel) extruders; APS-W (630-mm), Ta-W, or Re-W coatings on graphite crucibles (to melt oxide ceramics such as Al2O3, Al2O3.ZrO2, and Al2O3.Y2O3) and HEP-sprayed membranes.13 Table 19.17 shows the properties of ceramic rolls prepared by thermal spraying.13 Nuclear Industry. The nuclear industry uses plasma-sprayed coatings of both B4C (moderator) and W in electron beam facilities and in advanced fusion devices for low erosion rates,120 and VPS-W or IPS-Be coatings onto stainless steels, thereby offering remote repair work on damaged walls of magnetic fusion energy devices. Miscellaneous Applications. A variety of other applications have been developed for thermal spray coatings as given below: 1. Coatings used as nuclear moderators, catalytic surfaces, parting films for hot isostatic presses, and freestanding components such as rocket nozzles, crucibles, and molds 2. Thermal sprayed Al2O3-based ceramic coated bicycle rims for improved brake performance and wear resistance 3. Thermal sprayed Al2O3 and WC coated and filled with a phenolic-based sealer to penetrate and fill the porosity in yacht sailboat winches for traction control111 4. Thermal sprayed Cr3C2/NiCr-coated brake poles for lawn mower brake clutches with improved wear resistance115 5. Thermal spray on stainless steel fan blades for land-based gas turbines.93 6. Thermal sprayed polymers (such as polyethylene, polypropylene, polyester, polyamides, polyvinylidine fluoride, polytetrafluoroethylene, and ethylene methacrylic acid copolymer)83 in a thickness range of 0.04 to 6.35 mm (0.002 to 0.25 in.), on a wide variety of substrate materials for applications such as plow blades, tank linings, pump impellers, external pipe coatings, structural steel coatings, transfer chutes, light poles, vacuum systems, etc. * Hydroxyapatite is classed as an excellent calcium phosphate bioceramic with unique bioactive properties that promotes rapid chemical bonding to natural bone (i.e., bone joint integrity), enhancing bone growth on its surface, offering long-term joint stability, and exhibiting osteoconductive properties.

TABLE 19.17 Properties of Ceramic Rolls Prepared by Thermal Spraying13

Powder characteristics

Process characteristics

Ceramic characteristics (after all treatments) Density, kg/m3

19.50

No.

Grain size, mm

1*

-180

Y2O3

APS

†

2

-180

Al2O3-0.02SiO2

HEP

1823 K/4 hr (air)

3580

a-Al2O3

3†

-200

Al2O3-22.3ZrO2

HEP

As above

3373

a-Al2O3, (t + m) ZrO2

Sillimantin 60TM, bal. Al2O3, 25–27SiO2

Sintering

4†

Chem. composition, wt%

Method

Postspraying

Phase content

Ceramic properties

E, GPa

Fracture stress sf, MPa

8

4450

* C. E. Holcombe, Jr., Ceramic Bull., vol. 57, 1978, p. 610. † E. H. Lutz, Powder Metallurgy International, vol. 25, 1993, pp. 131–137. Reprinted by permission of John Wiley & Sons, England.

TEC (300–1300), 10-6/K

165

79.4 ± 9.9

95

76 ± 4.6

56 ± 12

37.2 ± 2.1

7.4–9.0 7–9 4.6–5.7

THERMAL SPRAY COATINGS

19.51

Process Combinations 1. CO2-gas laser beam to thermal sprayed WC-12% Co coatings provides dense film, metallurgical bond to the base metal, and a higher hardness.121 2. (Laser beam) nitriding after plasma-sprayed Al2O3 coatings or Ti coatings show a lower porosity, smoother surface (only one finishing process), increased wear, and corrosion resistance.122 The advantages of special gas nitriding after plasmasprayed Al2O3 ceramic coatings are no masking during nitriding and no removal of nitrided layer. 3. Thermal sprayed coating used as an undercoat for painting steel structures, antiabrasive components for machine parts, repairing materials by overlay, and heatresistant materials, exploiting the advantages of its strong adhesion to substrate and superior applicability. 4. For corrosion protection purposes, however, the conventional thermal spray has been used as a sacrificial protection material, not as an insulating material from the environment due to the porous structure of the coating.123 5. Ni electron brush plating on the arc-sprayed coating for improved wear properties of the mold coating surface.124

REFERENCES 1. R. C. Tucker, Jr., ASM Handbook, vol. 5, 10th ed., ASM International, Materials Park, Ohio, 1994, pp. 497–509. 2. R. C. Tucker, Jr., Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 253–258. 3. S. Boire-Lavigne, C. Moreau, and R. G. Saint-Jacques, in Thermal Spray Industrial Applications, eds. C. C. Berndt and S. Sampath, ASM International, Materials Park, Ohio, 1994, pp. 621–626. 4. L. Pejryd, J. Wigren, and N. Hanner, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 445–450. 5. A. R. Marder, ASM Handbook, vol. 20: Materials Selection and Design, ASM International, Materials Park, Ohio, 1997, pp. 470–490. 6. M. Sawa and J. Oohori, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 37–42. 7. H. Kreys, T. Ahlorn, J. Niebergali, and R. Schwetzke, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 127–131. 8. T. Miyamoto et al., in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 3–8. 9. A. Kumar, J. Boy, R. Zatorski, and P. March, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 83–90. 10. S. Steinhauser, B. Wielage, U. Hofmann, and G. Zimmermann, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 491–497. 11. R. Knight and R. W. Smith, ASM Handbook, vol. 8, 10th ed., ASM International, Materials Park, Ohio, 1998, pp. 408–419.

19.52

CHAPTER NINETEEN

12. R. W. Smith and R. Knight, JOM, vol. 47, no. 8, 1995, pp. 32–39. 13. L. Pawlowski, The Science and Engineering of Thermal Spray Coatings, John Wiley & Sons, Chichester, England, 1995. 14. S. Midorikawa et al., in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 43–46. 15. H. Yara, K. Miyagi, and A. Ikuta, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 163–167. 15a. R. C. Tucker, Jr., private communication, 2001. 16. J. Youngchang and Y. Feng, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 827– 832. 17. I. Kretschmer, P. Heimgartner, R. Polak, and P. A. Kammer, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 199–202. 18. J. Sheard, J. Heberlein, K. Stelson, and E. Pfender, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 613–618. 19. Thermal Spray Coatings Page Index, Gordon England, Surrey, England. 20. Z. Zurecki, D. Garg, and D. Bowe, J. Thermal Spray Tech., vol. 6, 1997, p. 417. 21. K. Wira, C. W. Lim, and N. L. Loh, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 465–470. 22. T. Nakano, R. Uchino, and T. Kusano, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, 1995, pp. 15–20. 23. S. Y. Hwang and B. G. Seong, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 59–63. 24. J. M. Park, S. W. Lee, and B. K. Kim, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 121–126. 25. H. Takeda, Ceramic Coating, Nikkan-Kogyo, 1987, pp. 179–205; H. Yajima, Y. Kimura, and T. Yoshioka, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 621–626. 26. T. Priem, R. Ranc, E. Rigal, J. Giral, G. Olalde, and J. F. Robert, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 725–729. 27. J. Larsen-Basse and P. Kodali, in Surface Modification Technologies XII, eds. T. S. Sudarshan, K. A. Khor, and M. Jeandin, ASM International, Materials Park, Ohio, 1998, pp. 501–506. 28. E. Lugscheider, P. Remer, C. Herbst, K. Yuschenko, Y. Borisov, A. Chernishov, P. Vitiaz, A. Verstak, B. Wielage, and S. Steinhauser, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 235–240. 29. G. Barbezat, S. Keller, and K. H. Wegner, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 9–13. 30. A. S. Khanna, C. Coddet, C. S. Harendranath, and K. Anuja, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, pp. 577–583.

THERMAL SPRAY COATINGS

19.53

31. K. Niemi, P. Vuoristo, E. Kumpulainnen, P. Sorsa, and T. Mantyla, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 687–692. 31a. R. Knight, private communication, 2001 32. S. Keller, P. Tommer, R. Clarke, and A. R. Nicoll, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 275–281. 33. T. S. Srivatsan and E. J. Lavernia, J. Mat. Sci., vol. 27, 1992, p. 5965. 34. M. E. Vinayo, in 7th International Symposium on Plasma Chemistry, Eindhoven, Netherlands, 1985, p. 1161. 35.

M. L. Lau, Thermal Spraying of Nanocrystalline Materials, Ph.D. thesis, University of California, Irvine, 2000.

36. H. Herman and S. Sampath, Thermal Spray Coatings, Web Page, pp. 1–13. 37. K. Honda, I. Chida, M. Saito, Y. Itoh, and K. F. Kobayashi, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 411–416. 38. H. J. Kim, B. L. Choi, and S. H. Hong, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 435–440. 39. H. Yajima, Y. Kimura, and T. Yoshioka, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 621–626. 40. S. Oki, S. Gohda, E. Lugscheider, P. Jokiel, and R. W. Smith, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 561–564. 41. Y. Tsunekawa, M. Okumiya, and T. Kobayashi, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 755–760. 42. S. Sodeoka, M. Suzuki, T. Inoue, X. Ono, and K. Ueno, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 283–288. 43. R. J. DuMola and G. R. Heath, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 427–434. 44. P. J. Meyer, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 217–222. 45. R. Chattopadhyay, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 31–34. 46. A. J. Sturgeon, M. D. F. Harve, F. J. Blunt, and S. B. Dunkerton, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 669–673. 47. M. C. Nestler, H. M. Hohle, W. M. Balbach, and T. Koromzay, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 101–106. 48. H. Kreye, S. Zimmermann, and P. Heinrich, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 393–398. 49. M. Thorpe and H. Richter, J. Thermal Spray Technol., June 1992. 50. A. J. Sturgeon and M. D. F. Harvey, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 933–938.

19.54

CHAPTER NINETEEN

51. M. Dvorak and J. A. Browning, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 405– 409. 52. M. L. Thorpe and H. J. Richter, in Thermal Spray: International Advances in Coating Technology, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1992, p. 137. 53. J. A. DeBarro and M. R. Dorfman, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 651–656. 54. K. Kadyrov and V. Kadyrov, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 417–424. 55. G. Naisbitt, T. Alderton, and C. Bruce, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 59–63. 56. D. W. Parker and G. L. Kutner, in Adv. Mater. Process, vol. 140, 1991, p. 68. 57. K. Sakaki, Y. Shimizu, and N. Saito, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 301–306. 58. S. Y. Hwang, B. G. Seong, and M. C. Kim, in Proceedings of the 9th National Thermal Spray Conference, October 7–11, 1996, p. 107. 59. G. M. Guilemany and J. A. Calero, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 717–721. 60. H. Fukutome, H. Shimizu, N. Yamashita, and Y. Shimizu, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 21–26. 61. S. E. Hartfield-Wunsch and S. C. Tung, in Proceedings 1994, National Thermal Spray Conference, Boston, MA, June 20–24, 1994, ASM International, Materials Park, Ohio, pp. 19–24. 62. E. Petrovicova, R. Knight, R. W. Smith, and L. S. Schadler, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 877–883. 63. Y. Fukuda and M. Kumon, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 107–110-1. 64. K. Yushehenko, E. Astakhov, Yu Borisov, G. Holmberg, and P. Kaski, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 137–140. 65. A. Utsumi, J. Matsuda, M. Yoneda, M. Katsumura, and T. Araki, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 325–330. 66. R. B. Bhagat, M. F. Amateau, A. Papyrin, J. C. Conway, Jr., B. Stutzman, and B. Jones, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 361–367. 67. J. L. Bacon, D. G. Davis, R. L. Sledge, J. R. Uglum, R. C. Zowarka, and R. J. Polizzi, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 399–406. 68. W. M. Steen, The Encyclopedia of Advanced Materials, Pergamon Press, Oxford, England, pp. 2755–2761. 69. D. E. Wolfe, M. B. Movchan, and J. Singh, Advances in Coating Technologies for Surface Engineering, eds. C. R. Clayton et al., The Metallurgical Society, Warrendale, PA, 1997, pp. 93–110.

THERMAL SPRAY COATINGS

19.55

70. K. A. Khor, Y. W. Gu, C. H. Quek, and Z. L. Dong, Surface Modification Technologies XII, eds. T. S. Sudarshan, K. A. Khor, and M. Jeandin, ASM International, Materials Park, Ohio, 1998, pp. 495–500. 71. J. Carr and J. Jones, Post Densified Cr2O3 Coatings for Adiabatic Engine, SAE Technical Paper 840432, Society of Automotive Engineers Warrendale, PA, 1984. 72. H. Ito, R. Nakamura, and M. Shroyama, Surf. Eng., vol. 4, 1988, pp. 35– 38. 73. A. Ohmori, K. Aoki, S. Sano, Y. Arata, and N. Iwamoto, Thin Solid Films, vol. 207, 1992, pp. 153–157. 74. C. Burman, T. Ericsson, I. Kvernes, and Y. Lindblom, Surf. Coat. Technol., vol. 32, 1987, pp. 127–140. 75. D. Matejka and B. Benko, Plasma Spraying of Metallic and Ceramic Materials, John Wiley & Sons Ltd., West Sussex, UK, 1989. 75a. H. Hu, Z. H. Lee, D. R. White, and E. J. Lavernia, Met. and Mats. Trans., vol. 31A, 2000, pp. 723–733. 76. V. V. Sobolev, J. M. Guilemany, and A. J. Martin, in Proceedings of the 15th International Thermal Spray Conference, Nice, France, ed. C. Coddet, ASM International, Materials Park, Ohio, 1998, p. 503. 77. L. C. Erickson,T. Troczynski, and H. M. Hawthorne, in Conference Proceedings:Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 743–748. 78. M. F. Smith, R. C. Dykhuizen, and R. A. Neiser, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 885–893. 79. O. O. Popoola, M. J. Zaluzec, and H. Haack, Surface Modification Technologies XII, eds. T. S. Sudarshan, K. A. Khor, and M. Jeandin, ASM International, Materials Park, Ohio, 1998, pp. 75–84. 80. J. Wigren, L. Pejryd, D. J. Greving, J. R. Shadley, and E. F. Rybicki, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 113–118. 81. L. Russo and M. Dorfmann, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 681–686. 82. P. Vuoristo, T. Mantyla, L.-M. Berger, and M. Nebelung, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 909–915. 83. B. J. Taylor and T. S. Eyre, Tribology, April 1979, p. 79. 84. J. A. DeBarro and M. R. Dorfman, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 651–656. 85. T. Rogne, T. Solem, and J. Berget, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 113–119. 86. K. J. Stein, B. S. Schorr, and A. R. Marder, Elevated Temperature Coatings: Science and Technology II, eds. N. B. Dohrte and J. M. Hampikian, The Metallurgical Society, Warrendale, PA, 1996, p. 99. 87. E. Lugscheider, H. Jungklaus, P. Remer, and J. Knuuttila, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 833–838. 88. K. Niemi, P. Vuoristo, T. Mantyla, E. Lugscheider, J. Knuuttila, and H. Jungklaus, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 675–680.

19.56

CHAPTER NINETEEN

89. J. DeFalco, L. Russo, and M. R. Dorfmann, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt,ASM International, Materials Park, Ohio, 1998, p. 990. 90. Z. H. Tong, C. X. Ding, B. Qiao, and K. W. Huang, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 713–717. 91. K. Hajmrie and A. P. Chilkovich, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 127–130. 92. T. Lester, S. J. Harris, D. Kingerley, and S. Matthews, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 183–189. 93. Sultzer Metco, Thermal Spray Coatings Bulletin, Westbury, N.Y. 94. S. T. Bluni and A. R. Marder, Corrosion, vol. 52, 1996, p. 213. 95. M. Tamura, M. Takahashi, J. Ishii, K. Suzuki, M. Sato, and K. Shimomura, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 323–328. 96. Y. Kojima, K. Wada, T. Teramae, and Y. Furuse, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 95–99. 97. W. D. Swank, J. R. Fincke, D. C. Haggard, S. Sampath, and W. Smith, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 451–458. 98. J. R. Finke, W. D. Swank, and D. C. Haggard, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt,ASM International, Materials Park, Ohio, 1998, pp. 451–458. 99. A. H. Bartlett, R. G. Castro, D. P. Butt, H. Kung, J. J. Petrovic, and Z. Zurecki, Ind. Heat., vol. LXIII, no. 1, 1996, pp. 33–36. 100. F. Fendler, R. Henne, and M. Lang, in Proceedings of the 8th National Thermal Spray Conference, ASM International, Materials Park, Ohio, 1995, pp. 533–537. 101. L. J. Westfall, in Thermal Spray: Advances in Coating Technology, Proceedings of the Second National Thermal Spray Conference, ASM International, Metals Park, Ohio, 1988, p. 417. 102. A. Notomi and N. Hisatome, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 79–82. 103. K. Mailhot, F. Gitzhofer, and M. I. Boulos, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt,ASM International, Materials Park, Ohio, 1998, pp. 21–25. 104. T. Miytamoto and S. Sugimoto, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 3–8. 105. M. Takahashi, Y. Itoh, and M. Miyazaki, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 83–88. 106. H. D. Steffens, J. Wilden, and I. A. Josefiak, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, p. 988. 107. R. Yamasaki, J. Takeuchi, A. Nakahira, M. Saitoh, and Y. Itoh, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 863–868.

THERMAL SPRAY COATINGS

19.57

108. K. S. Ravichandran, K. An, and R. Taylor, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt,ASM International, Materials Park, Ohio, 1998, pp. 291–298. 109. M. Sawa and J. Oohori, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 37– 42. 110. Y. Matsubara and A. Tomiguchi, in 13th International Thermal Spraying Conference, Orlando, FL, May 28–June 5, 1992, ed. C. C. Berndt, ASM International, Materials Park, Ohio, pp. 637–641. 111. Y. Kobayashi et al., in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 211–216. 112. Y. Wang, in Surface Modification Technologies XII, eds. T. S. Sudarshan, K. A. Khor, and M. Jeandin, ASM International, Materials Park, Ohio, 1998, pp. 525–532. 113. G. Irons, D. Poirier, and A. Roy, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 205–209. 114. L. Pawlowski, R. Zacchino, R. Dal Maschio, V. M. Saglavo, J. Andersen, and F. J. Driller, in Thermische Spritzkonferenz, Aachen, Germany, March 3–5, 1993, pp. 132–138. 115. W. J. Lenling, P. R. Gilson, and D. L. Ohmann, in Surface Modification Technologies XII, eds. T. S. Sudarshan, K. A. Khor, and M. Jeandin, ASM International, Materials Park, Ohio, 1998, pp. 519–524. 116. D. Matejka et al., in 1st Plasma Technik Symposium, Lucerne, Switzerland, May 18–20, 1988, pp. 247–257. 117. K. A. Khor and P. Cheang, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, pp. 769–774. 118. A. J. Sturgeon and M. D. F. Harvey, in Conference Proceedings: Thermal Spraying— Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 933–938. 119. T. Kameyama, M. Ueda, K. Onuma, A. Motoe, K. Ohsaki, H. Tanizaki, and K. Iwasaki, in Conference Proceedings: Thermal Spraying–Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 187–192. 120. W. K. W. M. Mallener, H. Gruhn, and H. Hoven, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 229–233. 121. N. Takasaki, M. Kumagawa, K. Yairo, and A. Ohmori, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 987–992. 122. H. D. Steffens, J. Wilden, and C. Buchmann, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 981–986. 123. T. Suzuki, K. Ishikawa, and Y. Kitamura, in Conference Proceedings: Thermal Spraying—Current Status and Future Trends, ed. A. Ohmori, High Temperature Society of Japan, Osaka, 1995, pp. 1033–1038. 124. L. Xianjun and M. Dan, in Thermal Spray: A United Forum for Scientific and Technological Advances, ed. C. C. Berndt, ASM International, Materials Park, Ohio, 1998, p. 986.

APPENDIX A

CONVERSION TABLE FOR UNITS, CONSTANTS, AND FACTORS IN COMMON USE

A.1

APPENDIX B

TEMPERATURE CONVERSIONS

Look up temperature to be converted in middle column. If in degrees Centigrade, read Fahrenheit equivalent in right-hand column; if in Fahrenheit degrees, read Centigrade equivalent in left-hand column. (Source: Republic Steel.)

B.1

B.2

APPENDIX B

TEMPERATURE CONVERSIONS

B.3

B.4

APPENDIX B

TEMPERATURE CONVERSIONS

B.5

INDEX

Ablation lasers, 18.31, 18.32 Abnormal grain growth (see Secondary recrystallization) Absorbed fracture energy, 14.66 Accelerated cooling, 15.3, 15.5, 15.6, 15.24 Accommodation helper mechanism, 15.95 Acicular austenite, 10.6 Acicular ferrite, 9.24–9.29, 13.88, 14.19, 15.33 characteristics and morphology, 9.24, 9.25 effect of allotriomorphic ferrite, 9.27, 9.28 growth mechanism, 9.25–9.27 lattice matching theory, 9.25, 9.28 role of inclusions and others on nucleation of, 9.28, 9.29 Activation energy, 4.30, 15.17 barrier, 4.30 Affine transformation, 8.4 Age hardening (see Precipitation hardening) Aging temperature, 6.30 Air cooling (see Cooling rate) Air hardening steels, 1.47, 12.23, 13.61, 13.65, 17.31 AISI designation, 1.36–1.38, 1.41, 1.48–1.50, 10.25, 10.28 AISI-SAE designations, 1.52, 1.53 Allotriomorphic ferrite, 7.21–7.27 growth kinetics, 7.22 interface characteristics and orientation relations, 7.26 nucleation and growth, 7.22 relief effects, 7.27 Alloy carbides, 1.17, 1.18 Alpha double prime iron nitride (a≤–Fe16N2), 4.30 Alpha iron: crystal structure, 1.1, 1.2 octahedral hole, 1.1 packing factor, 1.1 tetrahedral hole, 1.1 (See also Ferrite) Alternative layer deposition (see Atomic layer epitaxy) Aluminide formation, 18.59, 18.65, 18.67

Aluminum nitride (AlN) embrittlement, 14.81–14.83 panel cracking in ingots, 14.82 reduced hot ductility, 14.83 Amorphitization, 18.10 AMS specification, 1.52 Andersen formula, 18.12 Anelastic stress, 4.13 Anilox rolls, 19.24, 19.25, 19.47 coating specifications, 19.48 Anisotropic structure, 19.33, 19.36, 19.46 Annealing of bar, rod, and wire, 12.18, 12.19 batch, 12.18, 12.19 continuous, 12.19 Annealing of cold worked materials, 5.1–5.55 grain growth, 5.1, 5.39–5.50 laws of recrystallization, 5.32, 5.33 polygonization, 5.6, 5.7 recovery, 5.3–5.6 recrystallization, 5.1, 5.12–5.26 recrystallization temperature, 5.29–5.32 recrystallization texture, 5.33–5.39 secondary recrystallization, 5.50–5.55 semiconductors, 16.36 stored energy release, 5.1–5.5 Annealing of compact graphite iron, 12.34 Annealing of ductile iron, 12.32–12.34 effect of alloying elements, 12.32 full, 12.33 modified two-stage, 12.33 single stage (subcritical), 12.32 stress relieving, 12.34 two stage, 12.32 microstructure, 12.32, 12.33 Annealing of forgings, 12.19 Annealing of gray iron, 12.27–12.30 high temperature (or graphitizing), 12.27 low temperature (or ferritizing), 12.28 microstructure, 12.28 rate of ferritization, 12.28 medium temperature (or full), 12.27, 12.28 I.1

I.2

INDEX

Annealing of gray iron (Cont.): procedure, 12.27, 12.28 stress relieving, 12.28, 12.30, 12.31 Annealing of malleable iron, 12.34–12.36 ferritic malleable iron, 12.35 first stage graphitization, 12.35, 12.36 second stage graphitization, 12.36 martensitic malleable iron, 12.36 microstructures, 12.36, 12.37 nucleation and growth of temper carbon nodules, 12.36 pearlitic malleable iron, 12.36 Annealing of plate, 12.19 Annealing of steel sheet and strip, 12.13 batch (or box), 12.13–12.15, 12.18 continuous, 12.13, 12.14, 12.16 cooling methods, 12.16, 12.17 production line, 12.16, 12.17 zinc coatings, 12.16 Annealing of steels, 12.3–12.20 definition, 12.3 general, 12.3 purpose of, 12.3, 12.4 (See also specific types) Annealing of thermal sprayed coatings, 19.25, 19.26 ceramic coatings, 19.25, 19.26 metal and alloy coatings, 19.25, 19.26 Annealing of tubular products, 12.19 Annealing twins, formation, 5.47–5.50 mechanism for, 5.49 Anomalous metals, 2.54 Antiphase: boundary, 6.65–6.67 domains, 14.7 Anti (and selective) reflective coatings, 18.15, 18.35, 18.87, 18.94 AOD process, 10.88 API, 1.52 Arc spray process: advantages and disadvantages, 19.10, 19.11 applications, 19.11 procedure, 19.10 (See also Electric (or wire-) arc spray coating) Architectural stainless steel, 10.49, 10.52, 10.53, 10.55 Arrhenius equation, 2.40, 2.53, 5.47 Arrhenius parameter, 2.41 Arrhenius plots, 2.41, 2.43, 2.48, 2.49, 2.54, 2.56 Ashby equation, 4.25

Ashby-Orowan strengthening, 9.35, 9.36 Aspect ratio, 6.5, 6.6, 7.25, 7.27 ASTM grain size number, 10.9–10.14 ASTM specification, 1.51, 4.9, 10.75 Atmospheric plasma spray (APS) coatings: advantages and drawbacks, 19.13 applications, 19.13 characteristic features, 19.12 principle, 19.12, 19.14 Atom probe field ion microscopy, 2.37, 14.2–14.4, 14.7 Atomic diameter, ideal ratios, 10.80 Atomic layer epitaxy (ALE), 18.103, 18.105–18.107 advantages and limitations, 18.105, 18.106 application, 18.105 equipment: rotating substrate reactor, 18.107 traveling wave reactor, 18.106, 18.107 principles, 18.105 Atomization treatment, 15.55 Ausforming of steel, 15.29 application, 15.30 characteristic features of, 15.29, 15.30 disadvantages, 15.30 Austempering of ductile iron, 13.86–13.94 advantages and applications, 13.92–13.94 classification, 13.86, 13.90 comparison of mechanical properties, 13.90 effect of incomplete transformation, 13.90 process control, 13.90 reaction, 13.88, 13.89 saturated salt baths, 13.86, 13.88 section thickness, 13.86 Austempering of steel, 13.77–13.86, 13.138 advantages and limitations, 13.79–13.82 applications, 13.83–13.85 comparison of process and products, 13.82 comparison of properties, 13.78, 13.79, 13.82, 13.83 hardness and section thickness of products, 13.78, 13.79 mechanical properties, 13.78, 13.79 procedure, 13.77, 13.78 supersaturation method, 13.82 Austenite: atomic packing factor, 1.2 crystal structure, 1.2 dendrites, 12.37 formation, 10.1–10.7 formers (or stabilizers), 1.14, 1.15

INDEX

growth, 12.2 mechanical properties, 1.2 nucleation and growth process, 12.2, 12.3 solidification, 10.25 stabilization, 8.17 unit cell, 1.2 Austenite-ferrite band, 10.95 from annealed fully pearlitic structure, 10.2 grain growth rate, 10.2 isothermal austenitizing diagram, 10.3 nucleation rate, 10.3 nucleation sites, 10.3, 10.5 from bainitic structure, 10.6 mechanism, 10.6 morphology, 10.6 from ferrite-pearlite structure, 10.6 mechanism, 10.6 in ferrite-spheroidized cementite structure, 10.5 nucleation sites, 10.6 from martensitic structure, 10.6 mechanism, 10.6 morphology, 10.6 nucleation sites, 10.6 from normalized fully pearlitic structure, 10.3 time-temperature relationship, 10.3, 10.5 Austenite grain boundaries, development of, 10.7–10.9 delineation, 10.7, 10.8 McQuaid-Ehn test, 10.7 oxidation etching, 10.7, 10.8 picric acid solutions, 10.7, 10.8 thermal etching, 10.7, 10.8 Austenite grain boundary per unit volume, SV, 15.22, 15.23 Austenite grain size, 10.7–10.14, 15.10 effect on hardenability, 10.7 effect on mechanical properties, 10.7 measurement techniques, 10.9, 10.12–10.14 automated image analyzer, 10.13 lineal intercept (Heyn) method, 10.12 planimetric (Jeffries) method, 10.12, 10.13 Shepherd fracture grain size, 10.14 Snyder-Graff intercept method, 10.13, 10.14 standard chart method, 10.9 starting, 15.10 Austenite/pearlite reaction front, 7.5, 7.6

I.3

Austenitic stainless steels, 10.19–10.86 applications, 10.49, 10.52–10.54 basic 18-8 grades, 10.49 classification, 10.25, 10.27 compositions, 10.28–10.32 constituent phase diagrams: Fe-Cr, 10.21 Fe-Cr-Ni, pseudobinary, 10.26 effect and measurement of d ferrite, 10.63–10.66 Fe-18Cr-8Ni-C, 10.24 Fe-18Cr-4Ni-C, 10.24 Fe-Ni, 10.23 Ni-Cr, 10.23 definition, 10.20, 10.21 effect of cold working on tensile properties, 10.49, 10.50 high corrosion resistant, 10.54–10.56 martensite formation, 10.81, 10.82 mechanical properties, 10.33–10.46 metastable, 10.84 Mn-substituted, 10.27, 10.47–10.49 solubility of N in, 10.47, 10.48 stress-relieving treatments, 10.27, 10.47 structure-property relationships, 10.82–10.86 strengths, 10.82–10.86 toughness, 10.86 Austenitic stainless steels, precipitation: borides, 10.74, 10.75 chi (c) phase, 10.80, 10.81 d-ferrite, 10.63–10.65 chromium equivalent, 10.63, 10.64 ferrite number, 10.63–10.65 harmful effects of, 10.65 hot tearing, 10.65 metallography, 10.63 microfissuring, 10.65 nickel equivalent, 10.63, 10.64 Schaeffler diagram, 10.63, 10.64 eta (h) and beta (b) precipitates, 10.77, 10.78 G phase, 10.81 gamma prime (g ¢), 10.76, 10.77 coarsening properties, 10.76 crystal structure, 10.76 orientation relationship, 10.76 Laves (h) phases, 10.79, 10.80, 14.108 atomic diameter ratios, 10.80 conditions necessary for formation of, 10.79, 10.80 crystal structure, 10.79 orientation relationship, 10.80

I.4

INDEX

Austenitic stainless steels, precipitation (Cont.): MC carbide, 10.73 nucleation sites, 10.73 M6C carbide, 10.73, 10.74 M7C3 carbide, 10.74 M23C6 carbide, 10.67–10.73 crystal structure, 10.67 grain boundary precipitation, 10.70–10.72 intragranular precipitation, 10.72, 10.73 kinetics, 10.68–10.70 morphology, 10.69–10.72 nucleation sites, 10.68, 10.73 orientation relationship, 10.71, 10.72 time-temperature-precipitation curve, 10.68, 10.69 mu (m ) phase, 10.81 nitride precipitation, 10.75, 10.76 s-phase, 10.78, 10.79 correction, 10.78 crystal structure, 10.78 harmful effects, 10.78, 10.79 nucleation sites, 10.78 occurrence, 10.78 orientation relationship, 10.78 Austenitic stainless steels, sensitization, 10.59–10.63 control of, 10.62, 10.63 polythionic acid cracking, 10.60 SCC, 10.59 theory: Cr-depletion, 10.62 occurrence, 10.59 segregation, 10.62 thermodynamic, 10.62 time-temperature-sensitization curves, 10.60, 10.61 weld decay, 10.61 Austenitization, rapid, 10.7 Autocatalytic nucleation, 8.21, 8.44 Autotempering, 8.79, 9.42, 14.38 Avrami equation, 5.18, 5.51, 5.52, 7.11, 7.12, 10.2, 15.17, 15.23, 15.24

Bain distortion (or mechanism or strain), 8.27, 8.28 usefulness, 8.28 weakness, 8.28 Bainite, 9.1–9.24 Bf, 9.18, 9.19, 9.33 B50, 9.33

Bs, 9.8, 9.19, 9.33 C curve, 9.1, 9.16, 9.18, 9.19 crystallography, 9.3 definition, 9.1, 9.2 distribution (or precipitation) of, 9.1, 9.8 effects of alloying elements, 9.23, 9.24 Reynold-Aaronson model, 9.23 solute drag like effect, 9.21, 9.23 volume diffusion of carbon, 9.22 as function of isothermal temperature, 9.18, 9.19 growth by shear, 9.14 habit planes, 9.7, 9.10–9.12, 9.17, 9.24 inhomogeneous shear, 9.14, 9.17 interfacial structure, 9.17 interphase precipitation, 9.10 kinetic definition, 9.18–9.21 empirical equation for Bs temperature, 9.19 gR Æ bainite, 9.19, 9.20 incomplete transformation (stasis phenomenon), 9.1, 9.20, 9.21 TTT diagram, 9.20, 9.21 lath, 9.3–9.7, 9.13, 9.14, 9.17, 9.18 lattice invariant shear, 9.11 LBm, 9.12 LBs, 9.7, 9.8, 9.24 ledge growth mechanism, 9.12, 9.17, 9.22 massive (or granular), 9.11 mechanisms, 9.21 lower bainite: Aaronson model, 9.23 Bhadeshia model, 9.23 diffusionless shear (or displacive) model, 9.23 Ohmori model, 9.23 upper bainite: atom probe studies, 9.22 displacive mechanism, 9.22 in ferrous alloys, 9.22 ledge growth mechanism, 9.22 in nonferrous alloys, 9.22 Ohmori classification, 9.22 microstructural definition, 9.2 blocky bainite, 9.12 ferrous bainite, 9.2–9.12 grain boundary allotriomorphic bainite, 9.12 granular bainite, 9.11, 9.12 inverse bainite, 9.12 lower bainite, 9.7–9.11 LBm, 9.12 nodular bainite, 9.11

INDEX

nonferrous bainite, 9.12–9.14 in Cu–based alloys, 9.14 in Ti-X alloys, 9.13, 9.14 upper bainite, 9.3–9.7 morphologies, 9.2–9.6, 9.9, 9.11–9.14 plate, 9.7–9.12, 9.14, 9.16, 9.17 similarity with tempered martensite, 9.2 spine plate, 9.12, 9.23 strengthening mechanisms, 9.33–9.36 surface relief definition, 9.14–9.18 multiple surface relief, 9.14, 9.16 nonferrous bainite, 9.16 single surface relief, 9.14, 9.15 tent or V-shaped tilt, 9.16 terminology, 9.1, 9.11 transition temperature between bainites, 9.5, 9.6 TTT diagram near Ms, 9.20 Widmanstätten ferrite, 9.2, 9.3, 9.6, 9.16 Bainite transformation in ductile iron, 13.88–13.90 ausferrite, 13.90 lower bainite (or ausferrite), 13.89 mechanical properties, 13.89, 13.90 stabilization of gR, 13.88 upper bainite (or ausferrite), 13.88 Bainitic forging steels, 9.38, 9.39 advantages, 9.38, 9.39 heat treatment cycles, 9.39 Bainitic high strength, carbide-free steels, 9.39, 9.40 Bainitic rail steels, 9.40–9.43 composition, 9.40, 9.41 wheel spin burning, 9.43 Bainitic steels, 9.29–9.45 currently used, 9.29, 9.30 high-carbon: advantages, 9.38, 9.43, 9.44 applications, 9.44 composition, 9.30, 9.32, 9.43 disadvantages, 9.42, 9.44 strength, 9.43 low-carbon and ultra-low carbon: advantages, 9.37, 9.38 alloying element selection, 9.30, 9.31 applications, 9.37, 9.38 Ashby-Orowan strengthening, 9.35, 9.36 equation for ITT determination, 9.36 strength, 9.31, 9.33 strengthening mechanism, 9.33–9.36 TMT, 9.37 toughness (impact and fracture), 9.36, 9.37

I.5

wear resistance, 9.37 weldability of, 9.44, 9.55 CE value, 9.44, 9.45 cold (or HAZ) cracking, 9.44, 9.45 Bake hardening steels, 1.33, 1.34 Ballistic mechanism, 18.11, 18.12 Band gap, 2.103 of heterostructures, 18.81, 18.98 Bar-code scanning, 18.99 Barium activated liquid carburizing (see Salt bath carburizing) Barium carbonate (see Pack carburizing) Barium chloride (see Salt bath carburizing) Barkhausen effect (or noise), 17.28 Barrier layer for semiconductor metallization, 18.35 Baths (see individual baths, also cooling and quenching entries) Bay, 15.29 Bearing steels, 1.41, 1.43 Bend gliding, 5.7 Bending: fatigue life, 13.71 strength, 17.10 Beta-titanium alloy structures, 15.69 Bipolar transistors, 18.56 Blistering, hydrogen-induced, 14.101 Blue brittleness, 4.33 Blueing annealing, 12.19, 12.20 Boltzmann–Matano solution, 2.58–2.60, 2.68, 2.99 Bond coat, 19.10, 19.25, 19.33, 19.43, 19.46 Bond strength, 19.2, 19.3, 19.9, 19.10, 19.12, 19.15, 19.18, 19.20, 19.23, 19.25, 19.29, 19.30, 19.33, 19.35, 19.38, 19.42, 19.46 Boride: coatings, 18.77 formation, 18.59, 18.65, 18.66 Boron, 1.23 in bainitic steels, 9.30–9.32 factor, 13.125, 13.126 in H-steels, 13.134, 13.135 hardenability, 1.23, 13.123 optimum content, 1.23, 13.123, 13.138 steels, 1.33, 13.125 Boronizing (or boriding), 16.114–16.138 advantages, 16.114–16.119 applications, 16.136–16.138 boriding reactions, 16.123 disadvantages, 16.119–16.121 effect of alloying elements, 16.123, 16.125, 16.126 of ferrous materials, 16.121, 16.123

I.6

INDEX

Boronizing (or boriding) (Cont.): characteristics of FeB and Fe2B layers, 16.121, 16.122 typical properties, 16.121, 16.122 multicomponent, 16.134, 16.135 of nonferrous materials, 16.125 techniques, 16.127 fluidized bed, 16.133, 16.134 gas, 16.132, 16.133 liquid, 16.131 pack, 16.127 Borudif process, 16.130 case depth, 16.128, 16.130 furnaces, 16.128 paste, 16.130, 16.131 plasma, 16.132 pulsed plasma, 16.133 Bosze-Trivedi treatment, 7.32 Boundary: high angle, 5.8 low angle, 5.8 pinning, 15.8 tilt, 5.8 Bradley-Aaronson treatment, 7.25, 7.26 Branching mechanism, 7.2, 7.4 time sequential, 7.4, 7.5 Bridge ropes, 7.60 Bright annealing, 12.20 Brine or caustic solutions, 13.10 advantages, 13.10 cooling rate, 13.10 disadvantages, 13.10, 13.111 Brittle behavior (CVN, DBTT, DT, DWT, NDT), 4.9 Brittle fracture, 4.9, 14.88 Brittle materials, 4.7 Buffer layer, 18.44, 18.45 Burning of low alloy steels, 17.2 detection and effects of, 17.5 occurrence, 17.2 prevention, 17.8, 17.9

Cahn-Hilliard, definition, 6.19 theory, 6.19–6.21 Cahn-Hillig-Sears model, 7.30, 7.34, 7.35 Calendar rolls, 19.48 Carbide: coatings, 16.138–16.144, 18.24, 19.35 formation, 18.59, 18.65, 18.67 formers, 7.8 network, 16.112 spheroidal, 12.7–12.11, 16.111

Carbides, alloy, 1.17, 1.18 chi (c) carbide, 1.17 (See also stainless steels, precipitation hardening) e-carbide, 1.17 (See also e-carbide) h-carbide (see h-carbide) Carbon, atomic radii, 1.12 Carbon-carbon composites, 18.108, 18.109 Carbon diffusivity, 7.25 Carbon equivalent of steel, 9.44, 9.45, 13.95 Carbon flux, 7.22 Carbon potential, 16.59, 16.60, 16.63, 16.66, 16.67, 16.71 dew point method, 16.63 hot wire method, 16.63 infrared method, 16.64, 16.65 oxygen probe method, 16.64, 16.65 Carbon profile, computer controlled, 16.65 advantages, 16.65 Carbonitriding, 16.75–16.80 fluidized bed, 16.80, 16.81 gaseous, 16.76–16.80 advantages and disadvantages, 16.77 applications, 16.80 case composition, 16.78 case depth, 16.78, 16.79 control of gR, 16.79 furnaces, 16.79 of P/M parts, 16.80 quenching media, 16.79 void formation, 16.79 temperature selection, 16.79 tempering, 16.79, 16.80 liquid, 16.76 plasma, 16.80 Carburizing, 16.42 case, 16.42 case depth measurement, 16.45 diffusion period, 16.43 direct quenching, 16.44 double quenching, 16.43, 16.44 driving force for diffusion, 16.43 effect of alloying elements in shifting Acm line, 16.44 high temperature, 16.59, 16.67 single stage, 16.62 single quenching, 16.44 steel grades, 16.42, 16.43 temperature, 16.43 Cascade mixing, 18.12 Cast iron, 1.6

INDEX

applications, 1.6 carbon equivalent, 1.55, 1.57 classifications, 1.6, 1.7, 1.53–1.55 compact graphite, 1.64, 3.141 eutectic growth morphology, 3.141 (See also Compacted graphite iron) ductile, 1.60–1.64 eutectic growth morphology, 3.138– 3.140 eutectic cell, 1.57 Fe-Fe3C-Si system, 1.55, 1.56 graphitization, 1.60 gray (see Gray iron) high alloy, 1.65–1.67 malleable, 1.64, 1.65 (See also Malleable iron) Mn in, 1.6 other elements, effects of, 1.6 P in, 1.6 S in, 1.6 Si in, 1.6 solubility of carbon in, 1.6 white, 1.57, 1.66, 3.136, 3.137 Catalysis, 18.59, 18.68, 18.69 Cavitation, internal, 15.87, 15.88, 15.90 Cell size, 4.22, 14.59 Cell structure, 4.21 Cell walls, 4.22, 5.6 Cells (or subgrains), 4.22 Cellular growth, 3.41, 3.42, 6.49 substructure, 4.23 Cementite, 3.6 morphology, 7.41, 7.42 network, 7.2 Ceramic coatings, 18.27, 18.75, 18.78, 19.25, 19.30, 19.42, 19.43, 19.46, 19.47 Ceramic cutting tools, 18.75, 18.77 Ceramic rolls, 19.49, 19.50 Ceramics, 2.94, 15.94, 18.9, 18.10, 18.92 Cermet coatings, 19.20, 19.21, 19.38, 19.40, 19.45, 19.47 Charpy shelf energy (CSE), 4.9, 7.43, 7.46, 15.38, 15.39, 15.53 Chemical beam epitaxy (CBE): advantages, 18.100 application, 18.103 schematic CBE system, 18.103, 18.104 versus other processes, 18.100, 18.102 Chemical potential gradient, 5.40, 5.41, 5.46 Chemical transport, 18.59, 18.67, 18.68, 18.70 Chemical vapor deposition (CVD):

I.7 advantages and disadvantages, 18.70, 18.71 applications, 18.71–18.88 antireflective (AR) coatings, 18.67 boride coatings, 18.77 CVD–coated fibers, 18.79 CVD-coated powders, 18.78, 18.79 cutting tool industries, 18.72–18.77 electrical characteristics of materials, 18.79, 18.80 electronics, 18.79–18.81 electronic devices and their functions, 18.80 semiconductor diamond, 18.81 ferroelectrics, 18.86 fibers, 18.79 hot (infrared)– and cold (visible)mirror CVD coatings, 18.87 indium-tin oxide (ITO) coatings, 18.88 monolithic metallic structures, 18.77 optical, 18.86, 18.87 optoelectronic applications, 18.82–18.86 critical properties, 18.82 device materials, 18.86 group III–V and III-nitride compound semiconductors, 18.82 photodiodes, properties, 18.82, 18.83 transparent conductive oxide (TCO) coatings, 18.87, 18.88 ultrafine powders, 18.78 wear, erosion, and corrosion resistant properties, 18.71, 18.72, 18.74 whiskers, 18.79 chemical reactions, 18.59 classification of chemical reactions: boride, carbide, aluminide formation, 18.65–18.67 catalysis, 18.68 chemical transport, 18.67 combined reactions, 18.69, 18.70 disproportionation, 18.68 hydrolysis, 18.63 nitridation, 18.63, 18.64 oxidation, 18.61, 18.62 photolysis, 18.69 reduction, 18.61 synthesis, 18.68, 18.69 thermal decomposition/pyrolysis, 18.59–18.61

I.8

INDEX

Chemical vapor deposition (CVD) (Cont.): classification of CVD processes (see individual process) criteria for CVD sources, 18.58, 18.59 principles, 18.57 reactor configurations, 18.58 typical CVD system, 18.57 Chemical vapor infiltration (CVI), 18.107–18.110 advantages and disadvantages, 18.110 application, 18.108, 18.110 types, 18.107, 18.108 Cleavage crack propagation, 9.36 Closed-field unbalanced magnetron sputtering, 18.41, 18.42 Coarse grained structure, 4.15, 5.52 effect of coercive force, 5.52 soft magnetic materials, sheets, 5.52 texture, 5.52, 5.53 Coarse structure, 16.27 Coarsening theory, 10.76 Coating characteristics, 19.33–19.36 anisotropy, 19.36, 19.46 classification, 19.33 mechanical properties, 19.36–19.39 microstructures, 19.33, 19.36 metastable phases, 19.35 oxides, 19.35 porosity, 19.33–19.35 unmelted particles, 19.35 Cobalt-based superalloys, 6.53, 6.63, 10.68 Coble creep mechanism, 2.80, 2.82, 4.23, 4.25 Coercivity, 6.28 Coherency, loss of, 6.6, 6.15, 6.45, 9.22 Coherency dislocations, 8.28 Coherent, embryo, 6.3, 6.6 facets, 7.25 interface, 6.6, 8.46 nucleation, 6.14, 6.15, 9.3 critical free energy near dislocations, 6.15 on dislocations, 6.14, 6.15 at sharp interfaces, 6.4, 6.6 precipitate, 6.4, 6.6, 6.67 aspect ratio, 6.6 barrier to nucleation, 6.6, 6.7 misfit strain, 6.5 orientation–habit plane relationship, 6.6 partially, 9.14 shape dependent, 6.5 strain energy, 6.4–6.6 Cohesive strength (see Bond strength) Cold die quenching, 13.64, 13.65

Cold gas dynamic spray (CGDS) process procedure, 19.22, 19.23 Cold rolled motor lamination (CRML) steels, 12.13 Cold rolling texture, 4.46 Cold-wall reactor, 18.88 Cold worked materials: effect of variables, 4.28 release of stored energy, 5.1–5.3 stored energy of, 4.27, 4.28 Cold working, 4.27 Combined reactions, 18.59, 18.69, 18.70 Combustion flame spray process (see Flame spray and fuse process) Compacted graphite iron, 1.64 applications, 12.61, 12.63 damping capacity, 12.61 fatigue properties, 12.60 impact toughness, 12.59, 12.60, 12.62 machinability, 12.60, 12.61 mechanical and physical properties, 12.58–12.60 stress-strain curves, 12.58, 12.59 thermal fatigue properties, 12.60 Complementary wrap forming, 15.92 Composite coating, 18.111, 18.139 Composition modulation, 6.69 Composition plane, 4.35 Compound coatings, 18.18 Concentrated profile, implanted ions, 18.3 Congruent point, 7.38 Contact angle, 3.12, 3.13, 6.9–6.11 (See also Dihedral angle; Wetting angle) Continuous casting, 3.98 of nonferrous alloys, 3.101 direct chill casting, 3.103–3.105 electromagnetic casting, 3.104, 3.105 hot top casting, 3.101, 3.102 low head composite (LHC) casting, 3.104 Ohno continuous casting, 3.101, 3.102 of steel, 3.98–3.100 continuous cast structure, 3.100 Continuous coarsening kinetics, 7.15 Continuous cooling transformation diagrams, 11.10 correlation with IT diagrams, 11.12 derived, 11.11–11.13 experimentally determined, 11.13–11.17 for C-Si-Mn-Cr-Mo-Ni steels, 11.14, 11.16 dilatometric method, 11.13 elongation-temperature curve, 11.14

INDEX

for SAE 4140 and 4150 steels, 11.15 modified, 11.17 for C-Si-Mn-Cr-Mo-Ni steel, 11.19, 11.20 effect of composition and bar diameter, 11.19 limitations, 11.19 for 1035–1040 steel, 11.18 for rail steel, 7.49, 7.50 Continuous transformation, 6.1 Controlled atmospheric plasma spray (CAPS) process, 19.13 Conventional molecular beam epitaxy (MBE), 18.44–18.48 advantages and disadvantages, 18.47, 18.48 applications, 18.48 schematic growth kinetics, 18.44, 18.45 typical system for Si and III–V semiconductor growth, 18.44, 18.46 Cooling curves, 13.6, 13.8 alternative methods of determination, 13.18, 13.19 hot wire test, 13.16 magnetic quenchometer test, 13.16 for oil, 13.10, 13.13, 13.15 of quenchant, 13.16 silver ball (or cylinder) method, 13.6 steel probe, 13.6 for water and soluble oils, 13.10, 13.11 Cooling rate (or power), 13.6, 15.24 Castrol index, 13.21 factors determining, 13.21 IVF hardening power, 13.19 for quenching media, 13.6–13.9, 13.12–13.16 role of additives, 13.21 Copper-base alloys, 6.51, 15.77, 15.78 Corona rolls, 19.24, 19.47 Correlation factors, 2.33, 2.43 Cosworth process, 3.159, 3.161 Cottrell atmosphere, 14.85, 14.88 Coupling gaps, 16.30 Crack length, effective, 14.90 Cracking, 13.10, 13.12, 13.16, 13.27 Cracking cell, plasma, 18.54 Crazed pattern, 17.50 Critical concentration, 7.36 Critical nucleus, 6.4, 6.7, 6.8, 6.19, 7.40 Critical resolved shear stress, 4.13, 4.42, 6.57, 6.59, 6.62, 6.65–6.69 Critical stress, upper, 14.90, 14.91 lower, 14.90, 14.91

Critical temperatures, 1.7–1.9 coherent, 6.19, 6.20 range, 1.7 Critical thickness, 6.6, 6.7 Cross slip, 4.13, 4.21, 4.23, 4.27, 15.12 Crowdion mechanism, 2.29, 2.30 Cryogenerator, 10.18 Cryopanel, 18.44, 18.49 Curie point, 1.2, 13.16, 16.12 Cutting tools, CVD-coated, 18.72–18.77 Cyaniding, 16.76 bath composition, 16.76 bath properties, 16.76 case depth, 16.76 procedure, 16.76

Damping capacity, 8.48, 8.73, 12.44, 12.58, 12.61, 12.64 Darken’s analysis, 2.62–2.64, 2.66 Darken’s equation, 2.64, 2.66, 2.68 DBTT, 4.9, 4.33, 7.43, 7.46, 8.81, 9.36, 9.37, 10.86, 10.98, 14.62, 14.63, 14.72, 14.76, 15.31, 15.38, 15.39, 15.53 Decarburization, 12.20, 12.21, 13.36, 13.95, 16.16 control, 12.20 determination of, 12.21 types, 12.21 Decarburization annealing, 12.12, 12.13 Decarburized steels, 17.31 Decorative coatings, 18.20, 18.23, 18.30, 18.31, 18.36, 19.48 Defect cascades, 2.104 Defects: extrinsic, 2.95 intrinsic, 2.95 Deformation: homogeneous, 4.6 inhomogeneous, 4.16, 4.17 Deformation bands, 4.31 Deformation texture, 4.44–4.49 fiber texture, 4.47, 4.48 plastic anisotropy, normal and planar, 4.44, 4.45 rolling, 4.47–4.49 Deformation twinning, 4.27, 4.33–4.44 in bcc materials, 4.40–4.42 characteristics of, 4.35 crystallography, 4.37–4.39 in fcc materials, 4.42, 4.43 in hcp materials, 4.43, 4.44 nucleation of, 4.35

I.9

I.10

INDEX

Deformation twinning (Cont.): plane, 4.35 shuffles, 4.35 systems, 4.35, 4.36 Dehydrogenation annealing, 12.12 Delayed cracking, 14.100, 14.101 Delta ferrite, 1.2 crystallographic structure, 1.2 mechanical properties, 1.2 DK, 15.40 DKth, 15.50 Dendrite growth, 3.41–3.44 anisotropy, 3.49, 3.50 cell-dendrite transition, 3.53 models: in alloy melt, 3.44–3.49 in pure supercooled melt, 3.49 primary dendrite spacing, 3.50–3.52 secondary dendrite arm spacing, 3.52 twinned dendrites, 3.53 Dendrite arm spacing, 3.89–3.93 Detonation gun (D Gun) spray process: applications, 19.21, 19.22 characteristic features, 19.20, 19.21 process, 19.20, 19.21 Detwinning, 8.46 Diamond-like carbon (DLC) coatings, 18.15, 18.16, 18.29–18.31, 18.41, 18.72 Dielectric coatings, 18.35 Differential dilatometry, 2.37, 2.38 Diffraction gratings, 18.36 Diffusion: applications, 2.1 Boltzmann-Matano solution, 2.58–2.60 Coble creep, 2.80, 2.82 in concentrated alloys, 2.58 correlation factors, 2.33, 2.43 Darken’s analysis, 2.62–2.64, 2.66 in dilute substitutional alloys, 2.56 discontinuous precipitation, 2.89, 2.90 Einstein relation, 2.32 Fick’s first law, 2.2–2.4, 2.9, 2.33 measurement: permeability, 2.10, 2.11 steady state method, 2.9, 2.10 Fick’s second law, 2.11–2.14, 2.86 non-steady state solutions, 2.14–2.23 decarburization, of steel, 2.18, 2.20 homogenization, 2.14–2.16 infinite (or diffusion) couple method, 2.20, 2.21 semi-infinite system, 2.16–2.18 thin layer (or instantaneous source) methods, 2.21–2.23

in ionic solids, 2.93–2.96 in semiconductors, 2.96–2.103 in thin films, 2.69–2.76, 2.92, 2.93 long range, 7.38 random walk theory, 2.32, 2.33 self-, in pure metals, 2.53–2.56 bcc metals, 2.56 fcc and hcp metals, 2.54–2.56 normal and anomalous, 2.53, 2.54 short range, 4.28, 7.36 ternary, 2.66–2.69 uphill (see Uphill diffusion) vacancy, 2.33–2.40 direct determination, 2.37, 2.38 formation, experimental measurement, 2.36, 2.37 indirect determination, 2.39 migration rates of defects and atoms, 2.36 in thermodynamic equilibrium, 2.33–2.36 vacancy wind correction, 2.66 volume, 7.25, 14.85 Diffusion along short circuits, 2.76, 2.77 dislocations, 2.77, 2.80 grain boundary, 2.80–2.86 induced grain boundary migration (DIGBM), 2.86–2.90, 5.40, 5.41 surface, 2.90–2.92 TLK model, 2.90, 2.91 Diffusion barrier, 18.81 Diffusion bonding, 15.104, 15.105, 18.20 with interface aids, 15.105 Diffusion coefficient, 2.50–2.84 Arrhenius parameter, 2.41 chemical (or interdiffusion), 2.51, 2.53, 2.59, 2.60, 2.63, 2.64, 2.66–2.68, 2.91 effect of pressure, 2.40, 2.43, 2.48 effect of temperature, 2.40–2.42 effective, 2.83 impurity or solute, 2.52 indirect methods of measurement, 2.23–2.26 Gorsky effect, 2.24–2.26 magnetic relaxation, 2.26 Mössbauer effect, 2.23, 2.24 nuclear magnetic resonance, 2.23 nuclear reaction analysis, 2.26 quasi-elastic neutron scattering, 2.23 relaxation method, 2.24–2.26 Rutherford back scattering (RBS) analysis, 2.26 Secondary ion mass spectroscopy (SIMS), 2.26

INDEX

Snoek relaxation, 2.24, 2.25 Zener relaxation, 2.24–2.26 intrinsic, 2.3, 2.60, 2.65, 2.74 isotropic mass effect, 2.50 Onsager’s phenomenological, 2.66 self-, 2.51, 2.52 solute, 2.54 solvent 2.57 steady state method of measuring, 2.9 permeability, 2.10, 2.11 tracer self-, 2.51, 2.60, 2.64, 2.66 Diffusion in concentrated alloys Boltzmann–Matano solution, 2.58–2.60 Darken’s analysis, 2.62–2.64 Kirkendall effect, 2.60–2.62 experiment, 2.61, 2.62 vacancy wind effect, 2.64–2.66 Diffusion in dilute substitutional alloys, anomalous dilute systems, 2.57, 2.58 solute diffusivity, 2.57 solvent diffusivity, 2.57 Diffusion in ionic solids, 2.93–2.96 defects, 2.94, 2.95 diffusion theory, 2.95, 2.96 Diffusion in semiconductors, 2.96–2.103 GaAs and AlAs-GaAs materials, 2.103 intrinsic point defects, 2.97, 2.98 silicon, 2.98–2.103 mechanisms, 2.99–2.103 phenomenological concepts, 2.98, 2.99 Diffusion in silicon: diffusion mechanisms, 2.99 dopant diffusion, 2.101 silicon self-, 2.100, 2.101 dopant diffusion: Fermi level effect, 2.101 interstitial related, 2.101 misfit dislocations, 2.102 interstitial-substitutional diffusion, 2.102, 2.103 phenomenological concepts, 2.98, 2.99 Diffusion-induced grain boundary migration, 2.86–2.90 discontinuous or cellular precipitation, 2.89, 2.90 liquid film migration, 2.90 Diffusion mechanisms, 2.26–2.32 direct exchange and ring, 2.26, 2.27 Frenkel pairs, 2.32 interstitial, 2.27 interstialcy, 2.27, 2.28 involving extended defects, 2.32 mixed, 2.30, 2.32

I.11

split interstitials, 2.28, 2.29 vacancy, 2.29, 2.30 Dihedral angle, 5.40 Dimensional changes, 17.34 machining related factors, 17.36 in steel parts during heat treatment, 17.36, 17.37–17.40 stress related factors, 17.34, 17.36, 17.37 due to thermal contraction and transformation expansion, 17.18, 17.19, 17.36, 17.37 Dimensional control, 16.28 Dimensional restoration, 19.13, 19.42 Dimensional stability, 17.50, 17.51 Direct quenching, 13.68 Directional solidification processing, 3.163, 3.164 Discontinuous (or cellular) precipitation, 2.80, 2.89, 2.90 Discontinuous stringers, carbide, 9.4 Dislocation: absorption, 6.15 annihilation, 6.58, 15.13 bending, 8.78 as catalyst for nucleation, 6.12–6.15 climb, 5.7, 5.8, 5.13, 6.15, 6.49, 15.87 cores, 6.13, 14.85 curved, 10.72 density, 4.15, 4.16, 4.21, 4.23, 4.25, 4.27, 4.28, 5.10, 8.33, 8.79–8.81, 9.33, 9.35, 14.60, 15.14 and grain size, 4.23 diffusion, 2.77–2.80 dissociation, 5.6, 6.49 dynamics model, 4.14 edge, 4.13, 6.61, 6.69, 10.72 geometrically necessary, 4.25 grain boundary, 4.28 helical, 6.49 immobile (or tangled), 4.16, 4.21, 4.22, 8.33, 8.34 interaction: force, 6.61 with dislocation, 4.21, 6.66 with grain size, 4.23–4.26 loop, 5.13, 6.58 mobile, 4.13, 4.15, 4.16, 4.26 movement, 6.67 multiplication, 4.13, 4.15 nonuniform, 4.21 in pairs, 6.67 partial, 4.40, 6.68 pinning, 4.17, 4.23, 4.26, 4.28, 6.59 prismatic punching mechanism, 10.72

I.12

INDEX

Dislocation (Cont.): screw, 4.13, 4.15, 6.69 Shockley partial, 10.73 split, 6.68 statically stored, 4.25 strengthening, 2.111, 2.112, 4.21, 8.79–8.81, 9.35, 14.60, 15.35 structure, 4.22 velocity, 4.16 Dispersed second phase particles, 15.31 Disproportionation, 18.59, 18.68 Dissociation mechanism, 2.30, 2.31, 2.54, 2.102, 2.103 Distortion in heat treatment, 17.34 in Al alloys and their control, 17.51, 17.52 allowable, 16.79, 17.34 in burnishing wheels, 17.42 in carburized steel roolers, 17.42 in case carburized steels, 17.42 change in bore of a finished gear, 17.41 in chisel, 17.41, 17.42 cold treatment or refrigeration, 17.50, 17.51 control of, 17.48, 17.49 crazing pattern, 17.50 definition, 17.34 dimensional changes: in carbon and high alloy steels, 17.37 during heat treatment, 17.36, 17.37 dimensional stability, 17.50 effect of grinding on, 17.49, 17.50 in half round files, 17.41 in head hardened rail, 17.43, 17.44 in helix and spur gears, 17.42, 17.43 and importance of design, 17.52, 17.53 Ishawaka diagram, 17.34, 17.35 in long pins and bars, 17.41 methods of preventing, 17.45 high pressure gas quenching, 17.46, 17.47 press quenching, 17.47 quenching fixture, 17.45, 17.46 restraining fixture, 17.45 rolling die quenching, 17.47, 17.48 straightening, 17.45 stress relieving, 17.48 in nitriding, 17.43 precautions, 17.44 and quenching, 17.30 in ring die, 17.40, 17.41 role of part design and ground rules, 17.52, 17.53

shape distortion, factors causing, 17.38, 17.40 size distortion, 17.38 in thin die, 17.41 in two pounder shot, 17.42 types and causes of, 17.34–17.37 Divacancy diffusion, equilibrium concentration, 2.35 Dopant diffusion, 2.97, 2.98, 2.101, 2.103 Double-cross slip, 4.15 Double hardening treatment, 13.36 Drop weight tear (DWT), 4.9 Dual phase steels, 15.45 application, 15.45 compositions of, 15.47 M–A constituents, 15.47 microstructure, 15.45, 15.46, 15.50 process routes, 15.45 continuous annealing line (CAL) method: advantages, 15.49 thermal cycles for, 15.49 hot rolling, 15.48 intercritical annealing method, 15.45 intercritical austenitization, 15.47, 15.48 properties, 15.50, 15.51 structure-property relationship, 15.51, 15.52 Ductile fracture, 4.5 Ductile iron, 1.60 alloying elements, 1.62, 1.64, 1.68, 1.69 applications, 1.62, 1.64, 12.57, 12.58 classification, 1.62, 1.64 damping capacity, 12.58 dynamic elastic modulus (DEM), 12.46, 12.50 fatigue limit, 12.55, 12.56 fracture toughness, 12.55 graphite spherulite, 1.61, 1.62 hardness, 12.47, 12.52, 12.53 impact properties, 1.64, 12.44, 12.53– 12.55 mechanical properties, 1.62, 1.64, 12.44–12.58, 12.60 modulus of elasticity, 12.46 morphology, 1.62, nodularity, 12.46, 12.50, 12.54 nodulizing, 1.60 strain rate and notch sensitivity, 12.55 stress-strain curves, 12.44, 12.47 tensile properties, 12.44, 12.45, 12.47, 12.52, 12.53

INDEX

TMT, 15.56 wear resistance, 12.55, 12.58 Ductility, 4.5–4.7 Duplex stainless steels, 10.88–10.95 advantages, 10.88 annealing temperature for, 10.89, 10.90 applications, 10.95, 10.96 cast alloys, 10.91, 10.92 compositions of different grades, 10.89, 10.90 embrittlement, 10.94 mechanical properties, 10.94, 10.95 microstructure, 10.88, 10.89 phase balance, 10.88 pitting resistance equivalent number (PREN), 10.89, 10.90 precipitate phases, 10.91, 10.93, 10.94 production, 10.88 sensitization, 10.88, 10.91 structure-property relationships, 10.96–10.98 superplastic properties, 10.89 TTT diagrams, 10.93, 10.94 wrought alloys, 10.89, 10.90 Dynamic recovery, 15.12, 15.58 characteristics, 15.12, 15.13 steady state flow stress, 15.14 Dynamic recrystallization, 15.14, 15.58, 15.65, 15.79 Avrami equation, 15.17 characteristics, 15.14–15.17 continuous flow curve, 15.15 critical grain size model, 15.19 critical strain, 15.15, 15.18 critical strain model, 15.17, 15.18 grain size, 15.16–15.18 necklace mechanism, 15.15, 15.16 occurrence, 15.14, 15.17 periodic flow curve, 15.15 progressive lattice rotation model, 15.19, 15.20 repeated, 15.17 Sah–Richardson–Sellars model, 15.18 Dynamic strain aging, 4.28, 4.31, 15.50 kinetics, 4.33 negative strain rate sensitivity, 4.33

Earing, 4.44 Einstein relation, 2.32 Elastic collisions, 18.3 Elastic constants, 4.3–4.5 anisotropic, 4.5

I.13

Elastic limit, 4.13 Elastic-plastic stress analysis, 14.87 Elastic strain energy, 6.3, 6.4 Electric- (or wire-) arc spray (or cold) coating: advantages and disadvantages, 19.10, 19.11 applications, 19.11 procedure, 19.10 processing parameters, 19.10 Electrical steels (see Steels, silicon) Electrically conductive: coatings, 18.20 diffusion barriers, 18.20 Electromagnetic powder deposition (EPD) process, 19.23, 19.24 Electromigration: damage, 2.76 electron wind effect, 2.69, 2.70 microelectronics device, 2.75, 2.76 in thin films, 2.72–2.75 Electron/atom ratio, 10.80 Electron beam evaporation: advantages, 18.25, 18.27 applications, 18.27, 18.28 guidelines for successful deposition, 18.25 Electron beam surface hardening, 16.37–16.42 advantages, 16.39–16.42 applications, 16.41, 16.42 characteristic features, 16.39 EB generation, 16.38, 16.39 equipment, 16.39, 16.40 principle, 16.39 programmable computer control system, 16.39 recommendations for using, 16.41 selection of, 16.37 Electron concentration, 10.80 Electron cyclotron resonance (ECR) plasma: advantages and disadvantages, 18.94, 18.95 principles, 18.94 schematic reactor, 18.95 Electron gun, 16.38, 16.39, 18.25 Electron irradiation, 5.5 damage recovery, 5.6 Electronics packaging, 18.10 Electroslag refining (ESR), 3.100, 17.4 Ellipsometry, 18.44, 18.102 Elongation, 4.5– 4.7 total, 4.7 uniform strain, 4.6, 4.7

I.14

INDEX

Embrittlement phenomena, 14.65 (See also specific types) Embryos, 6.3, 6.4, 6.7–6.11 End-quench hardenability test (see Jominy method) Endothermic (or endo) gas, 16.54, 16.56, 16.63, 16.70, 16.75, 16.80, 16.102, 16.103 Engineering strain, 4.1–4.3 Engineering stress, 4.1– 4.3, 4.5 Engineering stress-strain curve, 4.1–4.8, 4.10, 4.13, 4.15 Epitaxial deposition processes, definition, 18.43, 18.44 Epitaxial growth, 9.23, 18.97 Epitaxial single-crystal multilayers, 18.10 Epitaxy, 18.43 single layer, 18.105 e-carbide, 4.30, 9.10, 9.11, 14.2, 14.9, 14.11–14.14, 14.16 e-iron carbonitride, 16.102, 16.103 e-iron nitride, 16.81, 16.82, 16.98, 16.105, 16.110 Equilibrium: constant, 16.57 full local, or ortho (or PLE), 7.26 negligible partition under local, 7.26 para, 7.26 Error function, 2.17–2.19 Eshelby’s approach, 6.4 h-carbide, 14.2, 14.11, 14.13, 14.14, 14.16 h-phase, 10.78–10.80 (See also Laves phases) h-precipitate, 10.77 Eutectic carbide, decomposition, 12.27, 12.28, 12.32, 12.33, 12.35, 12.36 Eutectic cell, 1.57, 1.60 Eutectic composition (see Eutectic point) Eutectic growth morphology, cast iron, 3.136 compacted graphite iron, 3.141 coral graphite, 3.140, 3.141 gray iron, 3.137, 3.138 nodular iron, 3.138–3.140 white iron, 3.136, 3.137 Eutectic modification of Al-Si alloys, 3.125–3.135 chemical modification, type and amount, 3.126–3.131 effect of impurities, 3.131 effect of P, 3.132, 3.133 effect of Si content of the alloy, 3.131 effect on mechanical properties, 3.126, 3.129

fading of modifier, 3.132 impurity modified fiber, 3.131 incubation period, 3.133 mechanism, 3.133, 3.134 microporosity and shrinkage, 3.134, 3.135 modification efficiency, 3.131 with grain refinement, 3.134 overmodification, 3.132 quench modifier fibers, 3.131 Eutectic point, 1.4, 1.55 iron–carbon system, 1.4–1.6 Eutectic solidification, 3.54–3.68 criterion for growth, 3.61, 3.62 eutectic alloys, importance, 3.54 extent of eutectic range, 3.63, 3.64 irregular eutectic, 3.67, 3.68 microstructures, 3.55, 3.56 phase diagram, 3.58 regular lamellar growth, 3.56–3.64 rod eutectics, 3.64–3.67 spacing, 3.57 spacing selection, 3.59–3.61 Eutectic spacing selection, 3.59–3.61 Eutectic temperature, 1.55 Eutectoid composition (see Eutectoid point) Eutectoid point, 1.4 effect on, 1.18–1.20 Eutectoid steel, 1.9, 1.10 cementite, 1.9, 1.10 ferrite, 1.9, 1.10 Eutectoid reaction, 1.19, 7.3 divorced, 7.4 ledgewise cooperative growth, 7.4 Eutectoid temperature, 1.19, 1.20 Eutectoid transformation in ferrous and nonferrous alloys, 7.3 Evaporated plume, 18.31 Evaporative coatings, 18.24 Exaggerated grain growth (see Secondary recrystallization) Exothermic gas, 16.55, 16.102, 16.103 Exhaustion hardening, 4.13, 4.15 Externally heated furnaces, 16.75

Fatigue properties, 16.16, 17.10 resistance, 13.80, 17.10 strength and life, 7.53, 15.29, 15.58, 16.1, 16.16, 17.10, 17.11 FATT, 14.76 Faulty heat treatment practice, 17.1 FCP resistance, 15.40, 15.50 Feathery appearance, 9.5

INDEX

Ferrite: acicular, 9.24–9.29 alloying elements, 1.14–1.17 formers (or stabilizers), 1.14–1.17 morphology, 7.21, 7.22 (see also specific types) solidification, 10.25 solubility of C and N, 1.12, 1.14 (See also Alpha iron; Proeutectoid ferrite) Ferrite number, 10.63–10.65 Ferrite–pearlite and pearlitic steels: effect of grain size on yield stress, 7.43–7.46 effect of pearlite volume fraction on yield stress, 7.45, 7.46 equation for transition temperature, 7.43, 7.46 Ferritic thermochemical surface hardening (see Nitriding; Nitrocarburizing) Ferroelectric device, 18.80, 18.86, 18.87 Fibers, 18.79 CVD-coated, 18.79 Fibrous carbide precipitation, 7.15 characteristic features, 7.20 morphology, 7.20 orientation relationship, 7.20 structure, 7.20, 7.21 Fick’s first law, 2.2–2.4, 2.9, 2.33 Fick’s second law, 2.11–2.14, 2.86 Field effect transistors (FETs), 18.44, 18.81 ferroelectric (FeFET), 18.86 MESFET, 18.81 MOSFET, 18.81 Fine structural superplasticity, 15.82–15.92 Fisheyes, 14.100 Flame hardening, 16.3–16.11 advantages, 16.8, 16.9 applications, 16.11 disadvantages, 16.9 fuel gas, 16.3, 16.4, 16.9 materials selection, 16.9, 16.10 oxygen/fuel ratios, 16.3, 16.4 pre- and post-heat treatment, 16.10 procedure, guidelines, 16.10, 16.11 progressive method, 16.6 progressive-spinning method, 16.8 selection of hardening methods, 16.9 spinning method, 16.6–16.8 spot or stationary method, 16.5, 16.6 Flame spray (FS) (or combustion flame spray) coating: advantages and disadvantages, 19.9

I.15

applications, 19.9 characteristic features, 19.6 procedure, 19.6, 19.8 process parameters, 19.7 Flame spray and fuse process, 19.9, 19.10 applications, 19.10 variants, 19.10 spray and fuse (hardfacing), 19.10 spray-fuse, 19.10 Flakes, 14.100 Fleisher model, 8.78, 8.79 Flow stress, 4.18– 4.26 change during dynamic recrystallization, 15.16 change during recovery, 5.9, 5.10 effect of grain size, 4.21 effect of temperature, 4.19, 4.21, 4.26 strain hardening exponent, 4.19 and subgrain size, 5.9, 5.10 true, 4.18 Fluidized bed: boriding, 16.133, 16.134 carbonitriding, 16.80, 16.81 carburizing, 16.75 advantages, 16.75 procedure, 16.75 chemical vapor deposition: advantages, 18.111 applications, 18.110, 18.111 typical, reactor, 18.111 nitriding, 16.101, 16.102 advantages, 16.101 procedure, 16.101, 16.102 nitrocarburizing, 16.110, 16.111 advantages, 16.110, 16.111 applications, 16.111 characteristic features, 16.110 procedure, 16.110 quenching, 13.62, 13.63 Formability, 4.7, 15.45, 15.50 Forming limit diagram (FLD), 4.20 Fracture: brittle, 4.9 ductile, 14.65 intergranular, 14.67, 14.68, 14.82, 14.91, 17.3 quasi-cleavage, 14.93 transgranular, 14.67 Fracture toughness, 4.9, 4.10, 14.111, 14.114 Frank–Read source, 4.13 Free-energy–composition diagram, 6.15–6.17 coherent, 6.19, 6.20 incoherent, 6.19, 6.20

I.16

INDEX

Frenkel defect, 2.94, 2.95, 2.104 anti-, 2.94 Frenkel pairs, 2.104 Frenkel product, 2.94 Friedel process, 6.59 Fuel cells, 2.94 Full annealing, 12.4, 12.5 procedure, 12.4 purpose of, 12.5 representation on Fe-Fe3C and I–T diagrams, 12.4, 12.5 structure of transformation product, 12.4 Functionally graded coatings (or materials), 19.11, 19.25, 19.29, 19.43, 19.44 Furnace atmosphere, 16.56 parameters, 16.56 Furnaces, 16.48, 16.52, 16.55, 16.56, 16.79, 16.128 batch type, 16.48, 16.55, 16.56, 16.88, 16.89 continuous type, 16.55, 16.56

G phase, 10.81 Gamma iron, 1.2 crystal structure, 1.2, 1.3 octahedral holes, 1.2 solubility of C and N in, 1.12 tetrahedral holes, 1.2 Gamma loop, 1.14 g ¢-Fe4N, 16.81, 16.82, 16.91, 16.98, 16.110 g ¢ phase, 6.48, 6.49, 6.51–6.53, 10.76, 10.77, 15.58 g ≤ phase, 6.51, 6.52, 10.77 Gas carbonitriding, 16.76–16.80 advantages, 16.77 applications, 16.80 case composition, 16.78 case depth, 16.78, 16.79 choice of temperature, 16.79 control of g R, 16.79 carbonitrided and quenched microstructure, 16.78 disadvantages, 16.77, 16.78 effect of time and temperature on case depth, 16.78, 16.79 furnaces, 16.79 of P/M parts, 16.80 process, 16.76, 16.77 quenching media, 16.79 tempering, 16.79, 16.80 void formation, 16.79 Gas carburizing, 16.54–16.65 activity coefficient of C, 16.57

alloy effects, 16.63 atmosphere, 16.54, 16.56 atmosphere reactions, 16.57, 16.58 carbon concentration gradient, 16.63, 16.64 carbon profile, 16.65 carbon potential control, 16.63–16.65 carburizing reactions, 16.57, 16.58 carburizing step, 16.57 case depth, 16.54, 16.63 diffusion step, 16.57 effect of temperature, 16.59 effect of time and temperature on case depth, 16.60 equilibrium constant at a given pressure, 16.57 furnaces, 16.55, 16.56 gas carburized microstructure, 16.55, 16.56 in situ atmospheres, 16.58, 16.59 process variables, 16.59 reaction coefficient, 16.57 removal of decarburizing agent, 16.58 surface carbon content, 16.58 two-step cycle, 16.57 Gas nitriding, 16.88–16.93 advantages, 16.89 ammonia supply, 16.88 appearance of nitrided parts, 16.89 applications, 16.91 bright, 16.91 case depth, 16.90 double stage, 16.89 Floe process, 16.89 furnaces, 16.88, 16.89 nitreg, 16.91, 16.92 advantages, 16.92 applications, 16.92, 16.93 disadvantages, 16.92 nitriding potential, 16.91, 16.92 nitrided microstructure, 16.89, 16.90 pressure nitriding, 16.90 advantages, 16.90 drawbacks, 16.90, 16.91 process, 16.88 purging, 16.88, 16.89 safety precautions, 16.88 single stage, 16.89 Gas nitrocarburizing, 16.102 application, 16.102, 16.103 atmospheres, 16.102 oxynitrocarburizing, 16.105 with post oxidation, 16.105

INDEX

processing steps, 16.102 proprietary methods, 16.103 nitrotech, 16.103 application, 16.104 atmosphere, 16.103 properties, 16.103, 16.104 process, 16.103, 16.104 structures, 16.103, 16.104 Gas-source molecular beam epitaxy (GSMBE): advantages and disadvantages, 18.55, 18.56 application, 18.56 physical mechanism, 18.54 process, 18.53, 18.54 Gaussian distribution, 2.21, 18.3 Gibbs-Duhem equation, 2.51 Gibbs–Thompson equation, 3.60, 5.40, 7.6 Globular g, 10.6 Glow discharge sputter deposition: advantages and disadvantages, 18.36 fundamentals, 18.36 procedure, 18.36 GP zones: in Al-Ag alloys, 16.48–16.50 in Al-Cu alloys, 6.38–6.41 approximate composition, 6.39, 6.41 crystal structures, 6.39–6.41 detecting method, 6.39 Gerold’s model, 6.24, 6.41 in Al-Cu and Al-Cu-Mg alloys, 6.30, 6.31, 6.38, 6.39, 6.41– 6.45 in Al-Li base alloys, 6.46–6.48 in Al-Mg-Si alloys, 6.45, 6.46 in Al-Zn-Mg alloys, 6.46 in Cu-base alloys, 16.51 orientation–matrix relationships, 6.41 shapes, 6.38, 6.39, 6.41 Graded interface, 18.19 Gradient energy, 6.19, 6.21– 6.24 coefficient, 6.21 Grain boundary: diffusion coefficient and mobility, 5.45 orientation effect, 5.45 as dislocation sinks, 5.13 free energy, 5.43 coherent inclusion, 5.43 incoherent inclusion, 5.43 high angle, 5.13, 8.33 low angle, 5.14, 5.51, 8.33 migration, 5.40, 5.42, 5.44 chemical potential, 5.41 free energy difference, 5.41

I.17

in grain growth, 5.41 migration velocity, 5.41, 5.46 mobility, 5.40, 5.41 impurity effect, 5.42 orientation effect, 5.42, 5.45 second phase particles, 5.42, 5.44 temperature effect, 5.42, 5.46 texture effect, 5.46 thickness effect, 5.46 vacancy effect, 5.45, 5.46 oxidation, 10.88 strengthening, 4.25 Grain coarsening, 15.8, 15.10, 15.15 temperature, 15.8, 15.10 Grain boundary diffusion, 2.80–2.86 Harrison’s A-B-C-classification, 2.81, 2.83–2.86 Hart-Martlock equation, 2.84, 2.85 Lavine-MacCallum model, 2.85 measurement techniques: accumulation, 2.81–2.86 profiling, 2.80, 2.81 role, 2.80 Grain growth, normal, 5.39–5.50 driving force, 5.40, 5.44 effect of second phase particles, 5.42–5.45 equation, 5.46, 5.47 fault, 5.49 grain boundary migration, 5.40, 5.41 grain boundary mobility, 5.41, 5.42 grain coarsening behavior, 5.47 impurity effect, 5.42 inhibiting conditions, 5.50, 5.51 kinetics, 5.46, 5.47 maximum grain size, 5.44 other variable effects, 5.45, 5.46 restraining force, 5.43, 5.44 types, 5.39 Grain refinement, 15.19, 15.22, 15.67 Grain refinement of Al-Si alloys, 3.120–3.125, 3.135, 3.136 effect of, 3.135 inoculation methods, 3.122–3.124 mechanisms, 3.124, 3.125 thermal methods, 3.121 by vibration, 3.121 Grain refinement of Mg alloys, 3.125 Grain size: initial, 5.30 limiting, 5.44 Grain structure, equiaxed: soap-bubble experiment, 5.40 two-dimensional representation, 5.41

I.18

INDEX

Granular bainite, 9.11, 9.12 Graphite: compacted, 1.64 flakes, 1.6 factors affecting, 1.60, 1.61 length, 1.60 morphology, 1.57 nucleation and growth of, 1.60 spherulite, 1.60–1.62 effect of minor elements, 1.62, 1.68 microstructures, 1.59, 1.62 morphologies, 1.62 temper carbon nodules, 1.65 Graphitization, 1.6 Graphitization of steel, 12.21–12.23 Graphitizing potential, 1.61 Gray iron: alloying elements in, 1.66–1.68 applications, 12.44, 12.48, 12.49 classification, 1.60 compressive strength, 12.39 damping capacity, 12.44, 12.46 eutectic growth morphology, 3.137, 3.138 graphite flakes, 1.57, 1.60 fatigue limit, 12.42, 12.43 fatigue notch sensitivity, 12.42 fatigue properties, 12.41 general, 1.57 hardness, 12.41 inoculation, 1.6 machinability, 12.44, 12.55 mechanical properties, 12.36, 12.40, 12.60 modulus of elasticity, 1.60, 12.38 plane strain fracture toughness (KIC), 12.67 stress–strain curves, 12.37–12.39 structure–property relations, 12.67 tensile strength, 1.60, 12.36–12.38 tensile strength/hardness ratio, 12.39, 12.41, 12.42 torsional shear strength, 12.41 wear resistance, 12.43, 12.44 Greninger–Troiano model, 8.28 Griffith–Owen elastic-plastic stress analysis, 14.70 Grinding: of decarburized and nitrided layers, 17.31, 17.49, 17.50 in selective case hardening, 17.31, 17.49, 17.50 of thermal sprayed parts, 19.10, 19.31, 19.33 Grit blasting, 19.6, 19.22, 19.33, 19.45

Growth fault, 5.49

Hackney-Shiflet theory, 7.4 Hall effect, 18.39 Hall–Petch equation, 4.22, 4.23, 8.79, 8.80, 15.31, 15.34 Hard coatings, 18.22, 18.23, 18.25, 18.30, 18.31, 18.40, 18.57, 18.72–18.74, 18.88 Hardenability: bands, 13.118, 13.134 boron effect, 13.100 computer calculation of Jominy, 13.131 CHAT process, 13.131 Creusot-Loire method, 13.131 definition, 13.98, 13.99 DeRetana and Doane approach, 13.114 DI calculation for nonboron steels, 13.125, 13.126 DI for boron steels, 13.125 Grossman method, 13.101–13.110 DI values for various steel grades, 13.101 effect of grain size and composition, 13.101, 13.112 empirical equation for DI, 13.103 50% martensite location, 13.101 ideal critical diameter (DI), 13.101 ideal quench, 13.101 limitations, 13.110 multiplying factors for alloying elements, 13.104–13.113 procedure, 13.101 reexamination of, 13.110 relation between DI and D for given H values, 13.101, 13.102 hot brine test method, 13.113–13.116 additive effects of minor elements, 13.113, 13.114 based on 90% martensite structure, 13.113 hot-wire brine test, 13.113 Jatczak approach, 13.113, 13.116 case hardenability, 13.113 core hardenability, 13.113 importance of Mo, 13.113 Jominy end-quench method, 13.116, 13.117 advantages, 13.116 controlling parameters during heat treatment, 13.116, 13.117 hardenability band, 13.118, 13.134 deep hardenable, 13.138, 13.139

INDEX

shallow hardenable, 13.134, 13.138 modified, 13.118, 13.119 procedure, 13.117, 13.118 shortcomings, 13.116 Jominy inflection points: calculation of Jominy curves, 13.123–13.130 DI for boron steels, 13.125, 13.126 DI for nonboron steels, 13.123– 13.125 Just method for calculating Jominy curves, 13.129 Kirkaldy method for calculating Jominy curves, 13.130 Kramer approach, 13.115, 13.116 methods of determining, 13.116–13.133 Minitech computerized hardenability predictor, 13.133 Moser–Legat approach, 13.116 and nucleation theory, 13.99, 13.100 shallow hardenability tests, 13.118 air hardenability test, 13.120–13.122 modified Jominy tests, 13.118, 13.119 SAC hardenability test, 13.119 use of Jominy curves, 13.130, 13.131 Hardenability (or H-) steels, 13.133–13.140 advantages of alloy H-steels, 13.139 alloy steel selection guidelines based on, 13.140–13.147 boron hardenability, 13.123 methods of specifying, 13.139, 13.140 Hardenable diameter (DH), 13.113 Hardening power (see Cooling rate; Severity of quench) Hardfacing coatings, 19.10, 19.16, 19.17 (See also Flame spray and fuse process; Plasma transferred-arc spraying) Hardness in hardened steels, 13.96–13.98 in form of Hall-Petch relationship, 7.47 of martensite using different indentors, 13.94, 13.96 versus C content for different amounts of martensite, 13.97, 13.98 Harris equation, 16.60, 16.63 HAZ cracking, 3.112, 9.44 Head hardening, 7.48, 7.49, 9.37, 9.42 Heat affected zone (HAZ), 9.31, 15.26 delayed or underbead cracking, 14.100, 14.101 Heat treatment, general, 12.1 classification, 12.1, 12.3 (See also specific types) definition, 12.1

I.19

importance, 12.1, 12.2 processes, 12.3 purpose of, 12.2 selection, 12.1, 12.2 Herring-Nabarro diffusional creep, 15.17, 15.87 Heteroepitaxial structure, multilayer, 18.54 Heterogeneous nucleation, 6.8–6.15 at container walls, 6.9, 6.10 critical free energy for, at grain boundary, 6.10 of grain corners, 6.10–6.12 on grain edges, 6.10–6.12 of martensite: potential nucleation sites, 8.43 by a continuous strain modulation path, 8.43 preexisting embryos, 8.43 superdislocations, 8.43 Heterojunction bipolar transistors, 18.43, 18.56, 18.82, 18.85, 18.99 Heusler alloys, 8.48, 8.57 High-alloy steels, 17.37 High-carbon, high chromium cold work tool steels, 1.47, 1.49 High electron mobility transistors (HEMTs), 18.99 High energy plasma (HEP), 19.35, 19.49 High-frequency resistance hardening, principle, 16.30, 16.31 High pressure plasma spraying (HPPS), 19.16 High-speed tool steels, 1.47–1.49, 14.28–14.36 advantages, 1.47, 14.28 applications, 14.28–14.30 classification, 1.47–1.49 precipitation sequence on tempering T1 steel, 14.34 quenching and tempering of: M2 steel, 14.35–14.37 T1 steel, 14.32–14.35 wear resistance comparison, 14.28, 14.31 High strain rate superplasticity, 15.92, 15.94–15.98 High-strength low-alloy (HSLA) steels, 1.45, 1.46, 15.31–15.39 applications, 14.28–14.30 classification, 1.45, 1.46, 14.28, 15.33, 15.34 mechanical properties, 15.34–15.37 processing methods, 15.31–15.33 strengthening mechanisms, 15.34, 15.38

I.20

INDEX

High-strength low-alloy (HSLA) steels (Cont.): toughness. 15.38, 15.39 High-Tc superconductors, 18.23, 18.31, 18.41, 18.59, 19.25, 19.26 High-temperature thermomechanical treatment (HTMT), 15.28, 15.29 advantages, 15.28 applications, 15.28, 15.29 procedure, 15.28 High-velocity oxyfuel (HVOF) spray coating: advantages and disadvantages, 19.18, 19.19 applications, 19.19, 19.20 coating characteristics, 19.18 process, 19.17, 19.18 process parameters, 19.17, 19.18 types of, 19.17 Hillert-Staffansson regular solution treatment, 7.25, 7.26 Hillert-Trivedi growth equation, 7.32 HIPing or Hipping (see Hot isostatic pressing) Hole expansion ratio, 15.51 Holes and hillocks in thin films, 2.73, 2.75, 2.77 Homo- and heterostructure materials, 18.44 Homogeneous nucleation, classical, 6.2–6.7 coherent nucleus, 6.4 critical free energy, 6.4 critical number of atoms, 6.4 critical radius, 6.4 of martensite, 8.41, 8.42 critical activation free energy barrier, 8.42 critical dimensions, 8.42 nucleation rate, 8.42 strain energy factor, 8.41 Homogenize annealing, 12.4–12.6 Hertwich continuous process, 12.6 Hot cathode triode (see Triode sputter deposition) Hot deformation parameters, 15.3 Hot isostatic pressing, 18.31, 19.25, 19.28, 19.29 Hot-mill, 15.48 Hot rolling, 15.3, 15.22, 15.48 precipitation kinetics of microalloying during, 15.23, 15.24 Hot-wall reactor, 18.88 Hot work tool steels, 1.47, 1.49 Hot workability, 15.17

Hot working, 15.12 Hydrogen attack (HA), 14.97–14.100 control using Nelson curves, 14.97–14.99 and internal and surface decarburization, 14.97 methods for minimizing, 14.97, 14.98, 14.100 occurrence, 14.97 Hydrogen damage of steels, 14.84–14.101 classification, 14.84 other forms of, 14.101 Hydrogen embrittlement (HE), 14.84–14.97 control of, 14.95–14.97 metallurgical modification, 14.96 microstructural control, 14.96 reversibility of HE, 14.97 TMT, 14.96, 14.97 forms of: delayed failure cracking, 14.90, 14.91 grain boundary impurity concentration, 14.91–14.93 hydrogen induced cracking (HIC), 14.88 characteristics, 14.88 stages of, 14.89, 14.90 and stress intensity factor, 14.89, 14.90 various terminologies, 14.88 in weldments, 14.88 hydrogen induced ductility losses, 14.94, 14.95 embrittlement index (or RA loss), 14.94 factors affecting, 14.95 sulfide stress cracking (SSC), 14.93, 14.94 guidelines for safe operation, 14.94 susceptibility, 14.94 hydrogen entry, transport, trapping and, 14.85, 14.86 theories of, 14.85, 14.86 decohesion mechanism, 14.87 hydride formation, 14.88 hydrogen-enhanced local plasticity (HELP), 14.86, 14.87 internal pressure mechanism, 14.88 local hydrogen pressure mechanism, 14.86 surface energy mechanism, 14.87 Hydrogen fugacity, 14.90 Hydrolysis, 18.59, 18.62, 18.63 Hypereutectic Al-Si alloys, primary Si refinement: effect of modification, 13.136

INDEX

effect on properties, 3.135 Hysteresis loop, 4.13 Hysteresis loss, 1.22, 5.53

Ideal critical diameter (DI), 13.101–13.104 correlation with critical diameter, 13.101, 13.102 as function of composition and grain size, 13.101 as function of critical diameter, 13.101, 13.103 multiplying factors, 13.102–13.109 Ideal quench, 13.101 Idiomorphic ferrite, 7.21, 7.41 Imanite steel, 16.83, 16.84 Immiscible alloy systems, 18.13 Impact properties of ausformed steels, 15.29 bainitic steels, 9.36, 9.37 effect of grain size, 10.7 effect of tempering, 14.23–14.25 of ferrite-pearlite and pearlitic steels, 7.43, 7.45, 7.46 HSLA steels, 15.38, 15.39 HTMT, 15.28 microalloyed forging steels, 15.27 Impact transition temperature (see DBTT) Impregnation (or sealing), 19.24 application, 19.30 types: inorganic sealers, 19.30 organic sealers, 19.30 vacuum, 19.30 Improved low pressure (ILP) casting process, 3.161, 3.163 Impurity effect: on embrittlement, 14.68–14.70, 14.72, 14.74–14.81, 14.85, 14.91 on grain boundary mobility, 5.42 Inclusions: and grain boundary mobility, 5.42–5.45 MnS type, 1.21, 1.22 shape control, 15.33, 15.34 Incoherent nucleation, 6.6, 6.7 vacancy effect, 6.7 Incoherent boundary, 7.40 Incubation time, 3.133, 5.10, 5.15, 5.19, 5.23, 5.24, 6.8, 15.21 Indium tin oxide coating (ITO), 18.88 Induction hardening, 16.11–16.30 advantages and disadvantages, 16.15–16.17

I.21

applications, 16.28–16.30 coil impedance matching, 16.23 computer simulation, 16.28, 16.29 current frequency, 16.12 depth of penetration, 16.12, 16.13 equipment, safety practices, 16.30 gear hardening, 16.15 heating pattern, 16.18, 16.19 heating time, 16.12 induction coil design, 16.17–16.22 clamshell coils, 16.18 hairpin coils, 16.18 internal coils, 16.18 longitudinal-flux heating coils, 16.21, 16.22 low-frequency, 16.17 medium to high frequency, 16.17, 16.18 quick-change coil design, 16.20 scan inductor coils, 16.20 split coils, 16.19 transverse-flux heating coils, 16.20, 16.21 magnetic flux concentrators, 16.22 advantages and disadvantages, 16.22, 16.23 applications, 16.22, 16.23 requirements of, 16.22 selection of materials, 16.22 power rating (or density), 16.12, 16.14 power supplies, 16.23–16.26 capacitors, 16.26 diode, 16.23, 16.24 IGBT, 16.26 MOSFET, 16.26 thyristor, 16.23, 16.24 transformers, 16.26 transistor, 16.23–16.25 vacuum tube oscillators, 16.26 pre- and postheat treatment, 16.27, 16.28 principles of, 16.11 procedure, 16.14 progressive hardening, 16.15 single shot (or SI), 16.14 selection of power and frequency, 16.12 skin effect, 16.12 steel grades, 16.26, 16.27 Inductively coupled plasma (ICP), 18.42 Induction tempering, 16.27, 16.28 Inert (or shrouded) plasma spraying (IPS or SPS), 19.15, 19.16 advantages and disadvantages, 19.16 Infrared gas analyzer, 16.64

I.22

INDEX

Ingot casting, 3.93 classification of steel ingot, 3.93 capped steel, 3.95 killed steel, 3.93, 3.94 rimmed steel, 3.94, 3.95 semikilled steel, 3.94 Ingot structure, 3.95 chill zone, 3.96 columnar-to-equiaxed transition, 3.97, 3.98 columnar zone, 3.96, 3.97 equiaxed zone, 3.97 Inhomogeneous (or complementary) shear, 8.28, 8.31, 8.32 Initial transient period, 6.8 Inoculant, 3.13 Inoculation, 1.60 methods, 3.122, 3.123 Instability, plastic, 4.10–4.12, 15.50 Insulating coatings, 18.20 Intense quenching of steel, 13.69–13.72 advantages and disadvantages, 13.69–13.71 design of a typical fixture, 13.71, 13.72 guidelines for successful operation, 13.71 Interband spacing, 7.16 Interconnected porosity, thermal spray coatings, 19.12, 19.13, 19.21, 19.30, 19.33–19.35 Intercritical annealing, 15.45–15.49 Interfaces: coherent, 7.40 incoherent, 7.35, 7.40 partially coherent, 7.34 Interfacial energy, 12.10 Intergranular embrittlement by AlN control of, 14.82 occurrence of AlN phase, 18.41 panel cracking in ingots, 14.82 reduced hot ductility, 18.43 rock candy fracture, 14.82 Intergranular fissures, 14.97 Intergranular fracture: effect of P, Sb, Sn, 14.72, 14.74–14.76 in filamentary wires, 5.52 in hydrogen embrittlement, 14.91 mode of TME, 14.67, 14.68 overheating and burning, 17.2–17.5, 17.8 temper embrittlement, 14.72, 14.74– 14.76 Interlamellar spacing, 4.1, 7.1, 7.2, 7.6–7.9, 7.45–7.47

critical, 7.6, 7.7 dependence on supercooling, 7.6, 7.7 effect of pearlite volume fraction, 7.9 effect on mechanical properties, 7.45– 7.47 maximum growth rate criterion, 7.6, 7.7 mean true, 7.47, 7.48 Interledge spacing, 7.4, 9.23 Intermetallic phase precipitation, 10.76–10.81 Intermetallics, superplasticity, 15.93, 15.94 Internal cavitation, 15.87–15.90 Internal friction, 8.73 curves, 14.81 mode, 2.24 Internal stress (see Residual stress) Internal twin, 8.28 Interphase precipitation, 7.15 Baker-Nutting orientation relationship, 7.17 curved, 7.17, 7.18 interband spacing, 7.16, 7.20 ledge mechanism, 7.17 in microalloyed steels, 7.16 morphology, 7.16 partially coherent facets, 7.16, 7.17 planar, 7.16, 7.17 structure, 7.15–7.17 Interrupted quenching, 8.13, 8.14 Interstitial free steels, 1.33, 1.34, 15.26 Interstitial mechanism, 2.27 Interstitial-substitutional diffusion, 2.102, 2.103 Interstialcy mechanism, 2.27, 2.28 Intrinsic point defects, 2.97 equilibrium condition, 2.97 nonequilibrium condition, 2.97, 2.98 self-interstitials, 2.97 Invariant plane strain, 9.2, 9.14 Inverse bainite, 9.12 Inverse Kirkendall effects, 2.110, 2.111 Ion beam accelerator, 18.2, 18.3 Ion beam assisted deposition (IBAD) process: adhesion characteristics, 18.2, 18.17 advantages and limitations, 18.15 applications, 18.15 DLC coatings, 18.15, 18.16 friction and wear resistance, 18.14, 18.17 high temperature oxidation resistance, 18.17 ion induced CVD, 18.17

INDEX

optical films, 18.15 polymers, 18.15 sputter deposition, 18.14 types of coatings, 18.15 wear and corrosion resistant coatings, 18.16, 18.17 classification, 18.13 ion energy range, 18.14 methods, 18.13 Ion beam mixing deposition: advantages, 18.13 cascade mixing, 18.11, 18.12 main configurations of, 18.11 mechanisms, 18.11 principle, 18.10, 18.11 Ion beam processes, 18.1–18.21 classification, 18.1 Ion channeling, 18.5 Ion gun (or source), 18.2, 18.17 Ion gauge, 18.46 Ion implantation: adhesion characteristics, 18.7 advantages and disadvantages, 18.5– 18.7 applications: ceramics, 18.10 material synthesis, 18.10 metals, 18.5, 18.6, 18.8, 18.9 nuclear, 18.10 optical materials, 18.10 other metallic coatings, 18.8 polymers, 18.10 semiconductors, 18.8, 18.10 sputter deposition, 18.21 concentration profile of implanted ions, 18.3 formation of amorphous alloys, 18.4 formation of metastable phases, 18.4 fundamentals, 18.1 ion beam accelerator, 18.2, 18.3 ion channeling, 18.5, 18.8 multiple, 18.6 penetration depth, 18.2, 18.7 plasma immersion (PIII), 18.5 plasma source (PSII), 18.5, 18.6 metal vapor vacuum arc (MEVVA) source, 18.5 stopping power, 18.2, 18.3 Ion implanted layer, 18.2 Ion induced CVD, 18.17 Ion nitriding (see Plasma nitriding) Ion plating (or ion vapor deposition): advantages and limitations, 18.19, 18.20

I.23

applications, 18.20, 18.21 methods used to bombard the film, 18.17 reactive plasma, 18.17, 18.18 types: cathodic-arc (or ion bond)-based, 18.17, 18.19 plasma-based, 18.17, 18.18 vacuum-based, 18.17, 18.18 Ionic solids, 2.93–2.96 Ionized PVD (I-PVD), 18.41–18.43 schematic of, 18.42 Iron: bcc structure, 1.1, 1.2 fcc structure, 1.1, 1.2 flow stress, 4.18– 4.23, 4.25, 4.26 plastic flow, 4.12– 4.18 slip systems in crystalline solids, 4.12– 4.14 temperature dependence of stress–strain curves for mild steel, 4.26 Iron alloy phase diagrams: Fe-C phase diagram, 1.4–1.6 Fe-Cr phase diagram, 10.21, 10.22 Fe-Cr-Ni-C phase diagram, 10.24 Fe-Cr-Ni phase diagram, 10.26 Fe-Fe3C phase diagram, 1.4, 1.5 Fe-Fe3C-Si ternary phase diagram (sectioned), 1.55, 1.56 Fe-N phase diagram, 16.81, 16.82 Fe-Ni phase diagram, 10.23 Iron nitrides, 16.81, 16.82 Irradiation damage, 5.5, 5.6 Irradiation defects, 2.104–2.108 bubbles, 2.106–2.108 Cascade defects, 2.105, 2.106 defect clusters, 2.106 voids, 2.106 Irradiation-induced creep, 18.10 Irradiation-induced segregation and precipitation, 2.109, 2.110 Ishawaka diagram, 17.34, 17.35 Isoforming, 15.30, 15.31 Isothermal annealing, advantages and application, 12.12 Isothermal martensitic transformation, 8.17–8.21 Isothermal reaction time, 6.8 Isothermal transformation (IT) diagram: partial, 9.20, 9.21, 13.4, 13.5 (see also TTT diagrams) Isotope mass effect, 2.50 ITT, 7.43, 7.44, 7.46, 9.36 Ivantsov’s treatment, 3.45, 3.48, 7.32

I.24

INDEX

Jackson-Hunt analysis, 3.61, 3.62 Jackson-Hunt criterion, 3.60 Jeffries method, 10.12, 10.13 Jerky flow, 4.17 Johnson-Mehl-Avrami-Kodmogorov (JMAK) equation, 5.15, 5.17–5.20, 5.22 Jominy method, 13.116–13.118 curves, 13.116–13.118, 13.123–13.131, 13.133 distance, 13.126, 13.128, 13.129 end quench method, 13.116–13.118 modified, 13.116, 13.119, 13.120 Jones-Trivedi treatment, 7.35

KI, 14.89 KIC, 4.9, 4.10, 14.64, 14.65, 14.89, 14.90 Kth, 14.89, 14.90 k-carbide, 9.11 Kaufman-type ion gun/source, 18.13, 18.15 Kickout mechanism, 2.31, 2.102, 2.103 Kinetics: of allotriomorphic growth, 7.22–7.25 of grain growth, 5.46, 5.47 of pearlite formation, 7.9–7.13 of recovery, 5.10–5.12 of recrystallization, 5.15–5.23 of Widmanstätten plate growth, 7.30–7.35 Kirkendall effect, 2.60–2.63 experiment, 2.61, 2.62 inverse, 2.110, 2.111 voids and porosity, 2.61, 2.62, 15.105 Kirkendall void, 15.105 Knudsen cells (or effusion cells, effusion ovens, or k-cells), 18.44, 18.54, 18.55, 18.102 Kroeger-Vink notation, 2.95

Lamellar pearlite growth, 7.5–7.7 Lamellar pearlite reaction, 7.2–7.5 Lamellar splats, 19.1 Lamellar structure, 1.4, 7.1, 7.2 Lamellar tearing, 15.39 Lamp power switching, 18.91 Lapping, 19.31 Laser ablation: advantages and disadvantages, 18.33 application, 18.31 characteristics of film deposition, 18.31, 18.32 lasers, 18.32

typical deposition systems, 18.31, 18.32 Laser CVD (LCVD): advantages and disadvantages, 18.95, 18.96 growth mechanism, 18.95, 18.96 laser writing, 18.96 Laser diodes, 18.43, 18.48, 18.50, 18.56, 18.82, 18.83 Laser spraying process, 19.22 Laser surface hardening, 16.31–16.37 advantages and disadvantages, 16.35–16.37 applications, 16.36 elements of LBHT machine, 16.31–16.33 energy absorbing coating, 16.33 factors affecting surface hardness and case depth, 16.33–16.35 focusing of beam, 16.32, 16.33 selection of, 16.37 theoretical models, 16.37 Laser treatment: laser engraving, 19.25, 19.28 laser remelting, (or glazing), 19.25, 19.27 Laser writing, 18.96 Lath martensite, 8.28, 8.29, 8.33, 8.34 Lattice invariant shear, 8.30, 8.31 Lattice matching theory, 9.28, 9.29 Lattice shear, mode of, 4.39 Lattice strain, 17.27 Lattice structures: bcc, 4.1, 4.2 bct, 8.24–8.26 fcc, 4.2, 4.3 Laue pattern, 5.6, 5.7 Laves phases, 10.79, 10.80, 14.108 Layered alloys, 18.105 Layered structures, 18.45, 18.105 LBm, 9.12, 9.24 LBs, 9.7, 9.8, 9.24 Lead bath, 13.57, 13.58 Ledeburite, 1.6 Ledge growth mechanism and kinetics, 7.17, 7.18, 7.23, 7.25 Ledge mechanism, 7.34, 7.35, 9.17 Lenticular martensite, 8.35–8.37 Lever rule, 1.10, 1.11 Lifshitz–Slazov–Wagner theory, 10.76 Light emitting diode (LED), 18.44, 18.48, 18.49, 18.56, 18.82–18.84, 18.98 Limiting draw ratio (LDR), 4.44, 4.45 Liquid carbonitriding (see Cyaniding) Liquid carburizing (see Salt bath carburizing) Liquid infiltration, 19.30

INDEX

Liquid metal embrittlement, 14.102–14.106 forms of, 14.102 prerequisites for occurrence of, 14.102 steel, by various liquid embrittlers, 14.102–14.106 Liquid nitriding (see Salt bath nitriding) Liquid nitrocarburizing (see Salt bath nitrocarburizing) Lissajous curves, 16.32 Lithographic masking, 18.5 Local equilibrium, 7.22, 7.26 Locked-in stress (see Residual stress) Lorentz force, 19.24 Low-angle boundaries, 9.4, 14.19 Low-energy electron diffraction (LEED), 18.94 Low-pressure plasma spraying (LPPS), 19.15 applications, 19.15 basic characteristics, 19.15 process, 19.15 Low-temperature superplasticity, 15.93, 15.95, 15.99 Lower bainite, 9.7, 9.8, 9.10, 9.11, 9.20–9.23 (See also Bainite, mechanisms, lower) Lower yield point, 4.15, 4.16, 4.18 Lubricant coatings, dry, 18.23, 18.31, 18.36 Lüders strain (or band), 4.17, 4.18, 4.23, 4.30, 4.31, 4.33, 15.66 Jerky flow, 4.17 serrated yielding, 4.17, 4.32 Ludwik-Hollomon equation, 4.18

Macroparticles (MP), 18.28 Magnetic aging, 6.28 Magnetic bubble memories, 18.10 Magnetic films, 18.25, 18.35 Magnetic fusion energy devices, 19.49 Magnetic hardening, 4.30 Magnetic permeability, 10.48 Magnetic property, 1.43, 10.99, 10.104 Magnetic soft (ferrous) sheets, 5.52 Magnetic storage device, 18.40 Magnetron sputter deposition: advantages and disadvantages, 18.40 application, 18.40 closed-field unbalanced, 18.41 configurations, 18.39 definition, 18.39 principles, 18.39 Magnetron sputtering (see Magnetron sputter deposition)

I.25

Malleable iron: application, 12.64–12.66 choice over ductile iron, 12.64 damping capacity, 12.64 fatigue properties, 12.64 impact properties, 12.64 machinability, 12.64 mechanical properties, 12.63, 12.65, 12.66 modulus of elasticity, 12.64 stress–strain curves, 12.63 tensile properties, 12.63 wear resistance, 12.64 Manning’s equation (or vacancy wind term), 2.64, 2.66 Maraging steels, 1.51, 14.108–14.119 age hardening constituents, 14.113 applications, 14.118, 14.119 compositions, 14.108, 14.110 developments of, 14.110 disadvantages, 14.118 heat treatment, 14.108, 14.112–14.114 role of alloying elements and precipitation reactions, 14.113, 14.115 austenite reversion, 14.116, 14.117 Co/Mo interaction, 14.115 ordering reaction, 14.116 thermal embrittlement, 14.118 typical hardness vs. aging curves, 14.115 iron–nickel phase diagram, 14.108, 14.109 salient features, 14.111, 14.112 types of, 14.110, 14.111 Martempering: of gray cast iron, 13.74, 13.77 of steel, 13.72–13.77 advantages, 13.73, 13.74 applications, 13.74–13.76 carbomartempering, 13.74 modified, 13.74 procedure, 13.72, 13.73 superimposed on TTT diagram, 13.66 Martensite: athermal, 8.9 as a function of temperature, 8.10–8.12 critical nucleus radius, 8.42 critical semithickness, 8.42 definition, 8.1 embryos, 8.40–8.43 ferrous: Bain distortion (or strain), 8.27, 8.28 Bain mechanism, 8.27, 8.28

I.26

INDEX

Martensite (Cont.): banded morphology, 8.36 butterfly martensite, 8.36 c/a ratio as a function of C content, 8.25 crystal structure, 8.24, 8.25 effect of C content on lattice parameter, 8.25, 8.26 Greninger–Troiano model, 8.28 inhomogeneous shear of crystallographic theory, 8.28 lattice invariant deformation, 8.28 growth of plates: glissile semicoherent interface movement, 8.44 lattice invariant shear transformation, 8.44 hardness as a function of C content, 13.96, 13.97 hardness compared to various carbides, 14.28, 14.31 importance of properties, 8.2 lath (Type I), 8.4, 8.33, 8.34 dislocation densities, 8.33, 8.34 effect of carbon content, 8.34 features, 8.33 packet, 8.33 structure, 8.28, 8.29 lenticular, 8.35–8.37 methods for detecting nucleating defects, 8.43, 8.44 nucleation of, 8.40–8.43 free energy barrier, 8.42 plate (Type II) habit plane, 8.4–8.6, 8.34 midrib, 8.4, 8.34–8.36 twinning, extent of, 8.4, 8.34, 8.35 structure, 8.4, 8.28, 8.29 thin plate, 8.28, 8.30, 8.36 strengthening mechanisms, 8.75–8.81 surface martensite, 8.36 toughness of, 8.81 nonferrous, 8.17, 8.36, 8.38, 8.39, 8.46 morphologies, 8.39 ordered structure, 8.46 Martensite, thermoelastic: assumptions, underlying, 8.46 characteristics, 8.45, 8.46 detwinning, 8.46 equation for dissipated frictional work, 8.46 shape memory alloys, 8.47–8.59 applications, 8.67–8.75

fine scale deformation, 8.48 internal faulting and twinning, 8.48 limitations, 8.48 in single crystals, 8.59 stacking sequence, 8.48 stages of deformation process, 8.59 stress–strain curve for Cu-alloys, 8.59, 8.62, 8.63 shape memory effect, 8.59–8.63 pseudoelastic effects (PE), 8.63 applications, 8.63, 8.64 characteristics required, 8.63 in Cu-Al-Ni alloy, 8.63 loop, 8.65 in other alloys, 8.50, 8.63 rubberlike behavior, 8.64–8.66 superelasticity, 8.64 two-way shape-memory (TWSM) effect, 8.59, 8.61–8.63 methods, 8.61–8.63 Martensite-(retained) austenite (M–A) islands, 15.45–15.48, 15.50–15.52 Martensite stabilization, 8.67 Martensite-to-martensite transformation, 8.66, 8.67 Martensitic transformation: athermal, 8.9–8.17 definition, 8.1 effect of applied stress, 8.15 effect of carbon content, 8.12 effect of temperature and time, 8.10–8.12 Md, 8.15, 8.23 Mf, 8.10, 8.12, 8.13 definition, 8.10 effect of carbon content, 8.12 Ms, 8.10, 8.12, 8.13, 8.15, 8.23 effect of applied stress, 8.15 effects of magnetic field and hydrostatic pressure, 8.15, 8.16 phenomenological theory of martensitic crystallography (PTMC), 8.28–8.30 mathematical expression, 8.31–8.33 retained austenite, 8.14, 8.15 reversibility of, 8.12, 8.13 stabilization of austenite, 8.14–8.16 thermal stabilization, 8.13, 8.14 burst: characteristics of, 8.21, 8.22 habit plane, 8.22 MB, 8.21, 8.22 mode of, 8.21 structure of, 8.22

INDEX

transformation curves, 8.22 definition, 8.1 first order, 8.4 general, 8.3–8.9 affine transformation, 8.4 fine inhomogeneous substructure, 8.9 growth velocity, 8.8, 8.9 habit plane, 8.3–8.6 invariant plane strain, 8.4 lattice orientation relationship, 8.6–8.8 midrib, 8.4 nature of g -martensite interface, 8.3 shape change and surface relief, 8.3, 8.4 strain components, 8.4 isothermal: characteristics, 8.18, 8.20, 8.21 effect of changes in grain size, 8.18, 8.19 effect of magnetic field, 8.18 in ferrous alloys, 8.17, 8.18 in nonferrous alloys, 8.17 transformation kinetics, 8.17, 8.18 mechanically induced, 8.22–8.24 strain-induced, 8.22–8.24 stress-assisted, 8.22, 8.23 nucleation and growth in, 8.39, 8.40 activation free energy barrier, 8.40 driving force, 8.40 free energy change, 8.39, 8.40 growth of martensite plate, 8.44 kinetics, 8.44, 8.45 methods of detecting nucleating defects, 8.43, 8.44 nucleation in Lin-Olson-Cohen theory, 8.45 Massive (ferrite) transformation, 7.35–7.41 characteristic features, 7.35, 7.36 coherent growth mode, 7.40 critical cooling rate, 7.36, 7.40, 7.41 effect of transformation temperature, 7.36, 7.40, 7.41 free-energy–composition curve, 7.36, 7.38 general aspects, 7.35–7.40 growth rate, 7.35 in ferrous systems, 7.37–7.41 in nonferrous systems, 7.38, 7.39 paraequilibrium, 7.41 schematic phase diagram, 7.37, 7.38 Materials synthesis, 18.10, 18.13 Maximum growth rate criterion, 7.6, 7.7 Maximum restraining force, 5.43, 5.44 McQuaid–Ehn test, 10.7 Mechanical bonding, 19.1, 19.17

I.27

Mechanically induced transformation, 8.15 Mechanodiffusion, 2.72 Metal-induced embrittlement, 14.102 difference between LME and SME, 14.108 liquid metal embrittlement, 14.102– 14.106 of steel, 14.102–14.106 solid-metal embrittlement, 14.106 characteristics of, 14.106, 14.107 Metal-matrix composite solidification, 3.149–3.154 dispersion process, 3.152, 3.153 infiltration casting process, 3.150, 3.152 reactive processing (in situ MMCs), 3.153, 3.154 spray casting process, 3.153 Metal oxide semiconductor (MOS), 18.51, 18.53, 18.81, 18.91 Metallic glass formation, 19.25 Metallization: plastics and ceramics, 18.24, 18.27, 18.31 semiconductor wafers, 18.29, 18.35 technology, 18.27, 18.28 Metalorganic CVD (MOCVD) or OMVPE or MOVPE advantages and limitations, 18.100, 18.101 applications, 18.98–18.100 growth of group III–V semiconductors, group II alkyls, and group V hydrides, 18.100 procedure and prerequisites, 18.98 schematic rotating-disk reactor, 18.100, 18.101 Metalorganic molecular beam epitaxy (MOMBE), 18.49–18.53 advantages and disadvantages, 18.53 deposition of group III–V compound semiconductors, 18.49 schematic diagram of MOMBE growth apparatus, 18.51, 18.52 Metalorganic vapor phase epitaxy (MOVPE) [see Metalorganic CVD (MOCVD)] Metastable austenite, 6.54 Metastable austenitic steels, 10.84 transformation behavior, 8.24 Metastable cementite, 1.3 Metastable g ¢ particles, 10.76 Metastable phase diagrams, 1.4, 1.5, 6.17, 6.19, 14.109 Metastable phase transformation, 9.16, 16.54–16.57, 16.59

I.28

INDEX

Metastable phases, 6.1, 18.10 Methane, 12.35, 16.65–16.68, 18.110, 18.111 Microalloy steels (see High-strength lowalloy steels) Microalloyed addition, 15.3, 15.8 Microalloyed ferrite-pearlite forging steels, 7.63 Microalloyed precipitates, 15.7 Microalloyed steels, 7.63, 15.4, 15.26–15.28 Microalloying and TMT: Cr-Mn austenitic steels, 15.28 4130 steel, 15.28 high C steel wire rod, 15.27, 15.28 high strength linepipe steels, 15.26 medium- and high-C microalloyed steels, 15.26, 15.27 microalloyed forging steels, 15.27 ultralow C steels or IF steels, 15.26, 15.27 weldable rebars, 15.27 Microalloying elements (MAEs), 15.4, 15.7, 15.11, 15.26 Microcracking, 16.31–16.33, 17.30, 17.31 Microduplex stainless steels, 15.92 Microelectronics, 18.23, 18.35, 18.57, 18.92 Microporosity and shrinkage, 3.134, 3.135 Migration process, 5.42 Mirror planes, 4.35 Miscibility gap, 6.15–6.17 coherent, 6.19, 6.20, 6.48 incoherent, 6.19, 6.20 Misfit: dislocations, 9.14 strain, 6.3, 6.61, 6.69 Mixed mechanisms, 2.30–2.32 Modification and refinement, 3.136 Modified austempering, 13.83 Modulated-beam mass spectrometry (MBMS), 18.44 Modulated structure, 6.27, 6.28 Modulus of elasticity, 4.3, 4.5 Moire fringes, 10.72 Molding pattern design, 12.36 Molecular beam epitaxy (MBE), 18.43, 18.44 definition, 18.43 Molecular layer (ML), epitaxy (MLE), 18.47, 18.105 Monolithic metallic structures, 18.77, 18.78 Molten salts (see Salt baths) Monotectic solidification, 3.68–3.72 directional, 3.69, 3.71, 3.72 types of, 3.68, 3.69

Morphological stability: of constitutional supercooling, 3.36–3.38 of Mullins and Sekerka (MS) instability theory, 3.38–3.41 MOSFET, 2.75, 2.76 m phase, 10.81 Multicomponent boriding, 16.134, 16.135 Multiple quantum well optical modulator, 18.56 Multiquantum wells (MQW) structures, 18.49, 18.99 III–V quantum wells, 18.48

Nabarro–Herring creep, 15.87 Nanocrystalline materials, 15.93 superplasticity, 15.95 Natural aging, 6.38 Necklace microstructure, 15.56 Nernst-Einstein equation, 2.73, 2.93 Network cementite, 12.5 Neutral baths, medium temperature, 13.54, 13.56 Neutron irradiation, 2.104 New solidification processes (see individual process) Newtonian viscous flow, 15.87 Nickel-base superalloys, 6.51, 6.52, 6.63 applications, 6.52 crystal structure, 6.51, 6.52 precipitate morphology, 6.52 Nickel equivalent, 10.63, 10.64 Niobium carbide, 10.73, 15.7, 15.8 Nitinol, 8.52 Nitrate–nitrite salts, 13.52–13.54 Nitridable steels, nitriding effect, 16.83 Nitridation, 18.59, 18.63–18.65, 18.70 Nitride coatings, 18.23, 18.24, 18.98, 18.105 Nitride films, 18.56 group III–V and III nitride materials, 18.56, 18.82, 18.98 Nitriding, 16.81–16.102 advantages, 16.81, 16.83, 16.84 compound and diffusion zones, 16.81 disadvantages, 16.84, 16.85 iron–nitrogen phase diagram, 16.82 process, 16.85 rate of, 16.83 (See also specific types) Nitrocarburizing, 16.102–16.111 austenitic, 16.111 compound layer, 16.102 fluidized bed, 16.110, 16.111

INDEX

advantages, 16.110, 16.111 application, 16.111 gas: application, 16.102, 16.103 with post oxidation, 16.105 processing steps, 16.102 proprietary methods, 16.103 nitrotech, 16.103, 16.104 oxynitrocarburizing, 16.105 plasma, 16.110 salt bath, 16.105–16.110 proprietary methods, 16.105 oxynit process, 16.109, 16.110 QPQ method, 16.105, 16.107, 16.109 Nitrogen: atomic radii, 1.12 diffusion, nitriding, 16.81 solubility in a- and g -irons, 1.12, 1.14 Nodular iron (See Ductile iron) Nondestructive methods, 17.27, 17.28 Nonequilibrium defect concentrations, 2.97, 2.98 Nonferrous bainite, 9.12–9.14 Nonferrous martensite, 8.17, 8.36–8.39, 8.46 Nonferrous pearlite, 7.4, 7.14, 7.15 Normalizing: of compacted graphite iron, 12.34 of ductile iron, 12.34 of gray iron, 12.30, 12.32 of steel, 12.23–12.26 vs. air hardening, 12.23 double, 12.25, 12.26 effect of hot working, 12.25 microstructures, 12.24, 12.25 procedure, 12.23 purpose of, 12.23 Notch sensitivity, 12.42, 12.55, 12.64 Nuclear magnetic resonance (NMR) spectroscopy, 14.2–14.4 Nucleation and growth process, 6.15–6.19, 7.9–7.11 heterogeneous, 6.39 homogeneous, 6.39 Nucleation in solids, 6.1 activation free energy barrier, 6.2, 6.3, 6.21, 6.38, 6.44, 6.48, 7.40 atomic sites per unit volume, 6.7, 6.8, 6.11 Becker–Doring theory, 6.1 classical homogeneous, 6.2–6.7 coherent, on dislocations, 6.14, 6.15 coherent, with sharp interfaces, 6.4 coherent nucleus, 6.4

I.29

critical free energy change, 6.4, 6.7, 6.8, 6.10–6.12 critical nucleus, 6.3, 6.4 incoherent, on dislocations, 6.13, 6.14 nucleation rate, 6.7, 6.8 nucleus shape, 6.8 strain energy, 6.4–6.7 vacancy effect, 6.7 at container wall, 6.9, 6.10 at dislocations, 6.12–6.15 driving force for, 6.3 on grain boundaries, edges, and corners, 6.10–6.12, 7.11, 7.12 sympathetic, 9.21 vacancy effect, 6.7 Volmer–Weber theory, 6.1 Nucleation rate, 6.7, 6.8 classical homogeneous, 6.7, 6.8 for grain boundary, edge, and corner, 6.11–6.13 steady state, 6.8 transient, 6.8 Nucleation sites, 6.8–6.15 Nucleation theory: classical homogeneous, 6.2–6.7 nonclassical homogeneous, 6.19–6.22 Cahn–Hilliard theory, 6.19–6.21 continuum, 6.19–6.21 sharp interface model, 6.21

Ohmic barrier, 2.76 Oil quenching: accelerators, 13.21 antioxidant, 13.12 characteristics, 13.11 conventional, features, 13.12 demerits, 13.16 inhibitors, 13.21 quenching power, 13.15, 13.16, 13.18–13.21 water-oil emulsion, 13.16 (See also Quenching media) Oils, 13.11–13.27 ash content, 13.27 classification, 13.11, 13.12 cooling curve, a typical, 13.15 degradation of quenching, 13.23, 13.24 dragout loss, 13.21 emulsion of soluble, 13.16 fast, 13.12 flash and fire points, 13.12, 13.24 hot quenching (or martempering), 13.12 infrared analysis, 13.27, 13.28

I.30

INDEX

Oils (Cont.): laundering, 13.21 mineral, 13.12 neutralization number, of quenching, 13.26 oxidation, 13.27 precipitation number, 13.26, 13.27 saponification number of quenching, 13.26 slow, 13.12 temperature, 13.21 vegetable, 13.16 viscosity, 13.24 water contamination, 13.24, 13.26 Omega (w ) transformation, 8.81–8.83 Optical coatings, 18.15, 18.23, 18.28, 18.29, 18.31, 18.35, 18.41, 18.72, 18.83, 18.86, 18.87 Optical materials, 18.10, 18.15, 18.72, 18.83, 18.86, 18.87 Optoelectronic devices, 18.57, 18.58, 18.82–18.86, 18.99 Orange peel appearance, 5.47 Order: –disorder transformation, 8.46, 8.47 phase, 6.51 strengthening, 6.65–6.67 structure, 8.46, 8.50, 10.22, 10.76–10.78 Ordered alloys, 4.33 Ordered precipitates, 6.67 Organometallic compounds, 18.98 Organometallic vapor phase epitaxy (OMVPE) (see Metalorganic CVD or OMVPE or MOVPE) Orientation distribution function (ODF), 4.45, 5.38 Orowan: mechanism, 6.57, 6.58 strengthening, 6.62–6.64 Oswald ripening, 5.47 Oval defects, 18.48, 18.100, 18.102 Overheating, 17.1, 17.2–17.9 adverse effect, 17.2 factors affecting, 17.8 fractured surface appearance, 17.8 mechanism, 17.2 methods of detecting, 17.2 fracture testing, 17.2–17.5 metallography, 17.2, 17.5, 17.6 prevention of, 17.8, 17.9 reclamation, 17.9 Oxidation, 18.59, 18.61, 18.62, 18.70, 19.35 Oxide coatings, 18.18, 18.19, 19.41, 19.42, 19.45 Oxygen probe method, 16.64, 16.65

Pack carburizing, 16.45–16.49 advantages, 16.45, 16.47 applications, 16.48 box, 16.45, 16.46 carbon potential and case depth, 16.48 carbonates, as energizer, 16.48 carburizing compounds, 16.47, 16.48 carburizing reactions, 16.47 containers, 16.48 direct quenching, 16.45 disadvantages, 16.47 effect of carburizing time on C gradient and case depth, 16.45 furnace, 16.48 procedure, 16.45 temperature, 16.45, 16.46 Paraequilibrium transformation, 9.20, 9.23 Paramagnetic iron, 10.22 PAS, 2.38, 2.39 Patenting (see Modified austempering) Pearlite, 1.4, 7.1–7.15 colony size, 7.1 degenerate, 7.4 divorced, 7.4 effect of alloying elements, 7.7, 7.8 formation, 7.1 growth rate measurement, 7.13 forced velocity-growth method, 7.14 hot stage technique, 7.14 isothermal reaction technique, 7.13, 7.14 interlamellar spacing, 7.9 kinetics, 7.9 Avrami equation, 7.11, 7.13 Cahn-Hagel equation, 7.12 Johnson–Mehl equation, 7.11 mechanism of reaction, 7.2 branching mechanism, 7.2, 7.4, 7.5 cooperative growth, 7.4 ledgewise cooperative (shared growth), 7.4 repeated nucleation, 7.2 nodules, 4.1, 7.10, 7.12 nonferrous, 7.4, 7.14, 7.15 nucleation rate, 7.9, 7.10 temperature dependence, 7.9, 7.10 orientation relationships, 7.14 structure, degenerate, 7.4 Pearlite growth, 7.5 alloy effect, 7.7, 7.8 boundary diffusion controlled, 7.8 critical interlamellar spacing, 7.6, 7.7 curves in two steel grades, 7.7, 7.8

INDEX

driving force for, 7.6, 7.7 interphase boundary energy, 7.6, 7.7 maximum growth rate criterion, 7.7 rate, measurement of, 7.13, 7.14 theories, 7.5–7.7 undercooling, 7.6, 7.7 Zener–Hillert model, 7.6 Zener’s hypothesis, 7.6 Pearlitic reaction, mechanism of, 7.2 branching, 7.2 ledgewise cooperative (or shared) growth, 7.4 multiple nucleation events, 7.2 repeated nucleation, 7.2 sidewise nucleation and edgewise growth, 7.2 time sequential branching, 7.4, 7.5 Pearlitic steels, DSA effect, 4.33 Peclet number, 3.45–3.49, 7.32 Peierls stress, 4.13, 14.59 Percent reduction, 4.6 Peritectic reaction, 1.4, 3.73–3.75 temperature, 1.4 Peritectic solidification, 3.72–3.78 direct, 3.75, 3.76 growth kinetics, 3.76 layered structure formation, 3.76–3.78 transformation, 3.75 Peritectoid reaction, 10.22 Permeability, 2.10, 2.11, 10.22 Permeation barriers, 18.20 Phase balance, 10.88 Phase diagram: of Al-Ag, 6.49 of Al-Cu, 6.29 of Al-Cu (partial, with metastable solvus curves), 6.43 equilibrium, 1.3–1.5, 6.19 of Fe-B, 16.115 of Fe-C, 1.4–1.6 of Fe-Cr, 10.21 of Fe-18Cr-4Ni-C, 10.24 of Fe-18Cr-8Ni-C, 10.24 of Fe-18Cr-Ni-N, 10.75 of Fe-Fe3C and Fe-C system containing 2.5% Si (estimated), 10.56 of Fe-Fe3C-Si-C, 1.4, 1.5, 1.56 of Fe-N, 16.82 of Fe-Ni, 10.23 metastable, 1.6, 6.19 of Ni-Cr, 10.23 schematic, showing massive transformation, 7.23

I.31

of Ti-6Al-4V (schematic), 15.70 Phosphorus valved cracker cell, 18.49, 18.51 advantages, 18.49 Photo CVD: advantages and disadvantages, 18.95, 18.96 photolysis (or photo laser) mechanism, 18.97 photothermal (or thermal) mechanism, 18.97 Photodiode detectors (structures), 18.84 Photodiode masks, 18.36 Photodiodes, 18.56, 18.82, 18.83 Photolithographic mask coating, 18.94 Photolysis, 18.59, 18.69 (or photolaser) mechanism, 18.97 Physical vapor deposition, 18.21–18.56 advantages and disadvantages, 18.21–18.23 applications, 18.23–18.25 characteristics of hard coatings, 18.22 classification, 18.21 Pinning force, particle, 5.43 Plane front solidification: lateral segregation, 3.36 morphological stability of, 3.36 constitutional supercooling, 3.36–3.38 Mullins-Sekerka instability theory, 3.38–3.41 solute redistribution during unidirectional, 3.26 complete liquid mixing, no solid diffusion, 3.29, 3.31 equilibrium solidification, 3.27, 3.28 limited liquid diffusion, no convection, 3.30–3.32 partial liquid mixing, with convection, 3.32, 3.33 zone refining, 3.33–3.36 Plasma carbonitriding, 16.80 Plasma carburizing, 16.71–16.74 advantages, 16.72–16.74 applications, 16.74 carburize–diffuse periods, 16.71, 16.72 disadvantages, 16.74 gas mixture, 16.72 procedure, 16.71, 16.72 pulsed, 16.74, 16.75 Plasma immersion ion implantation (PIII), 18.5 Plasma nitriding, 16.93–16.101 advantages, 16.98–16.100

I.32

INDEX

Plasma nitriding (Cont.): applications, 16.100 equipment, 16.94–16.96 limitations, 16.100 process, 16.94–16.98 pulse, 16.100, 16.101 advantages, 16.101 structural zones, 16.98 temperature–time curves, 16.97 Plasma nitrocarburizing, 16.110 procedure, 16.110 surface, layer structure, 16.110 Plasma spray (PS) coating: advantages and drawbacks, 19.12 application, 19.11 classification, 19.12 mechanical properties, 19.36–19.38 process, 19.11 Plasma transferred-arc (PTA) spraying (or hardfacing), 19.16, 19.17 advantages and disadvantages, 19.16 application, 19.16, 19.17 characteristics, 19.16 process, 19.14, 19.16 Plasmas, characteristics, 18.93 Plastic behavior, 4.1 Plastic deformation, 4.1, 4.27, 4.28 effect of grain size, 4.1, 4.28 effect of SFE, 4.1 irreversible, 8.46 LDR, 4.44 ODF, 4.45 pole figures, 4.47–4.49 Plastic instability, 4.10–4.12 Plastic strain, 4.16 rate, 4.16, 4.27 ratio, 4.44 variation of rm with grain size, 4.45, 4.46 Plate martensite (see Martensite, ferrous, plate) Plateau behavior, 7.36 PLC effect (or bands) (see Dynamic strain aging) Pole figures, 4.47–4.49 Polishing, 10.7, 10.8, 19.24, 19.31 Polyacrylate solutions, 13.33, 13.35–13.37 applications, 13.36, 13.37 characteristic features, 13.33, 13.35, 13.36 Polyalkylene glycol solutions, 13.31 Polyethyl oxazoline, 13.37 Polygonization, 5.6–5.9 Polymer: deposition on, 18.10, 18.17, 18.20

thermal sprayed, 19.49 Polymer quenchants, 13.27 advantages, 13.27 bath maintenance, 13.39–13.48 selection of, 13.37 stability, 13.37–13.40 chemical, 13.39 mechanodegradation, 13.37 polymer dragout, 13.38 thermal/oxidative, 13.38 synthesis of various, 13.28, 13.29 types of, 13.27 Polyvinyl alcohol, 13.29 concentration control, 13.30 cooling characteristics, 13.30 disadvantages, 13.30 Polyvinylpyrrolidone, 13.31–13.33 Positron annihilation, 2.37, 2.38 spectroscopy, 2.39 Post-spray treatment, 19.24–19.33 annealing treatment, 19.25 finishing: grinding, 19.31, 19.32 polishing and lapping, 19.31 hot isostatic pressing, 19.25, 19.28, 19.29 impregnation, (or sealing), 19.30, 19.31 inorganic sealers, 19.30 vacuum impregnation, 19.30, 19.31 other methods, 19.31 process, 19.24 Powders: CVD-coated, 18.78, 18.79 ultrafine, 18.78 Power density in induction hardening, 16.12–16.14 Power difference, 5.1–5.3 Precipitation hardening mechanism, 6.57–6.70 alloys exhibiting, 6.62, 6.63, 6.65, 6.67–6.70 chemical strengthening, 6.57, 6.64, 6.65 coherency strain hardening, 6.57, 6.61, 6.62 CRSS, 6.57–6.59, 6.62–6.67 effect of finite particle size, 6.60, 6.61 effective, dislocation line tension, 6.63, 6.68 hardening by spinodal decomposition, 6.57, 6.69, 6.70 interparticle spacing, 6.59, 6.60 maximum interaction force, 6.62, 6.64, 6.66 modulus strengthening, 6.57, 6.67 nondeformable particles, 6.63

INDEX

order strengthening, 6.57, 6.65–6.67 Orowan strengthening, 6.57–6.59, 6.62–6.64 stacking fault strengthening, 6.57, 6.68, 6.69 Precipitation hardening (PH) stainless steels, 6.53–6.57 applications, 6.35, 6.36, 6.56, 6.57 austenitic, 6.53, 6.55 classification, 6.53 heat treatment step, 6.53–6.55 martensitic, 6.53, 6.54 mechanical properties, 6.35, 6.36 precipitation sequence, 6.55, 6.56 semiaustenitic, 6.53, 6.54 sensitization, 6.55 Precipitation hardening systems: Ag-Al alloys, 6.48–6.50 metastable and equilibrium phases, 6.48 phase diagram, 6.48 Al-Cu alloys: effect of aging temperature, 6.42–6.44 effect of trace additions, 6.44 free-energy–composition diagram, 6.41–6.43 GP zones, 6.38–6.41 hardness vs. aging time, 6.30, 6.37 heat treatment, basic steps, 6.37, 6.38 mechanical properties, 6.32 metastable and equilibrium phases, 6.38–6.41 metastable solvus curves, 6.42–6.44 precipitation sequence, 6.30, 6.31, 6.38, 6.41–6.44 spinodal decomposition, 6.38 stability and solubility of phases, 6.41–6.44 Al-Cu-Mg alloys, 6.44, 6.45 GP zones, 6.44, 6.45 precipitation sequence, 6.44 Al-Li alloys, 6.46–6.48 metastable and equilibrium phases, 6.47, 6.48 Al-Mg-Si alloys, 6.45, 6.46 applications, 6.46 GP zones, 6.45 precipitation sequence, 6.45, 6.46 Al-Zn-Mg(-Cu) alloys, 6.46 GP zones and precipitation sequence, 6.46 Co-base alloys, 6.53 applications, 6.53 precipitation sequence, 6.53

I.33

Cu–base alloys, 6.51 mechanical properties, 6.32, 6.34 precipitation sequence, 6.31, 6.51 general characteristics, 6.30 mechanical properties and applications of some, 6.30, 6.32–6.34 necessary conditions for, 6.28 Ni–base systems, 6.51, 6.52 applications, 6.52 crystal structure, 6.51, 6.52 precipitate morphology, 6.52 precipitation sequence, 6.31–6.34, 6.51, 6.52 overaging, 6.52 peak hardness, 6.30 precipitation nucleation sequences, 6.30–6.32 precipitation sequence in, 6.30, 6.31 single- and two-stage hardening, 6.30, 6.37 Preferred orientation, 4.44–4.49 (See also specific type, texture) Press quenching, 13.65, 17.47 Pressure effect on diffusion, 2.42, 2.43, 2.48–2.50 Pressure gas quenching, 13.58–13.62 advantages and disadvantages, 13.59–13.61 applications, 13.58, 13.61, 13.62 factors required for design of, 13.58, 13.59 furnaces, 13.58, 13.59 Primary recrystallization (see Recrystallization) Primary Si refinement of hypereutectic Al-Si alloys, effect on properties, 3.135, 3.136 modification and, 3.136 with other elements, 3.136 Prismatic punching mechanism, 10.72 Process annealing, 12.12 Proeutectoid ferrite, 7.21–7.41 allotriomorphic, 7.22–7.27 effect of alloying elements, 7.26 growth kinetics, 7.22–7.26 growth rate models, 7.26 interface characteristics, 7.26 morphology, 7.21, 7.22 orientation relationship, 7.26 relief effects, 7.27 thickening kinetics, 7.22, 7.24, 7.25 idiomorphic (see Idiomorphic ferrite) massive [see Massive (ferrite) transformation] Widmanstätten plates, 7.27–7.35

I.34 Proeutectoid ferrite (Cont.): formation, mechanism, 7.29, 7.30 growth kinetics, 7.30–7.34 lengthening, 7.30–7.35 thickening, 7.34, 7.35 intragranular plates, 7.29 primary saw teeth, 7.29 primary side plates, 7.27 secondary saw teeth, 7.29 secondary side plates, 7.28, 7.29 Proof stress (see Yield strength) Propane, 16.65–16.68, 16.72, 18.110, 18.111 Proportional limit, 4.13 Pseudomorphic high-electron mobility transistors (PHEMTs), 18.48 Purging, in nitriding, 16.88, 16.89 Pyrolysis, 18.44, 18.59–18.61, 18.69, 18.70, 18.78

Quadrapole mass analyzer (QMA), 18.54 Quality descriptors, 1.26–1.28 Quantum-dot lasers, 18.48 Quantum effect devices, 18.105 Quantum-well, 18.48 Quantum-wire, 18.48 Quasi superplasticity, 15.99 Quench aging, 4.30 Quench cracking, 13.94–13.96, 17.28 in borided steel, 17.33 in carburized alloy steel, 17.31, 17.32 in decarburized steel, 17.31 microcracking, 17.32, 17.33 Ms-Mf range, 13.95 in nitrided steels, 17.33 part defects, 17.29, 17.30 part design, 17.28, 17.29 potential reasons for, 13.94, 13.95, 17.28 prior steel structure, 13.94, 13.95 skin tempering, 17.33 steel grades, improper selection, 13.95, 17.29 thumbnail checks, 17.30 tip cracking, 17.33 Quench hardening of iron castings, 17.67, 17.68 equation for A1, 13.68 method, 13.67, 13.68 Quench hardening of steel, 13.65, 13.67 advantages and limitations of water quenching, 13.65, 13.66 factors responsible for successful, 13.67 method, 13.65

INDEX

superimposed on TTT diagram, 13.65–13.67 Quench severity, 13.19–13.21 Quench tempering, 8.79 Quenched-in electrical resistivity measurement, 2.39 Quenching fixture, 13.74, 17.45, 17.46 Quenching media: brine and caustic solutions, 13.10, 13.11 advantages and disadvantages, 13.10, 13.11 cold die, 13.63, 13.64 fluidized bed quenching, 13.62 lead bath, 13.57, 13.58 main requirements of, 13.9 method of evaluation of cooling power, 13.6, 13.7, 13.9 oils, 13.11–13.27 additives, 13.21 classification, 13.11, 13.12 conventional (or straight), 13.11, 13.12 demerits of, 13.16 dragout loss, 13.21 fast, 13.12 hot quench (or martempering), 13.12, 13.15, 13.16 inhibitors, oxidation, 13.21 laundering, 13.21 maintenance of, 13.23, 13.24, 13.26, 13.27 analytical method, 13.27 ash content, 13.27 flash and fire points, 13.24 neutralization number, (TAN) 13.26 precipitation number or sludge, 13.26, 13.27 saponification number, 13.26 viscosity, 13.24 water contamination, 13.24, 13.26 measurement of heat-removal properties, 13.16 quench loads, 13.21, 13.22 quenching power: Castrol/Renault hardening power and Castrol Index, 13.21 hot wire test, 13.16, 13.18 IVF hardening power, 13.19 main factors responsible for, 13.21 other methods, 13.18, 13.19 selection of, 13.23 temperature, 13.21 vegetable-oil based, 13.16 water-oil emulsion, 13.16

INDEX

polymer solutions, 13.27–13.48 advantages, 13.27 bath maintenance of, 13.39, 13.40 biological degradation, 13.46 changes in, with use, 13.46–13.48 corrosion protection, 13.46 gaseous contamination, 13.48 membrane separation, 13.44 oil contamination, 13.46 polymer concentration, 13.40 polymer degradation, 13.40, 13.46–13.48 refractive index, 13.40–13.42 salt content, 13.42, 13.44 thermal separation and PH measurement, 13.44–13.46 types, 13.27–13.29 viscosity, 13.42 press quenching, 13.65, 17.47 pressure gas quenching, 13.58–13.62 applications, 13.49–13.51 salt baths, 13.48–13.58 advantages and disadvantages, 13.49–13.51 high temperature, 13.56 low temperature, 13.52–13.54 medium temperature, 13.54–13.56 maintenance of, 13.57 safety precautions of, 13.56, 13.57 water, 13.9, 13.10, 13.65, 13.66 advantages and disadvantages, 13.9, 13.10 rapid, 13.65, 13.66

Radiation damage, 5.4–5.6, 18.93 Radiation effects and diffusion, 2.103–2.112 defects production, 2.104 bubbles, 2.106–2.108 cascade defects, 2.105, 2.106 defect clusters, 2.106 voids, 2.106 inverse Kirkendall effects, 2.110, 2.111 Radiation-enhanced diffusion, 2.109 mixing, 18.12 Radiation-induced phase transformation, 2.111 experimental results, 2.111, 2.112 segregation and precipitation, 2.109, 2.110 void swelling, 2.108, 2.109 Rail-gun technology, 19.23, 19.24 Rail steels, 7.46 bainitic, 7.53

I.35

grades and specifications, 7.51, 7.52 hardness, 7.47 head hardening, 7.48, 7.49 interlamellar spacing, 7.47–7.49 rail corrugation, 7.54, 7.55 residual stress, grinding, and catastrophic failure, 7.55–7.57 rail wear, 7.53, 7.54 rolling contact fatigue, 7.47 tensile strength, 7.51 wear resistance, 7.47 Random walk theory, 2.32, 2.33 Raleigh instabilities, 5.32 Rapid solidification processes, 3.141, 15.55, 15.65 conduction processes, 3.143–3.146 chill block melt spinning, 3.143 crucible melt extraction, 3.145, 3.146 crucible melt overflow, 3.146 free-flight melt spinning, 3.143 free-jet melt spinning (or jet casting), 3.144 planar flow casting, 3.145 convection processes, 3.146–3.149 plasma spray deposition, 3.148, 3.149 spray deposition, 3.146–3.148 ultrasonic gas atomization, 3.149 Rare earth additions, 1.60, 1.64, 1.68 Reactive arc PVD, 18.31 Reactive plasma spraying, 19.16 Read–Shockley equation, 5.8 Recovery, 5.3 activation energies, 5.11, 5.12 anneal, 5.6 dislocation cell walls, 5.6 dynamic (see Dynamic recovery) kinetics of, 5.10–5.12 mechanism, 5.6 property changes, 5.5, 5.6 stress relief annealing, 5.5 Recrystallization, 5.4–5.26 activation energy, 5.26 in Al alloys, 5.23, 5.24 annealing texture, 5.33–5.39 critical strain, 5.26, 5.32 dynamic (see Dynamic recrystallization) effect of deformation, 5.30 effect of second phase particles, 5.42–5.45 growth rate, 5.15–5.19, 5.21–5.26 kinetics of, 5.15–5.22 laws of, 5.32, 5.33 main features of, 5.15–5.18

I.36

INDEX

Recrystallization (Cont.): mechanisms: JMAK model, 5.15–5.19 microstructural path methodology, 5.19–5.22 orientation aspect, 5.32, 5.33 property changes during, 5.12, 5.13 secondary (see Secondary recrystallization) temperature, 5.25, 5.29–5.32 deformation mode, 5.30 deformation temperature, 5.30 heating rate, 5.30 initial grain size, 5.29 pores and bubbles, 5.30–5.32 prior strain, 5.30 purity, 5.29 texture, 5.33, 5.35, 5.39 application, 5.39 correlation of microstructure, 5.35 effect of coiling temperature and heating rate, 5.35 origins of, 5.33, 5.35 of two-phase alloys, 5.27, 5.28 particle stimulated nucleation, 5.27, 5.28 segregation and precipitation, 5.28 Recrystallization annealing, 12.12 Recrystallized grain size, 5.25, 5.26 Recrystallized nuclei, origin of, 5.13–5.15 multiple twinning mechanism, 5.15 preexisting or preformed nuclei, 5.15 strain-induced boundary migration, 5.14, 5.15 subgrain coalescence via evaporation, 5.13 subgrain coalescence via pipe diffusion, 5.13, 5.14 Reduction of area, 4.6, 4.7 (See also Ductility) Reduction reaction, 18.59–18.61, 18.69, 18.70 Reflection high energy electron diffraction (RHEED), 18.44, 18.51, 18.54, 18.102 Reflectometry, 18.102 Refractive index, 13.40, 13.44 Refrigeration, 10.17, 10.18 Relaxation methods, 2.24–2.26 Residual stress, 17.9–17.28 compressive, 12.10, 12.11, 17.14 control, 17.51 design of parts, 17.52, 17.53 development of, 17.14 effect of, 17.10 Koistinen’s theory, 17.21 measurement, 17.26

destructive method, 17.26, 17.27 nondestructive methods, 17.27 Berkhausen method, 17.28 magnetic method, 17.28 ultrasonic method, 17.28 x-ray diffraction method, 17.27 in nonferrous alloys, 17.25 in other processing, 17.23–17.25 patterns, 17.15–17.19 after surface hardening, 17.20–17.25 due to thermal and transformational volume changes, 17.18, 17.19 due to thermal contraction, 17.15–17.17 tensile, 17.11, 17.13, 17.14 TERSA, 17.28 thermal stresses, 17.15 types of, 17.10 volume changes, 17.15 Resistance evaporation, 18.25 Restraining fixture, 17.45 Retained austenite, 8.13–8.15, 10.14–10.19, 13.73, 15.40, 15.51, 15.52, 17.21, 17.37, 17.50 beneficial effect, 10.15, 10.16, 15.40, 15.43 decomposition of, 14.13–14.16 deleterious effect, 10.15 determination of, 10.19 elimination of, 10.17 factors affecting formation of, 10.16, 10.17 morphology of, 15.43, 15.47, 15.48, 15.50 stability of, 15.43, 15.44 stabilization, 8.13–8.15 mechanical, 8.15 thermal, 8.13–8.15 RF magnetrons, 18.39 RF plasma reactor, 18.93 RF sputter deposition advantages and disadvantages, 18.38, 18.39 applications, 18.39 procedure, 18.37 Rock candy fracture, 14.82 Rolling die quenching, 17.47 Rolling texture, 4.48, 4.49 Rotating disk-reactor (RDR), 18.100, 18.101 Rotating substrate reactors, 18.107, 18.108 Rough threading, 19.4 Rutherford backscattering (RBS) analysis, 18.8, 18.11

Saddle displacement, 2.73 SAE specifications (see AISI–SAE designation)

INDEX

Salt bath carbonitriding (See Cyaniding) Salt bath carburizing, 16.49–16.54 advantages and limitations, 16.51, 16.52 applications, 16.53 barium chloride activated, 16.50, 16.51 bath composition, 16.49, 16.50 carburizing agent, 16.49 carburizing potential, 16.50 case depth and carbon concentration, 16.52 cyanide based baths, 16.50, 16.51 effect of temperature and time, 16.52, 16.53 furnaces, 16.52 nonactivated, 16.49, 16.50 nonyanide baths, 16.53, 16.54 advantages and disadvantages, 16.54 variations of, 16.54 silicate-activated, 16.51 Salt bath nitriding, 16.85–16.88 special processes, 16.86–16.88 Salt bath nitrocarburizing, 16.105–16.110 oxynit and sursulf process, 16.107, 16.109, 16.110 Tufftride TF1/AB1 and QPQ treatment, 16.105, 16.107, 16.109 Salt baths, 13.48–13.58 applications, 13.49–13.51 cyanide and noncyanide bearing, 13.56, 16.50, 16.51, 16.53, 16.54 guidelines for safe operations, 13.56, 13.57 high temperature, 13.56 low temperature, 13.52–13.54 maintenance of, 13.57 medium temperature neutral, 13.54–13.56 saturated, 13.86 Schaeffler diagram, 10.63, 10.64 Schockley partial dislocations, 10.73 Schottky barrier, 2.76, 18.48 Schottky defect, 2.94 SDLE, 7.26 Sealing (see Impregnation) Secant modulus method, 12.38 (Secondary) graphitization of steel, 12.21–12.23 Secondary hardening, 14.26–14.29 embrittlement, 14.81 peaks, 14.26, 14.27 stable carbides, 14.26 Secondary hardening steels: chromium bearing, 14.37–14.40 Cr-Mo-V bearing, 14.40 high Co-Ni steels, 14.40

I.37

high speed steels, 14.28–14.36 advantages, 1.47, 14.28 applications, 14.28–14.30 characteristic properties, 14.28 classification, 14.28 heat treatment of, 14.32–14.37 type M2 steel, 14.35–14.37 type T1 steel, 14.32–14.35 martensitic stainless steels, 14.43, 14.44 nitrogen bearing steels, 14.41, 14.43 Secondary ion mass spectroscopy (SIMS), 18.44, 18.47 Secondary recrystallization, 5.50–5.55, 15.8, 15.9 applications, 5.52–5.55 coarsening temperature, 5.51 factors promoting, 5.50, 5.51 grain-oriented silicon steel sheets, 5.54 nucleation and growth process, 5.51, 5.52 requirements of, 5.50, 5.51 and sintering, 5.55 surface controlled, 5.55 texture, 5.52–5.55 Selected area deposition (SAD) or epitaxy (SAE), 18.56, 18.97, 18.106 Selective hardening, 16.15 Self interstitials, 2.97 Semiconductor: group II–VI materials, 18.48, 18.54, 18.56, 18.81–18.83, 18.97, 18.98, 18.105 group III–V materials, 18.48, 18.55, 18.56, 18.81, 18.82, 18.92, 18.97, 18.98, 18.100, 18.105 group IV materials, 18.48 Semiconductor diamond, 18.81 Semiconductor metallization, 18.35 Semisolid metal forming process, 3.157–3.159 semisolid metal casting, 3.158, 3.159 parts and properties, 3.159 raw materials, 3.159 Sensitization (See Austenitic stainless steels, sensitization) Sensors, 8.74–8.77, 18.3 Severity of quench, 13.16, 13.19–13.21, 13.50, 17.15, 17.20 Shallow junction diffusion technology, 18.8 Shape factor, 3.12, 3.13 Shape memory alloys, 8.47–8.59 applications in: aerospace systems, 8.70 automotive industry, 8.74 biomedical: dental, 8.71, 8.72

I.38

INDEX

Shape memory alloys (Cont.): medical instruments and tools, 8.72, 8.73 orthopedic, 8.72 stents, 8.71 consumer products, 8.74 conversion of thermal energy to mechanical energy, 8.71 coupling and electrical connectors, 8.68, 8.69 electric power systems, 8.69, 8.70 electrical actuators, 8.70 electrical seals and packaging, 8.69 high force devices, 8.70 mechanical dampers, 8.73 as microactuators, 8.71 shape recovery, 8.68 smart materials, 8.74, 8.75 as structural dampers, 8.74 thermal sensors and actuators, 8.70 valves, 8.69 Au-Cd and Au-Zn alloys, 8.55 Co-based alloys, 8.57, 8.58 Co-Cr-bearing alloys, 8.57 Co-Ni-bearing alloys, 8.57 strain-induced transformation, 8.57, 8.58 Cu-based alloys, 8.54, 8.55 CuAlMn alloys, 8.54 CuSn alloy, 8.55 CuZnAl and CuAlNi alloys, 8.54 CuZnAlMnZr alloy, 8.54 CuZr alloy, 8.55 Fe-based alloys, 8.55–8.57 Fe-Mn-bearing alloys, 8.55, 8.56 Fe-Cr- and Fe-Ni-bearing alloys, 8.56, 8.57 fine scale deformation, 8.48 improvements of, 8.58 constitutive relations of, 8.58, 8.59 grain-refined, 8.58 high-temperature, 8.58 internal faulting and twinning, 8.48 limitations, 8.48 Mn-Cu based alloys, 8.57 Ni-based alloys, 8.57 Heusler structure, 8.57 stacking sequence, 8.48, 8.49 Ti-Ni-based alloys, 8.48–8.50, 8.52–8.54 advantages, 8.48 disadvantages, 8.52 near equiatomic composition, 8.52 pseudoelastic b Ti-alloy, 8.54

two-stage martensitic transformation, 8.54 Shape memory effect, 8.48, 8.59, 8.61–8.63 alloys, 8.50, 8.51 two-way: applications, 8.63 methods, 8.61–8.63 Shatter cracks, flakes, and fisheyes, 14.100, 14.101 Sheaves, 9.3, 9.4 Shockley partial dislocations, 10.73 Shot peening, 17.10–17.13 Shrinkage stress, 19.36 Shrinkage voids, 1.7, 3.155 Shrouded plasma spraying (SPS), 19.15, 19.16 Shuffles, 4.35, 4.43 Sigma phase, 10.78, 10.79 Single crystal growth from melt: Bridgman method, 13.115 Chalmer method, 3.117 Czochralski pulling method, 3.117–3.119 floating zone melting method, 3.119, 3.120 gradient freeze method, 3.115, 3.116 Skin tempering, 17.33 Slab reheating temperature, 15.7 Slip, 4.12, 4.13 Soft spot, 13.10, 13.27, 17.30 Solar cells, 18.43, 18.48, 18.49, 18.55, 18.82, 18.83, 18.85, 18.94, 18.98, 18.99 Solid-metal embrittlement, 14.106 characteristics of, 14.106–14.108 delayed failure type, 14.108 occurrence of, 14.106, 14.107 Solid-oxide fuel cells (SOFCs), 19.43, 19.44 Solid source molecular beam epitaxy (SSMBE) application, 18.49, 18.50 growth kinetics of InAsP, 18.49 phosphorus valved cracker cell, 18.49, 18.51 ten-cell configuration/system, 18.48, 18.49 Solidification, 1.6, 1.55, 1.60, 3.1 activation energy, 3.10 heat transfer coefficient, 3.4 heat transfer in, 3.1 Chvorinov constant, 3.6 conducting molds, 3.2, 3.3 heat diffusivity, 3.6 insulated molds, 3.5–3.7

INDEX

limited by interface resistance, 3.3, 3.4 hot cracks, 10.62 iron castings, 1.6, 1.60 rate, white iron, 1.57 Solidification, fusion welds, 3.105–3.114 characteristics, 3.106, 3.107 weld cracking, 3.112–3.114 weld microstructure, 3.109–3.112 weld shape and macrostructure, 3.107, 3.109 Solidification, growth from melt: interface growth, 3.24, 3.26 heat and solute transport, 3.25, 3.26 interface conditions, 3.24, 3.25 interface kinetics, 3.17–3.23 continuous growth, 3.18, 3.19 growth by: propagation of twin plane, 3.23 SDG, 3.21, 3.22 2DN, 3.20, 3.21 lateral growth, 3.19, 3.20 unified theory, 3.22, 3.23 interface structure, 3.14–3.17, 3.23 Cahn’s theory, 3.16 other models, 3.16, 3.17 Solidification, nucleation: heterogeneous, 3.11 contact (or wetting) angle, 3.12, 3.13 free energy change during, 3.12 inoculants, 3.13 nucleation rate, 3.13, 3.14 supercooling, 3.13 homogeneous, 3.7–3.9 critical embryos, 3.10 nucleation rate, 3.9–3.11 critical size clusters, 3.10 metastable equilibrium concentration, 3.10 steady state, 3.10 Solidification, segregation: macro-, 3.78–3.86 banding, 3.82, 3.83 Boussinesq method, 3.79 centerline, 3.81 channel, 3.84–3.86 D’Arcy’s law, 3.79 gravity, 13.83, 13.84 inverse, 3.81, 3.82 Lutwig-Soret effect, 3.86 normal, 3.81, 3.82 Scheil equation, 13.79, 13.80 micro-, 3.88–3.93 cellular, 3.89, 3.90

I.39

dendritic, 3.90–3.92 grain boundary, 3.93 Scheil equation, 13.90 Solidification, segregation patterns in: continuous steel casting, 3.87, 3.88 axial segregation, 3.87, 3.88 steel ingots, 3.86, 3.87 types of macrosegregation, 3.86, 3.87 Solidification processes and cast structures, 3.93–3.114 continuous casting, 3.98–3.105 of nonferrous alloys, 3.101–3.105 direct chill (DC) casting, 3.103, 3.104 electromagnetic casting, 3.103–3.105 hot top casting, 3.102, 3.103 LHC casting, 3.103, 3.104 OCC process, 3.101, 3.102 of steel, 3.98–3.101 structure, 3.100, 3.111 ingot casting, 3.93–3.98 classification of, steel ingots, 3.93–3.95 ingot structures, 3.95–3.98 Solute drag: like effect (SDLE), 7.8, 7.23, 7.26 theory, 5.42 Spalling, 16.85, 16.141, 18.93 Spheroidize annealing, 12.6–12.11 advantages and disadvantages, 12.9, 12.10 mechanism, 12.10, 12.11 methods, 12.7, 12.8 microstructure, 12.8, 12.9 rate of reaction, 12.10 temperature calculation, 12.10 Spherulitic graphite (SG) iron (see ductile iron) Spinodal decomposition, 6.15–6.28, 6.38 applications, 6.28 binary phase diagram, 6.15, 6.16 coherent, 6.19, 6.20 composition profile, 6.18, 6.19 diffusion equation solution, 6.23, 6.24 free energy composition diagram, 6.16, 6.17 heat treatment steps, 6.16, 6.17 kinetics, 6.23, 6.25–6.27 line, 6.15, 6.17, 6.22 versus nucleation and growth process, 6.17–6.19 nucleation theory, 6.19–6.22 spinodal alloys, 6.27, 6.28, 15.77 structures, 6.27, 6.28 Splat morphology, 19.1, 19.33, 19.36 Spring wires, 7.60, 7.61

I.40

INDEX

Sputter deposition: advantages and disadvantages, 18.34, 18.35 applications, 18.35, 18.36 glow discharge, 18.36 magnetron, 18.39–14.41 advantages and disadvantages, 18.38–18.40 applications, 18.39 balanced and unbalanced configurations, 18.40, 18.41 closed field unbalanced magnetron (CFUMS), 18.41 principle, 18.37, 18.38 various configurations, 18.35, 18.39, 18.40 procedure, 18.34 RF, 18.37–18.39 triode, 18.37 Squeeze casting, 3.155–3.157 Stacking fault, 4.27, 6.49, 9.17 contrast, 6.49 energy, 4.19– 4.22, 4.47, 4.49, 5.6, 6.68, 15.14 strengthening, 6.68, 6.69 Stainless steels: austenitic, 10.19, 10.20 cast, 10.20 duplex, 10.19 ferritic, 10.19 machinability, 10.104–10.107 martensitic, 10.20 superaustenitic, 10.21 superduplex, 10.19 precipitation hardening, 6.53–6.57, 10.19 Static equilibrium conditions, 6.9, 6.10 Static processes, 15.20 metadynamic, 15.20–15.22 recovery, 15.20 recrystallization, 15.20, 15.21 Static recovery, 15.12 Static recrystallization, 15.18 Steelmaking practice, 1.20 Steels, 1.1, 1.6, 1.9–1.12 bearing, 1.41, 1.43 C and C-Mn, 1.28–1.33 Cr-Mo heat resistant, 1.43 classifications, 1.6, 1.26, 1.51 effect of alloying and other elements, 1.14–1.17, 1.20–1.25 high carbon, 1.33 HSLA, 1.45, 1.46 hypereutectoid, 1.6 hypoeutectoid, 1.6

low alloy, 1.34, 1.35 low alloy ultrahigh strength, 1.45 for low temperature and cryogenic service, 1.44, 1.45 maraging, 1.51 designations for, 1.51–1.53 medium-carbon, 1.32, 1.33 silicon, 1.43, 1.44 grain-oriented (or anisotropic), 1.43, 1.44 classification, 1.44 magnetic properties, 1.43 nonoriented (or isotropic), 1.43, 1.44 ultrahigh carbon, 1.33 ultrahigh strength, 1.35, 1.41 ultralow or extralow-carbon steels, 1.33 bake hardening (BH), 1.33 interstitial free (IF), 1.34 Sticking coefficient, 18.55 Stokes’ equation, 3.84 Stored energy release, 5.1–5.3 impurity effect on, 5.2, 5.3 methods, 5.1, 5.5 power difference, 5.1–5.3 Straightening, 16.16, 17.45, 17.49 Strain: critical, 5.26, 15.15, 15.17, 15.18 elastic, 4.13, 6.19 energy, 6.4–6.6, 6.10, 6.21, 6.24 engineering, 4.2, 4.3, 4.19 hardening (see Work hardening) hardening exponent, 4.6, 4.7, 4.18–4.21, 4.31 peak, 15.15, 15.16, 15.18 plastic, 4.13 rate, 4.27, 15.82 rate sensitivity, 4.7, 4.33, 15.83, 15.84 true, 4.3, 4.18, 15.83–15.85 uniform, 4.6, 15.52 zero gage length, 4.7 Strain aging, 4.28–4.33 dynamic, 4.28, 4.31–4.33 and strain rate, 4.33 static, 4.28–4.30 Strain rate sensitivity, 4.33, 15.83, 15.84 Stress: concentration, 12.37 critical resolved shear, 4.13 engineering, 4.3, 4.19 raisers, 4.9, 12.42, 12.55, 17.13 true, 4.3, 4.18 Stress corrosion cracking (SCC), 14.88, 15.58, 15.60

INDEX

Stress relieving: for austenitic stainless steels, 10.27, 10.47 of ductile iron, 12.34 of gray iron, 12.28, 12.30, 12.31 of silicon steels, 1.43, 1.44 of steel, 12.11 vibratory, 17.51 Stress-strain curve, 15.13 engineering, 4.3, 4.10 true, 4.3, 4.7, 4.10, 4.19 Structure-property relationships: in austenitic stainless steels, 10.82–10.86 in bainitic steels, 9.33–9.36 in dual phase steels, 15.51, 15.52 in duplex stainless steels, 10.96–10.98 in gray iron, 12.67 in HSLA steels, 15.34, 15.38. 15.39 in low-carbon ferrite–pearlite steels, 7.43, 7.44 in martensite, 8.75–8.81 in pearlite in ferrite-pearlite steels, 7.44–7.46 in tempered martensite and bainite, 14.59–14.63 Subboundaries (or tilt boundaries or polygon walls), 5.6 Subgrain: formation, 5.6 growth, 5.6 size and flow stress, 5.9, 5.10 Subgrain boundary coalescence, 5.13 atomistic mechanism, 5.13 and nucleation, 5.13, 5.14 and SIBM, 5.14 Substructure strengthening, 5.10 Subzero treatment, 10.17, 10.18 Sulfide stress cracking, 14.93, 14.94 Super D-Gun spray, 19.20 Superaustenitic stainless steels, 10.86–10.88 applications, 10.86–10.88 definition, 10.86 microbiologically influenced corrosion (MIC) resistance, 10.87 pitting resistance equivalent, 10.86 Supercarburizing, 16.111–16.114 Superconducting coatings, 18.31 Supercooling (see Undercooling) Superduplex stainless steels, 10.89, 10.96, 10.98 Superlattices, 18.41, 18.49, 18.91, 18.99, 18.105 Superplastic alloys, 15.79–15.82 Superplastic flow stress vs. strain rate, 15.85

I.41

Superplastic forming (SPF)/diffusion bonding (DB) aerospace applications, 15.108, 15.109 Al alloys, 15.109 duplex stainless steels, 15.110 Ti–6Al–4V structures, 15.105 basic shapes, 15.105, 15.107, 15.108 UHC structures, 15.109 Superplastic forming methods, 15.99–15.110 blow forming, 15.101, 15.102 gas mass forming, 15.103 movable tool forming, 15.103 Superplastic forging process, 15.95 Superplastic forming temperature, 15.79–15.81 Superplastic internal cavitation, 15.87–15.89 Superplasticity, 15.1, 15.78–15.99 in Al alloys, 15.91, 15.92 in Inconel alloys, 15.92 internal cavitation, 15.87–15.90 mechanisms: diffusion flow, 15.85, 15.87 grain boundary migration, 15.85, 15.87 grain boundary sliding (GBS), 15.84, 15.87, 15.89 transition model concept, 15.87 in microduplex stainless steels, 15.92 prerequisites, 15.83 strain rate sensitivity, 15.83, 15.84 types: fine structural, 15.82–15.92 internal stress, 15.82 Superplasticity, recent advances: high strain rate superplasticity (HSRSP), 15.92, 15.94–15.99 in ceramics, 15.94 in intermetallics, 15.93, 15.94 low-temperature superplasticity (LTSP), 15.92, 15.93, 15.95, 15.99 in nanocrystalline materials, 15.93, 15.94 quasi superplasticity, 15.99 Supersaturation parameter, 7.3 Surface diffusion, 2.90–2.92 Surface free energy, 6.3, 6.4, 6.6, 6.9–6.11 Surface martensite, 8.36 Surface modification and thin film deposition, 18.1 Sympathetic nucleation, 7.4, 7.26, 7.29 Synthesis, 18.59, 18.68, 18.69

Tangent modulus method, 12.38, 12.39 Tie line, 1.10

I.42

INDEX

Temper embrittlement, 14.71–14.81 control and prediction, 14.80 embrittlement estimative diagram, 14.78–14.80 detection and measurement of, 14.77 kinetics, 14.80 metallurgical variables, 14.72 segregation theory, 14.80, 14.81 Temper martensite embrittlement, 14.66–14.71 characteristics, 14.66, 14.67 control of, 14.70, 14.71 effect of variables, 14.68 types of, 14.67 Temper rolling, 1.43 Temperature hysteresis, 8.12, 8.13, 8.45–8.47 Temperature-time-sensitization curves, 10.60, 10.61 Tempered martensite, ferrous alloys, 14.59–14.65 effect of embrittlement on toughness, 14.63–14.65 strengthening mechanisms, 14.59–14.62 toughness of, 14.62, 14.63 Tempering methods, 14.50, 14.55–14.58 Tempering of steel, 14.1 aging reaction stages, 14.2–14.9 chi (c) carbide, 14.16 effect of alloying elements on, 14.24–14.26 effect on mechanical properties, 14.1, 14.20–14.24 faulted cementite plates, 14.17, 14.18 fifth stage, 14.2 first stage of, 14.2, 14.9–14.13 formation of gR, 14.13–14.16 fourth stage, 14.19, 14.20 habit planes, 14.16 low-carbon martensite, 14.4, 14.7, 14.9 microsyntactic intergrowth mechanism, 14.17 orientation relationships, 14.9, 14.10, 14.13, 14.14, 14.16 precipitation and growth of cementite, 14.19, 14.20 precipitation of e-carbides, 14.2, 14.3, 14.9–14.14, 14.16 precipitation of h-carbides, 14.2, 14.3, 14.9, 14.11, 14.13, 14.14, 14.16 purpose of, 14.1 recovery mechanism, 14.19 recrystallization of ferrite, 14.19 second stage, 14.2, 14.13–14.16 structural changes on, 14.1, 14.2

third stage, 14.2, 14.16, 14.17 Tempering parameter, 14.45–14.58 Grange-Hribal-Porter correlation, 14.47–14.49 Hollomon–Jaffe correlation, 14.45–14.47 Tensile properties, 4.1–4.12 Tensile strength, 4.5 Tension test, 4.1, 4.2 Ternary diffusion, 2.66–2.69 couple experiments, 2.68, 2.69 Onsager’s phenomenological diffusion coefficients, 2.66 Texture (see specific types) Thermal barrier coatings, 19.11, 19.13, 19.15, 19.16, 19.22, 19.25, 19.43, 19.46 Thermal conductivity, 13.4, 13.5, 17.15 Thermal CVD, 18.88 atmospheric pressure (APCVD), 18.88, 18.90 low-pressure (LPCVD), 18.88, 18.90 rapid (RTCVD): advantages and disadvantages, 18.92 applications, 18.91 ultralow-vacuum (UHCVD), 18.90 Thermal decomposition (or pyrolysis), 18.59–18.61, 18.69, 18.70 Thermal diffusivity, 13.5, 13.6, 17.15 Thermal evaporation: electron beam evaporation, 18.25–18.28 resistance evaporation, 18.25, 18.26 Thermal laser mechanism, 18.97 Thermal spray coatings: advantages and disadvantages, 19.2, 19.5 classifications, 19.5, 19.6 coating characteristics, a comparison, 19.1, 19.2 comparison of typical, 19.4 processes and process parameters, 19.6–19.24 Thermal spray techniques, applications, 19.36–19.51 automotive industry, 19.46 ceramic industry, 19.49 corrosion resistance, 19.42 decorative coatings, 19.48 dimensional restoration, 19.42 electrical applications, 19.43, 19.45 electronics industry, 19.45 energy industry, 19.45 friction control, 19.41, 19.42 functionally graded materials, 19.43 infrastructure maintenance, 19.42

INDEX

iron and steel industries, 19.46 machine building industries, 19.47 medical industry, 19.49 mining industry, 19.49 miscellaneous applications, 19.49 nonferrous metals industries, 19.47 paper industry, 19.48 petrochemical/chemical industry, 19.45 printing industry, 19.47, 19.48 process combinations, 19.51 nuclear industry, 19.49 shipbuilding industry, 19.47 thermal barrier coatings, 19.11, 19.42, 19.43 wear resistance, 19.38, 19.40, 19.41 Thermal stability, 13.50 Thermal surface hardening (see specific types) Thermochemical surface hardening treatments, 16.1 austenitic, 16.42–16.81, 16.111–16.114 ferritic, 16.81–16.111 (see also specific types) Thermomechanical treatment (TMT), 15.1 ferrous TMT: classification of, 15.1 controlled rolling, 15.3 conventional, 15.3, 15.7 dynamic recrystallization, 15.3, 15.7 recrystallization, 15.3, 15.24–15.26 procedure, 15.5 shortcomings, 15.7 stages of, 15.7–15.24 controlled rolling and TMCP, 15.3 effect of Sv and deformation on ferrite formation, 15.22, 15.23 procedure, 15.5, 15.6 TMCP, 15.3, 15.5 HTMT, 15.28, 15.29 isoforming, 15.30, 15.31 LTMT (or ausforming), 15.29, 15.30 nonferrous TMT, 15.56–15.78 Al-base alloys, 15.58–15.67 double recrystallization processing, 15.62, 15.63 FTMT cycle, 15.58–15.60 five-step TMT, 15.63 four-step TMT cycle, 15.60, 15.62 ITMT cycle, 15.58, 15.60, 15.62 reciprocating extrusion, 15.63, 15.64 RRAT, 15.60 Cu-base alloys, 15.77, 15.78 Ni-base superalloys, 15.56–15.58 necklace microstructure, 15.56, 15.57

I.43

Ti-Al-Fe-based alloys, 15.74, 15.75 Ti-Al-Sn-Zr-based alloys, 15.75 Ti-6Al-4V alloys, 15.69–15.76 microstructures, 15.67–15.69 processing maps, commercial and ELI grades, 15.73–15.76 Zn-base alloys, 15.67, 15.68 treatment of dual phase steels (see Dual phase steels, treatment of) treatment of TRIP or multiphase steels, 15.39–15.45 UHC steels, 15.53, 15.55, 15.56 Thermomigration, 2.71, 2.72 Thermoreactive deposition/diffusion (TRD) process, 16.138–16.144 applications, 16.142, 16.143 fluidized bed, 16.141 advantages, 16.141 applications, 16.144 salt bath TRD, 16.138 advantages and disadvantages, 16.139, 16.141 applications, 16.141 Thermotransport, 2.70, 2.71 Thin film, diffusion application, 2.92, 2.93 irreversible processes, 2.93 metastable phases, 2.93 nonlinear diffusion, 2.93 electromigration, 2.69, 2.72–2.75 microelectronics device, 2.75, 2.76 thermomigration, 2.69 Thin film metallization, 2.71, 2.75, 2.76 Ti-Al-Fe-based alloys, TMT, 15.74, 15.75 Ti-alloys, alpha, TMT, 15.75, 15.76 Ti-Al-Sn-Zr-based alloys, TMT, 15.75 Ti–6Al–4V alloys, 15.69 phase diagram, schematic, 15.69, 15.70 structures, various, 15.69, 15.71 thermomechanical treatments, 15.69, 15.71 microstructures, 15.69 bimodal, 15.70, 15.71 equiaxed, 15.69, 15.70 lamellar, 15.69 processing maps, 15.73 commercial grade, 15.73, 15.74 ELI grade, 15.73, 15.74, 15.76 variations of, 15.72, 15.73 Tiller criterion, 3.60 Time-temperature-deformation sequence, 15.3 Time–temperature–transformation (TTT) diagrams, 11.1–11.10

I.44

INDEX

Time–temperature–transformation (Cont.): for alloy steels, 11.7–11.10 applications, 11.7 for austempering, 13.77 effect of alloying elements and grain size, 11.4, 11.7 factors responsible for position and shape of, 11.6 for martempering, 13.66 for modified austempering, 13.77, 13.83 for modified martempering, 13.67 partial IT diagrams, 11.4, 11.5 for quenching and tempering, 13.66 for 1080 steel, 11.2, 11.3 for 0.47% C steel, 11.5, 11.6 Tip cracking, 16.31, 17.31, 17.33 Tool steels (see specific types) Toughness, 4.7, 8.81, 9.36, 9.37, 10.86, 10.98, 14.62–14.65, 15.38, 15.39 Transient period, 6.8 Transformation: delaying elements, 15.48 hysteresis, 8.13, 8.45 induced plasticity, 8.22 irreversible, strains, 8.45 reversible, 8.12, 8.13, 8.46, 8.47 strain-induced, 8.24 stress-assisted, 8.23, 8.24 Transgranular fracture, 14.67 Transistors, ultrahigh speed, 18.48 Transparent conductive oxide coatings, 18.87, 18.88, 18.105 Transparent electric conductor, 18.31, 18.35 Transparent permeation barrier, 18.36 Traveling wave reactor, 18.106, 18.107 Triode sputter deposition, 18.37 Trivacancy diffusion, equilibrium concentration, 2.35 Trivedi-Magnit-Kurz model, 3.59 Trivedi treatment, 7.30, 7.32–7.34 Tufftride treatment, 16.105–16.107, 16.109 Twin boundary, 4.43 coherent, 4.39, 4.43 formation of, 4.35, 5.50 interface, 4.39, 4.40 nucleation of, 4.35 secondary, 4.43 Twin spacing, 10.83 Twinned martensite, 8.34–8.36 Twinning, mechanical (or deformation), 4.33–4.44 bcc materials, 4.40–4.42

characteristics of, 4.35 fcc materials, 4.42, 4.43 and slip, 4.35 hcp materials, 4.43, 4.44 systems, 4.35, 4.36 Twins, transformation (internal), 8.28, 8.46

Ultimate tensile strength (see Tensile properties) Ultrahigh carbon steels, 15.53–15.56 applications, 15.55 effects of alloying elements, 15.55 structure-property relationships, 15.56 Ultrahigh strength low alloy steels, 15.52–15.54 Ultrasonic quenching, 17.49 Ultrasonic testing, 17.28 Ultrasonic vibration, 3.121, 10.18 Undercooling, 3.7–3.11, 3.18–3.25, 6.2, 6.12, 7.11, 7.27, 8.14, 8.45 Uniform strain, 4.6, 15.52 UNS designations, 1.53, 10.27 Uphill diffusion, 6.18, 6.19, 6.23 Uphill quenching, 17.25

Vacancy excess, 6.39 flux, 2.64, 2.66, 6.7 formation, 5.12 experimental measurements of enthalpy and entropy: direct determination, 2.37–2.39 indirect determination, 2.33–2.36 migration, 5.12 in nucleation, 6.7 quenched in, 6.38 -solute binding energy, 6.48 wind effect (or factor), 2.64–2.66 Vacancy diffusion in metals, 2.33–2.40 equilibrium concentration of divacancies, 2.35, 2.36 equilibrium concentration of trivacancies, 2.35, 2.36 migration rates of defects and atoms, 2.36 Vacancy wind correction/effect, 2.64–2.66 Vacuum arc deposition (VAD): advantages and disadvantages, 18.28, 18.29 applications, 18.30, 18.31 cathode spots, 18.28 filtered (FVAD), 18.29

INDEX

industrial batch deposition systems, 18.29 macroparticles, 18.28 other variants, 18.29, 18.30 procedure, 18.28 Vacuum arc remelted (VAR) steels, 17.4 Vacuum carburizing, 16.65–16.71 advantages and disadvantages, 16.70 carbon gradient and case depth, 16.71 comparison of, 16.69 Vacuum heat treatment, 19.25, 19.30 Vacuum impregnation, 19.30 Vacuum plasma spray (VPS) coating application, 19.13 advantages and disadvantages, 19.15 characteristic features, 19.13 process, 19.13–19.15 Vapor phase epitaxy (VPE), 18.92 Vertical cavity surface emitting layers (VCSELs), 18.48, 18.49 Voids: octahedral, 1.12, 1.13 tetrahedral, 1.12, 1.13 Volmer–Weber theory, 6.1

Warpage (see Distortion, shape) Water quenching (see Quenching media, water) Wave number, 6.23, 6.24, 6.69 critical, 6.22, 6.25 maximum, 6.24, 6.25 spinodal, 6.24, 6.25 Wavelength, 6.22, 6.24 critical, 6.22, 6.25 maximum, 6.24, 6.25 Wear, 12.43 Wear- and abrasion-resistant coatings, 18.20 Wear resistance: compact graphite iron, 12.60 ductile iron, 12.55, 12.58 gray iron, 12.43 malleable iron, 12.64 Weathering steels (see High-strength lowalloy steels) Wechsler-Lieberman-Read (WLR) theory, 8.28 Weld cracking, 3.112–3.114 chevron cracking, 3.114 HAZ cracking, 3.112–3.114 hydrogen-induced cracking, 3.114 lamellar tearing, 3.114 reheat cracking, 3.114 solidification (or hot) cracking, 3.112

I.45

Weld decay, 10.61 Weld metal (pool) solidification, 3.105 characteristics of weld pool, 3.106, 3.107 weld microstructure, 3.109–3.112 weld shape and macrostructure, 3.107–3.109 Weldability, 15.31 Wetting angle, 3.12, 3.13 Whiskers, 18.77, 18.79 White cast iron, 1.57, 3.136, 3.137 eutectic growth morphology, 3.136, 3.137 White layer, 16.89 Wide energy-gap devices, 18.53, 18.56, 18.98 Widmanstätten ferrite, 7.21, 7.27–7.35, 9.1, 9.2 formation of secondary side plates, 7.21, 7.28 growth kinetics of plates, 7.30 Cahn-Hillig-Sears (CHS) model, 7.30–7.35 lengthening kinetics, 7.30–7.34 thickening kinetics, 7.34, 7.35 Zener-Hillert equation, 7.31 intergranular plates, 7.29 ledge mechanism, 7.27 mechanism of plate formation, 7.29, 7.30 primary side plates, 7.21, 7.27 sympathetic nucleation, 7.29 Widmanstätten precipitate, 6.6, 7.27–7.35, 9.3, 9.6, 15.48 Wire, cold heading, 7.63 Wire for prestressed concrete, 7.62, 7.63 Wire rod steels, 7.57 W-ire ropes, 7.57–7.60 spring, 7.60, 7.61 tire bead and tire cords, 7.61, 7.62 Work hardening, 4.6, 4.19, 4.20, 4.22 and dislocation density/tangles, 4.21 and grain size, 4.22 rate, 4.21 theories, 4.21

X-ray diffraction methods, 4.47, 5.6, 6.25, 6.38, 6.48, 8.9, 8.28, 17.27 X-ray photoemission spectroscopy (XPS), 18.44

Yield behavior, 14.14 Yield drop, 4.15, 4.18, 4.26

I.46 Yield point: elongation, 4.16–4.18, 4.20, 4.26, 4.30 lower, 4.15, 4.16, 4.18 upper, 4.15, 4.16, 4.18 Yield strength, 4.5 Yielding: flow, inhomogeneous (or discontinuous), 4.17, 4.18 macro, 4.13, 4.15 micro, 4.13, 4.15

INDEX

Zeldovich nonequilibrium factor, 6.8 Zener-Hillert equations, 7.30, 7.31 Zener-Hollomon parameter, 15.17 Zener’s linearized concentration gradient approximation, 7.25 Zener’s local equilibrium model, 7.30, 7.31 Zener’s maximum growth rate criterion, 3.60, 7.6, 7.7 Zero-growth rate, 7.31 Zhdanov-Ramsdell notation, 8.48, 8.49 Zirconia coatings, 18.15